{ "generated_at": "2026-08-02", "schema_version": "1.0.0", "count": 125, "entries": [ { "subcategories": [], "tasks": [ "model-training", "retrieval-search" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [ "University of Warwick TIA Centre" ], "related": [ "quilt-1m", "pathvqa" ], "licence": "unknown", "licence_notes": "Derived from published figures; review the terms before redistribution.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "dataset", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "arch-dataset", "name": "ARCH", "tagline": "Pathology image-caption dataset drawn from textbooks and articles.", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "links": { "homepage": "https://warwick.ac.uk/fac/cross_fac/tia/data/arch" }, "summary": "Multiple-instance captioning dataset built from pathology textbooks and journal articles, used for image-text pretraining and retrieval evaluation.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "survival-prediction" ], "subspecialty": [ "histopathology" ], "organs": [ "breast" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Artera, Inc." ], "related": [ "arteraai-prostate" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": null, "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician", "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "arteraai-breast", "name": "ArteraAI Breast", "tagline": "Prognostic AI test using digital pathology images for breast cancer.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2026-05-04 (K254115). Product code SHW — pathology software algorithm device analysing digital images for cancer prognosis.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K254115", "verified_on": "2026-08-02" }, "links": { "homepage": "https://artera.ai/" }, "summary": "Prognostic algorithm applying the same digital-pathology approach as the company's prostate test to breast cancer.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "survival-prediction", "grading" ], "subspecialty": [ "histopathology" ], "organs": [ "prostate" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Artera, Inc." ], "related": [ "arteraai-breast", "paige-prostate" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": "Answers a different question from a detection tool: not 'is there cancer' but 'how is this likely to behave'.", "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician", "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "arteraai-prostate", "name": "ArteraAI Prostate", "tagline": "Prognostic AI test using digital pathology images for prostate cancer.", "regulatory": { "status": "fda-de-novo", "detail": "De Novo granted 2025-07-31 (DEN240068). Product code SFH — pathology software algorithm device analysing digital images for cancer prognosis.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/denovo.cfm?ID=DEN240068", "verified_on": "2026-08-02" }, "links": { "homepage": "https://artera.ai/" }, "summary": "Prognostic algorithm analysing digitised prostate biopsy images alongside clinical variables to inform risk assessment and therapy decisions. Created the SFH product code for prognostic pathology algorithms.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "data-management" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [ "windows", "linux" ], "maintainers": [ "Radboud UMC Computational Pathology Group" ], "related": [ "qupath" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": true, "min_ram_gb": 8, "clinician_note": "A no-frills slide viewer. Opens big scanner files quickly and lets you draw regions.", "caveats": null, "featured": false, "showcase": false, "id": "asap", "name": "ASAP", "tagline": "Fast whole-slide image viewer with annotation support.", "category": "software-viewer", "audience": [ "researcher", "clinician" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "GPL-2.0", "cost": "free", "links": { "repo": "https://github.com/computationalpathologygroup/ASAP" }, "origin": "academic", "summary": "Lightweight viewer for multi-resolution whole-slide images with a simple annotation workflow, widely used to inspect and label slides for challenge datasets.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "education" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "text" ], "granularity": [], "platforms": [ "web" ], "maintainers": [ "Awesome-Pathology-VLMs contributors" ], "related": [ "conch", "plip", "quilt-llava" ], "licence_notes": null, "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "Research-focused. Useful if you want the academic papers rather than usable tools.", "caveats": null, "featured": false, "showcase": false, "id": "awesome-pathology-vlms", "name": "Awesome-Pathology-VLMs", "tagline": "Curated list of pathology vision-language models, datasets and benchmarks.", "category": "meta", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "Apache-2.0", "cost": "free", "links": { "repo": "https://github.com/wenhaozhang0066/Awesome-Pathology-VLMs" }, "origin": "community", "summary": "Focused bibliography of pathology vision-language models organised by training paradigm, with a useful patch/ROI/WSI granularity axis. A good deep reference for the VLM literature specifically.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [], "related": [ "site-specific-signatures", "tcga" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "batch-effect-acquisition-site", "category": "ethics-safety", "name": "Predicting Acquisition Site from TCGA Images", "tagline": "Independent confirmation that acquisition site leaks into the pixels.", "links": { "paper": "https://pmc.ncbi.nlm.nih.gov/articles/PMC10189924/" }, "summary": "Demonstrates that networks can identify which institution produced a TCGA slide, reinforcing that site is a confounder that must be controlled in study design rather than assumed away.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "tissue-segmentation", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "lymph-node", "breast" ], "modality": [ "wsi", "he" ], "granularity": [ "G3" ], "platforms": [ "web" ], "maintainers": [ "Radboud UMC", "UMC Utrecht" ], "related": [ "tcga", "clam" ], "licence_notes": null, "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "camelyon16", "name": "CAMELYON16", "tagline": "Lymph node metastasis detection challenge dataset.", "category": "dataset", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "CC0-1.0", "cost": "free", "links": { "homepage": "https://camelyon16.grand-challenge.org/", "paper": "https://doi.org/10.1001/jama.2017.14585" }, "origin": "academic", "summary": "Sentinel lymph node whole-slide images with exhaustive metastasis annotations. The reference benchmark for weakly and fully supervised WSI detection.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "standard-interop", "subcategories": [], "tasks": [ "report-writing", "data-management" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "College of American Pathologists" ], "related": [ "snomed-ct", "hl7-fhir", "icd-o-3" ], "licence": "unknown", "licence_notes": null, "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "The reason synoptic reporting exists: every element you fill in becomes a field a registry or a trial can actually use.", "caveats": null, "featured": true, "showcase": false, "origin": "community", "stage": "production", "audience": [ "clinician", "developer" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "cap-ecc", "name": "CAP electronic Cancer Checklists (eCC)", "tagline": "Machine-readable cancer synoptic reporting protocols.", "links": { "homepage": "https://www.cap.org/protocols-and-guidelines/cancer-reporting-tools", "paper": "https://doi.org/10.5858/arpa.2020-0126-RA" }, "summary": "Machine-readable XML encoding of the CAP cancer case summaries, structured as coded question-and-answer pairs so synoptic reports become queryable data rather than prose. SNOMED CT encoded since 2007.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "quality-control" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "College of American Pathologists" ], "related": [ "leica-aperio-gt450-dx", "philips-intellisite" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "If your laboratory is going digital, this is the document your validation study will be judged against. Read it before you buy a scanner, not after.", "caveats": null, "featured": true, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "cap-wsi-validation-guideline", "category": "validation-regulatory", "name": "CAP Guideline — Validating Whole Slide Imaging", "tagline": "The reference guideline for validating a WSI system for diagnostic use.", "links": { "homepage": "https://www.cap.org/protocols-and-guidelines/cap-guidelines/current-cap-guidelines/validating-whole-slide-imaging-for-diagnostic-purposes-in-pathology" }, "summary": "Evidence-based guidance from CAP, with ASCP and the Association for Pathology Informatics, on how a laboratory should validate a whole slide imaging system before using it for diagnosis. Originally issued 2013 and updated in 2021 using the GRADE framework.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "task-specific-model", "subcategories": [], "subspecialty": [ "histopathology", "cytopathology" ], "organs": [ "any" ], "modality": [ "he", "ihc", "fluorescence" ], "granularity": [ "G1", "G2" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [ "HHMI Janelia" ], "related": [ "qupath-cellpose", "stardist", "instanseg" ], "licence": "BSD-3-Clause", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": "Usable from QuPath via an extension if you would rather not touch Python.", "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "audience": [ "researcher", "clinician" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "cellpose", "name": "Cellpose", "tagline": "Generalist cell and nucleus segmentation across imaging modalities.", "tasks": [ "nuclei-segmentation", "cell-detection" ], "links": { "repo": "https://github.com/MouseLand/cellpose", "paper": "https://doi.org/10.1038/s41592-020-01018-x" }, "summary": "Widely used generalist segmentation model that works across many microscopy types without retraining, with a human-in-the-loop retraining workflow when it does not.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "nuclei-segmentation", "cell-detection" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [ "University Hospital Essen IKIM" ], "related": [ "hovernet", "stardist" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "cellvit", "name": "CellViT", "tagline": "Vision-transformer nucleus segmentation and classification.", "category": "task-specific-model", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "Apache-2.0", "cost": "free", "links": { "repo": "https://github.com/TIO-IKIM/CellViT", "paper": "https://doi.org/10.1016/j.media.2024.103143" }, "origin": "academic", "summary": "Transformer-based nucleus instance segmentation built on pretrained ViT encoders, with inference tooling for scaling detection across whole slides.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "Radiology: Artificial Intelligence" ], "related": [ "tripod-ai", "stard-ai" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "claim-checklist", "category": "validation-regulatory", "name": "CLAIM", "tagline": "Checklist for artificial intelligence in medical imaging.", "links": { "homepage": "https://pubs.rsna.org/journal/ai" }, "summary": "Reporting checklist aimed specifically at medical imaging AI, covering data handling, ground truth definition, model description and evaluation. Widely requested by imaging journals at submission.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he" ], "granularity": [ "G3" ], "platforms": [ "linux" ], "maintainers": [ "Mahmood Lab, Harvard Medical School" ], "related": [ "trident", "titan" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": true, "showcase": false, "id": "clam", "name": "CLAM", "tagline": "Attention-based multiple-instance learning for weakly supervised whole-slide classification.", "category": "library-framework", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "GPL-3.0", "cost": "free", "links": { "repo": "https://github.com/mahmoodlab/CLAM", "paper": "https://doi.org/10.1038/s41551-020-00682-w" }, "origin": "academic", "summary": "Widely used reference implementation of attention-based MIL for slide-level classification from slide-level labels only, with heatmap interpretability. The default baseline in most WSI classification papers.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "vqa", "classification" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "tcga", "pathmmu", "patho-bench" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": "Worth knowing about because it takes seriously a problem most benchmarks ignore: whether a model can score well without actually looking at the slide.", "caveats": null, "featured": true, "showcase": false, "origin": "academic", "stage": "research", "category": "benchmark", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "cleanslide", "name": "CleanSlide", "tagline": "Leakage-audited pan-cancer whole-slide multiple-choice benchmark.", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "links": { "repo": "https://github.com/wenhaozhang0066/CleanSlide", "dataset": "https://huggingface.co/datasets/eric-1w/CleanSlide-features" }, "summary": "TCGA-derived benchmark of about 149,000 four-option questions over 9,985 diagnostic slides, with train/validation/test splits disjoint by patient AND by tissue source site, and a blind image-free baseline reported for every task.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "retrieval-search", "vqa", "tissue-segmentation" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [ "Mahmood Lab, Harvard Medical School" ], "related": [ "plip", "titan" ], "licence_notes": "Gated on Hugging Face; non-commercial research terms.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "conch", "name": "CONCH", "tagline": "Vision-language foundation model for pathology image-text tasks.", "category": "vision-language-model", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "gated", "cost": "free-for-academic", "links": { "repo": "https://github.com/mahmoodlab/CONCH", "paper": "https://doi.org/10.1038/s41591-024-02856-4" }, "origin": "academic", "summary": "Contrastively pretrained image-text model for pathology supporting zero-shot classification, cross-modal retrieval, captioning and segmentation.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "CONSORT-AI / SPIRIT-AI Working Group" ], "related": [ "decide-ai", "tripod-ai" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "consort-spirit-ai", "category": "validation-regulatory", "name": "CONSORT-AI and SPIRIT-AI", "tagline": "Trial reporting and protocol guidelines for AI interventions.", "links": { "paper": "https://doi.org/10.1038/s41591-020-1034-x" }, "summary": "Extensions of the CONSORT and SPIRIT statements for randomised trials evaluating interventions that include an AI component, published together in Nature Medicine.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "generative" ], "tasks": [ "vqa", "classification", "report-writing" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "slidechat", "wsi-llava" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "cpath-omni", "name": "CPath-Omni", "tagline": "Unified multimodal model spanning patch and whole-slide analysis.", "links": { "repo": "https://github.com/PathFoundation/CPath-Omni", "paper": "https://arxiv.org/abs/2412.12077" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Attempts to handle patch-level and slide-level tasks in a single model rather than training separate systems for each scale.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "agent" ], "tasks": [ "vqa", "classification", "report-writing" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "pathagent", "qupath" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "agent-mcp", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "cpathagent", "name": "CPathAgent", "tagline": "Agent that navigates a slide the way a pathologist moves a microscope.", "links": { "paper": "https://arxiv.org/abs/2505.20510" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Models the zoom-and-pan behaviour of slide review as agent actions, unifying patch, region and slide-level capability in one system and reporting an interpretable trajectory of what it looked at.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "data-management", "education", "telepathology" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [ "wsi", "brightfield" ], "granularity": [ "G2", "G3" ], "platforms": [ "web", "docker", "linux" ], "maintainers": [ "University of Liège" ], "related": [ "digital-slide-archive", "omero" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": "Runs on a server your institution controls, so several people can annotate the same slides in a browser. Needs IT support to set up.", "caveats": "Self-hosting requires a server and maintenance; not a laptop tool.", "featured": false, "showcase": false, "id": "cytomine", "name": "Cytomine", "tagline": "Web-based collaborative platform for annotating and analysing whole-slide images.", "category": "software-viewer", "audience": [ "researcher", "educator", "clinician" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "Apache-2.0", "cost": "free", "links": { "homepage": "https://cytomine.org/", "repo": "https://github.com/cytomine" }, "origin": "academic", "summary": "Server-hosted platform for multi-user slide annotation, review and analysis, designed for collaborative projects and teaching cohorts.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "DECIDE-AI Steering Group" ], "related": [ "tripod-ai", "consort-spirit-ai" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "The stage most pathology AI has never been through. Worth asking a vendor whether their product has.", "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher", "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "decide-ai", "category": "validation-regulatory", "name": "DECIDE-AI", "tagline": "Reporting guideline for early live clinical evaluation of AI decision support.", "links": { "paper": "https://doi.org/10.1136/bmj-2022-070904" }, "summary": "Covers the stage between offline validation and a full trial — the first time an AI system is used by real clinicians on real patients — with emphasis on safety and human factors rather than accuracy alone.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "spatial-transcriptomics", "molecular-prediction" ], "subspecialty": [ "molecular-pathology", "histopathology" ], "organs": [ "any" ], "modality": [ "he", "omics" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "istar", "hest" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "audience": [ "researcher" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "spatial-omics", "id": "deepspot", "name": "DeepSpot", "tagline": "Uses spatial context from surrounding tissue to improve expression prediction.", "links": { "paper": "https://www.medrxiv.org/content/10.1101/2025.02.09.25321567v2.full" }, "summary": "Predicts spot-level expression from H&E while explicitly modelling the neighbouring tissue context rather than treating each tile independently.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "standard-interop", "subcategories": [], "tasks": [ "report-writing", "data-management" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "DICOM Standards Committee" ], "related": [ "dicom-wsi", "hl7-fhir" ], "licence": "unknown", "licence_notes": null, "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "Ask an AI vendor how their results are stored. If the answer is a bespoke database, your results are locked in.", "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "developer" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "dicom-sr", "name": "DICOM Structured Reporting", "tagline": "Encoding measurements and findings as structured DICOM objects.", "links": { "homepage": "https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.17.html" }, "summary": "The mechanism for storing AI outputs — measurements, annotations, findings — as structured objects alongside the image rather than as a separate proprietary file.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io", "data-management", "telepathology" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [ "web" ], "maintainers": [ "DICOM Standards Committee" ], "related": [ "openslide" ], "licence_notes": "Open standard published by NEMA; the specification is free to read.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "The reason a slide from one scanner can be opened by another vendor's software. Ask about DICOM support before buying anything.", "caveats": null, "featured": true, "showcase": false, "id": "dicom-wsi", "name": "DICOM for Whole Slide Imaging", "tagline": "The vendor-neutral standard for storing and exchanging pathology slides.", "category": "standard-interop", "audience": [ "developer", "clinician", "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "homepage": "https://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_A.32.8.html", "docs": "https://dicom.nema.org/" }, "origin": "community", "summary": "The DICOM VL Whole Slide Microscopy Image IOD defines how scanned slides are represented, enabling storage and exchange independent of scanner vendor.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "data-management", "tissue-segmentation" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G2", "G3" ], "platforms": [ "web", "docker", "linux" ], "maintainers": [ "Emory University" ], "related": [ "cytomine", "histomicstk" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "digital-slide-archive", "name": "Digital Slide Archive", "tagline": "Server platform for managing, annotating and analysing large slide collections.", "category": "software-viewer", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "Apache-2.0", "cost": "free", "links": { "homepage": "https://digitalslidearchive.github.io/digital_slide_archive/", "repo": "https://github.com/DigitalSlideArchive/digital_slide_archive" }, "origin": "academic", "summary": "Web platform built on Girder for organising whole-slide image collections, with annotation tooling and a plugin system for running analysis jobs at scale.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "education", "telepathology" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "Digital Pathology Association" ], "related": [ "cap-wsi-validation-guideline" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "A good starting point if your department is considering going digital and wants peer experience rather than vendor material.", "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "clinician", "educator" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "education", "id": "dpa-telepathology-resources", "name": "Digital Pathology Association", "tagline": "Professional body publishing practical digital pathology guidance.", "links": { "homepage": "https://digitalpathologyassociation.org/" }, "summary": "Professional association publishing white papers, webinars and practical guidance on adopting digital pathology, including validation and telepathology workflows.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "generative" ], "tasks": [ "vqa" ], "subspecialty": [ "haematopathology" ], "organs": [ "bone-marrow" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [ "UC Berkeley" ], "related": [ "patho-r1" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "dr-llava", "name": "Dr-LLaVA", "tagline": "Instruction tuning grounded in symbolic clinical reasoning.", "links": { "repo": "https://github.com/AlaaLab/Dr-LLaVA", "paper": "https://arxiv.org/abs/2405.19567" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Constrains a vision-language assistant with symbolic representations of clinical reasoning steps, so that its answers follow a defensible diagnostic sequence rather than arriving at conclusions unaccountably.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "reasoning" ], "tasks": [ "vqa" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "patho-r1" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "eagle-pathology", "name": "EAGLE", "tagline": "Preference alignment to reduce hallucination in pathology VLMs.", "links": { "repo": "https://github.com/meidandz/EAGLE", "paper": "https://doi.org/10.18653/v1/2025.acl-long.711" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Builds a large set of preferred and rejected response pairs and applies iterative preference optimisation, targeting multimodal hallucination and biased answers — a failure mode that matters more in pathology than in general vision tasks.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "European Union" ], "related": [ "cap-wsi-validation-guideline" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "'CE marked' under the old directive and under IVDR are not the same claim. Check which one a vendor means.", "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "developer", "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "eu-ivdr", "category": "validation-regulatory", "name": "EU IVDR 2017/746", "tagline": "The regulation governing in vitro diagnostic devices in the European Union.", "links": { "homepage": "https://eur-lex.europa.eu/eli/reg/2017/746/oj" }, "summary": "Full legal text of the In Vitro Diagnostic Regulation, which governs CE marking for diagnostic software including pathology algorithms, and which reclassified much software upward in risk class compared with the previous directive.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "US FDA", "Health Canada", "MHRA" ], "related": [ "fda-samd-pccp", "iec-62304" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "Short, plain and non-technical. The quickest way to see what regulators expect of a medical AI product.", "caveats": null, "featured": true, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "developer", "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "fda-gmlp", "category": "validation-regulatory", "name": "Good Machine Learning Practice for Medical Device Development", "tagline": "Ten guiding principles agreed by the FDA, Health Canada and the MHRA.", "links": { "homepage": "https://www.fda.gov/medical-devices/software-medical-device-samd/good-machine-learning-practice-medical-device-development-guiding-principles" }, "summary": "Joint statement of ten principles covering data quality, representativeness, training and test independence, human factors and lifecycle monitoring for medical-device machine learning.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "US FDA" ], "related": [ "fda-gmlp" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "developer" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "fda-samd-pccp", "category": "validation-regulatory", "name": "FDA Predetermined Change Control Plans for AI Devices", "tagline": "How a cleared AI device may be updated without a new submission.", "links": { "homepage": "https://www.fda.gov/regulatory-information/search-fda-guidance-documents/marketing-submission-recommendations-predetermined-change-control-plan-artificial-intelligence" }, "summary": "FDA guidance on describing, in advance, the modifications an AI-enabled device may undergo after clearance and how they will be validated — the mechanism that makes a continuously-learning device regulable at all.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "spatial-transcriptomics", "molecular-prediction" ], "subspecialty": [ "molecular-pathology", "histopathology" ], "organs": [ "any" ], "modality": [ "he", "omics" ], "granularity": [ "G1", "G2" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "istar", "hest" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "audience": [ "researcher" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "spatial-omics", "id": "ghist", "name": "GHIST", "tagline": "Single-cell resolution spatial gene expression from histology.", "links": { "paper": "https://pmc.ncbi.nlm.nih.gov/articles/PMC12446070/" }, "summary": "Predicts spatially resolved expression at single-cell rather than spot resolution, using cell segmentation alongside image features.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "task-specific-model", "subcategories": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he" ], "granularity": [ "G3" ], "platforms": [ "linux" ], "maintainers": [ "University Hospital Cologne" ], "related": [ "histoqc", "tcga", "known-failure-modes" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": "Run this before trusting any result computed on a batch of scanned slides. Most 'model failure' turns out to be slide quality.", "caveats": null, "featured": true, "showcase": false, "origin": "academic", "stage": "production", "audience": [ "researcher", "clinician" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "grandqc", "name": "GrandQC", "tagline": "Tissue detection and multi-class artefact segmentation for whole slides.", "tasks": [ "quality-control", "tissue-segmentation" ], "links": { "repo": "https://github.com/cpath-ukk/grandqc", "paper": "https://doi.org/10.1038/s41467-024-54769-y" }, "summary": "Segments tissue and the common artefacts that ruin downstream analysis — tissue folds, pen marks, bubbles, edges, black spots, foreign objects and out-of-focus regions. Released with a manually annotated test set and QC masks for all of TCGA.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io", "telepathology", "education" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "Grundium Oy" ], "related": [ "openflexure-microscope", "openslide" ], "licence": "proprietary", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "paid", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "A realistic entry point for a department that cannot justify a bulk scanner. Small enough to sit on a desk.", "caveats": "Regulatory status varies by model and jurisdiction and was NOT verified against a primary regulator record for this entry — confirm before any diagnostic use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "audience": [ "clinician", "educator", "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "hardware-lowresource", "id": "grundium-ocus", "name": "Grundium Ocus", "tagline": "Portable single-slide scanner at a fraction of the cost of a bulk system.", "links": { "homepage": "https://www.grundium.com/" }, "summary": "Compact desktop scanner handling one slide at a time, aimed at telepathology, consultation, case documentation and teaching rather than high-volume clinical throughput.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "report-writing", "education" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [], "related": [ "pathologyoutlines" ], "licence": "unknown", "licence_notes": "Free to use; content is copyrighted by its publisher.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "Practical rather than clever — the sort of thing that saves ten minutes a day.", "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "clinician" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "app", "id": "hematogones", "name": "Hematogones.com", "tagline": "Free browser tools and synoptic reporting templates for laboratory work.", "links": { "homepage": "https://hematogones.com/" }, "summary": "Collection of browser-based calculators, cancer-protocol synoptic templates and reporting work aids, aimed at reducing the clerical load of report generation.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "spatial-transcriptomics", "model-training" ], "subspecialty": [ "molecular-pathology", "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "omics" ], "granularity": [ "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [ "Mahmood Lab, Harvard Medical School" ], "related": [ "tcga" ], "licence_notes": "Aggregates many source datasets, each with its own terms.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "hest", "name": "HEST-1k / HEST-Bench", "tagline": "Paired histology and spatial transcriptomics dataset and benchmark.", "category": "benchmark", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "repo": "https://github.com/mahmoodlab/HEST", "dataset": "https://huggingface.co/datasets/MahmoodLab/hest", "docs": "https://hest.readthedocs.io/", "paper": "https://arxiv.org/abs/2406.16192" }, "origin": "academic", "summary": "Harmonised collection of spatial transcriptomics profiles paired with histology, plus a benchmark for predicting expression from morphology.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "spatial-transcriptomics", "molecular-prediction" ], "subspecialty": [ "molecular-pathology", "histopathology" ], "organs": [ "any" ], "modality": [ "he", "omics" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "istar", "st-net" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "audience": [ "researcher" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "spatial-omics", "id": "hist2st", "name": "Hist2ST / HisToGene", "tagline": "Transformer and graph approaches to expression prediction from histology.", "links": { "paper": "https://doi.org/10.1093/bib/bbac297" }, "summary": "Two closely related lines of work applying vision transformers, and transformers combined with graph neural networks, to predict spatial expression from histology spots — modelling relationships between spots rather than each in isolation.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "id": "histomicstk", "name": "HistomicsTK", "tagline": "Python toolkit for histology image analysis and feature extraction.", "category": "library-framework", "subcategories": [], "tasks": [ "nuclei-segmentation", "stain-normalisation", "ihc-quantification", "wsi-io" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "ihc" ], "granularity": [ "G1", "G2" ], "platforms": [ "linux", "macos" ], "maintainers": [ "Kitware", "Emory University" ], "related": [ "digital-slide-archive", "tiatoolbox" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "Apache-2.0", "cost": "free", "links": { "repo": "https://github.com/DigitalSlideArchive/HistomicsTK", "docs": "https://digitalslidearchive.github.io/HistomicsTK/" }, "origin": "academic", "summary": "Library of histology-specific image analysis routines — colour deconvolution, stain normalisation, nucleus detection and morphometric feature extraction — usable standalone or as the analysis layer behind Digital Slide Archive.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "quality-control" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he" ], "granularity": [ "G3" ], "platforms": [ "linux", "macos", "windows" ], "maintainers": [ "Case Western Reserve University" ], "related": [ "qupath" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": true, "min_ram_gb": 8, "clinician_note": "Checks a batch of scanned slides for problems like blur or pen marks before anyone spends time on them.", "caveats": null, "featured": true, "showcase": false, "id": "histoqc", "name": "HistoQC", "tagline": "Automated quality control for whole-slide images.", "category": "task-specific-model", "audience": [ "researcher", "clinician" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "BSD-3-Clause", "cost": "free", "links": { "repo": "https://github.com/choosehappy/HistoQC", "paper": "https://doi.org/10.1200/CCI.18.00157" }, "origin": "academic", "summary": "Pipeline that flags slide artefacts — blur, pen marks, tissue folds, coverslip edges, bubbles — and produces an interactive report for triaging a cohort before analysis.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "standard-interop", "subcategories": [], "tasks": [ "data-management", "report-writing" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "HL7 International" ], "related": [ "dicom-wsi", "snomed-ct", "cap-ecc" ], "licence": "unknown", "licence_notes": null, "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "If a vendor cannot tell you how their output reaches your LIS, this is the vocabulary in which to ask the question.", "caveats": null, "featured": true, "showcase": false, "origin": "community", "stage": "production", "audience": [ "developer", "clinician" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "hl7-fhir", "name": "HL7 FHIR", "tagline": "The interoperability standard for exchanging health data, including reports.", "links": { "homepage": "https://hl7.org/fhir/" }, "summary": "Resource-based standard for health data exchange. DiagnosticReport, Specimen, Observation and ImagingStudy are the resources a pathology system needs to speak to the rest of the hospital.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "nuclei-segmentation", "cell-detection" ], "subspecialty": [ "histopathology" ], "organs": [ "colon", "breast", "any" ], "modality": [ "he" ], "granularity": [ "G1", "G2" ], "platforms": [ "linux" ], "maintainers": [ "University of Warwick TIA Centre" ], "related": [ "stardist", "cellvit", "pannuke" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "hovernet", "name": "HoVer-Net", "tagline": "Simultaneous nucleus segmentation and classification in histology.", "category": "task-specific-model", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "MIT", "cost": "free", "links": { "repo": "https://github.com/vqdang/hover_net", "paper": "https://doi.org/10.1016/j.media.2019.101563" }, "origin": "academic", "summary": "Network predicting horizontal and vertical distances to nuclear centres to separate touching nuclei, while simultaneously classifying nucleus type. A standard baseline for nucleus tasks.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "task-specific-model", "subcategories": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he" ], "granularity": [ "G1", "G2" ], "platforms": [ "linux" ], "maintainers": [ "University of Bern" ], "related": [ "hovernet", "cellvit", "pannuke" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "audience": [ "researcher", "clinician" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "hovernext", "name": "HoVer-NeXt", "tagline": "Faster successor to HoVer-Net for nucleus segmentation and classification.", "tasks": [ "nuclei-segmentation", "cell-detection" ], "links": { "repo": "https://github.com/digitalpathologybern/hover_next_inference" }, "summary": "Reimplementation aimed at making whole-slide nucleus segmentation and classification tractable at scale rather than on selected tiles.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "quality-control" ], "subspecialty": [ "histopathology" ], "organs": [ "prostate", "breast" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Ibex Medical Analytics Ltd." ], "related": [ "paige-prostate" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": "Runs after you have signed out, and raises a flag if it disagrees. Designed as a safety net, not a first opinion.", "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "ibex-galen-second-read", "name": "Ibex Galen Second Read", "tagline": "AI second-read quality control for biopsy whole-slide images.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2025-01-24 (K241232). Product code QPN — software algorithm device to assist users in digital pathology.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K241232", "verified_on": "2026-08-02" }, "links": { "homepage": "https://ibex-ai.com/" }, "summary": "Software that reviews already-diagnosed cases and flags discrepancies between the algorithm's assessment and the recorded diagnosis, intended as a quality-control second read rather than a primary diagnostic step.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "standard-interop", "subcategories": [], "tasks": [ "data-management" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "World Health Organization" ], "related": [ "cap-ecc", "snomed-ct" ], "licence": "unknown", "licence_notes": null, "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "clinician" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "icd-o-3", "name": "ICD-O-3", "tagline": "Topography and morphology coding for cancer registries.", "links": { "homepage": "https://www.who.int/standards/classifications/other-classifications/international-classification-of-diseases-for-oncology" }, "summary": "WHO classification coding tumour site and morphology, the coding backbone of cancer registration worldwide.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "IEC" ], "related": [ "iso-14971", "fda-gmlp" ], "licence": "proprietary", "licence_notes": "The standard must be purchased from IEC or a national body.", "cost": "paid", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": "Not free. Budget for it if you intend to build a regulated product.", "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "developer" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "iec-62304", "category": "validation-regulatory", "name": "IEC 62304", "tagline": "International standard for medical device software lifecycle processes.", "links": { "homepage": "https://www.iso.org/standard/38421.html" }, "summary": "Defines the lifecycle requirements for medical device software: planning, requirements, architecture, verification, release and maintenance, scaled by safety class.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "data-management", "ihc-quantification" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Indica Labs, Inc." ], "related": [ "proscia-concentriq-dx" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": null, "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician", "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "indica-halo-ap-dx", "name": "Indica Labs HALO AP Dx", "tagline": "Anatomic pathology workflow and image analysis platform.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2025-11-25 (K252762) as digital pathology image viewing and management software (QKQ); earlier clearance K232833 in 2024 as a whole slide imaging system (PSY).", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K252762", "verified_on": "2026-08-02" }, "links": { "homepage": "https://indicalab.com/" }, "summary": "Case management and viewing platform for anatomic pathology, from a company whose research image-analysis software is widely used in academic laboratories.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "task-specific-model", "subcategories": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he", "ihc", "fluorescence" ], "granularity": [ "G1", "G2" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [], "related": [ "qupath-instanseg", "cellpose", "stardist" ], "licence": "Apache-2.0", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "beta", "audience": [ "researcher", "clinician" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "instanseg", "name": "InstanSeg", "tagline": "Embedding-based cell segmentation that runs acceptably on a CPU.", "tasks": [ "nuclei-segmentation", "cell-detection" ], "links": { "repo": "https://github.com/instanseg/instanseg" }, "summary": "Instance segmentation approach designed for speed, able to run without a dedicated GPU — which matters more than raw accuracy when there is no workstation available.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "ISO" ], "related": [ "iec-62304" ], "licence": "proprietary", "licence_notes": "The standard must be purchased.", "cost": "paid", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": "Not free.", "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "developer" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "iso-14971", "category": "validation-regulatory", "name": "ISO 14971", "tagline": "Risk management standard for medical devices.", "links": { "homepage": "https://www.iso.org/standard/72704.html" }, "summary": "The framework regulators expect for identifying hazards, estimating risk and showing that residual risk is acceptable. Underpins nearly every other device requirement.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "spatial-transcriptomics", "molecular-prediction" ], "subspecialty": [ "molecular-pathology", "histopathology" ], "organs": [ "any" ], "modality": [ "he", "omics" ], "granularity": [ "G1", "G2" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "hest", "st-net", "deepspot" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 32, "clinician_note": "The idea worth knowing: morphology carries enough signal to interpolate molecular data between measured points — but it is inference, not measurement.", "caveats": null, "featured": true, "showcase": false, "origin": "academic", "stage": "production", "audience": [ "researcher" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "spatial-omics", "id": "istar", "name": "iStar", "tagline": "Super-resolution spatial gene expression from histology and spatial transcriptomics.", "links": { "repo": "https://github.com/daviddaiweizhang/istar", "paper": "https://www.nature.com/articles/s41587-023-02019-9" }, "summary": "Uses hierarchical image features to integrate spatial transcriptomics with high-resolution histology, raising expression predictions towards near-single-cell resolution rather than the coarse spots the assay produces.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "Awesome AI in Pathology" ], "related": [ "site-specific-signatures", "cleanslide", "decide-ai" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "Read this before trusting any performance figure. Most published accuracy is measured under conditions that do not resemble a working laboratory.", "caveats": "Maintained by this project rather than drawn from a single source; each failure mode links to its own primary literature.", "featured": true, "showcase": false, "origin": "community", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "clinician", "researcher", "educator" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "known-failure-modes", "category": "ethics-safety", "name": "Known Failure Modes in Pathology AI", "tagline": "A plain-language index of how these systems actually go wrong.", "links": { "homepage": "https://github.com/atultiwari/awesome-ai-pathology/blob/main/DISCLAIMER.md" }, "summary": "Curated index of documented failure modes: site and batch confounding, scanner and stain variability, distribution shift after deployment, miscalibrated confidence, hallucinated findings in generative models, and automation bias in the reader.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io", "data-management" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Leica Biosystems Imaging, Inc." ], "related": [ "philips-intellisite", "roche-ventana-dp200", "openslide" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": "The scanner itself, not an AI tool — it produces the digital slides everything else works on.", "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "leica-aperio-gt450-dx", "name": "Leica Aperio GT 450 DX", "tagline": "Whole-slide scanner cleared for primary diagnosis.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2024-04-16 (K232202). Product code PSY — whole slide imaging system.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K232202", "verified_on": "2026-08-02" }, "links": { "homepage": "https://www.leicabiosystems.com/" }, "summary": "High-throughput brightfield whole-slide scanner and associated software cleared as a whole slide imaging system for diagnostic use.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "generative" ], "tasks": [ "vqa", "education" ], "subspecialty": [ "histopathology", "clinical-pathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [ "Microsoft Research" ], "related": [ "quilt-llava", "pa-llava" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": "Biomedical-general rather than pathology-specific; histopathology is a minority of its training data.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "llava-med", "name": "LLaVA-Med", "tagline": "Biomedical vision-language assistant trained rapidly from PubMed figures.", "links": { "repo": "https://github.com/microsoft/LLaVA-Med", "paper": "https://arxiv.org/abs/2306.00890" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "General biomedical multimodal assistant covering many imaging modalities including histopathology. Frequently used as the baseline that pathology-specific assistants are compared against.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "standard-interop", "subcategories": [], "tasks": [ "data-management" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "Regenstrief Institute" ], "related": [ "snomed-ct", "hl7-fhir" ], "licence": "unknown", "licence_notes": null, "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "developer", "clinician" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "loinc", "name": "LOINC", "tagline": "Universal codes for laboratory observations and report sections.", "links": { "homepage": "https://loinc.org/" }, "summary": "Naming system for tests, measurements and document sections, used alongside SNOMED CT so that the question and the answer are both coded.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "benchmark", "subcategories": [], "subspecialty": [ "histopathology", "veterinary-pathology" ], "organs": [ "breast", "skin", "any" ], "modality": [ "he" ], "granularity": [ "G1", "G2" ], "platforms": [ "linux" ], "maintainers": [ "DeepMicroscopy" ], "related": [ "known-failure-modes", "site-specific-signatures", "cellvit" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": "Mitotic count drives grading in several tumours, and it is one of the least reproducible things pathologists do — which is exactly why this benchmark measures generalisation rather than headline accuracy.", "caveats": null, "featured": true, "showcase": false, "origin": "academic", "stage": "production", "audience": [ "researcher", "clinician" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "midog", "name": "MIDOG Challenge", "tagline": "Mitosis detection benchmarked across scanners, tumours and species.", "tasks": [ "mitosis-counting", "cell-detection" ], "links": { "homepage": "https://midog2025.grand-challenge.org/", "repo": "https://github.com/DeepMicroscopy/MIDOG", "paper": "https://arxiv.org/abs/2204.03742" }, "summary": "Challenge series built specifically around domain shift: models must count mitoses on scanners, tumour types and species they were not trained on. Successive editions have widened the domain gap deliberately.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "contrastive" ], "tasks": [ "classification", "retrieval-search", "vqa", "molecular-prediction" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [ "Stanford University" ], "related": [ "conch", "titan" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "musk", "name": "MUSK", "tagline": "Vision-language foundation model for precision oncology.", "links": { "repo": "https://github.com/lilab-stanford/MUSK", "paper": "https://doi.org/10.1038/s41586-024-08378-w" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Trained on large volumes of unpaired image and text data as well as paired examples, then applied to retrieval, visual question answering, classification and molecular biomarker prediction.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "model-training", "education" ], "subspecialty": [ "histopathology" ], "organs": [ "colon" ], "modality": [ "he" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [ "NCT Heidelberg" ], "related": [ "camelyon16" ], "licence_notes": null, "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "A good starting dataset if you are learning — small enough to work with on a laptop.", "caveats": null, "featured": true, "showcase": false, "id": "nct-crc-he-100k", "name": "NCT-CRC-HE-100K", "tagline": "Colorectal tissue-type patch classification dataset.", "category": "dataset", "audience": [ "researcher", "educator" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "CC-BY-4.0", "cost": "free", "links": { "dataset": "https://zenodo.org/records/1214456", "paper": "https://doi.org/10.1371/journal.pmed.1002730" }, "origin": "academic", "summary": "Around 100,000 colorectal H&E patches labelled into nine tissue classes. Small, permissively licensed and CPU-tractable, which makes it the usual first dataset for teaching and prototyping.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "standard-interop", "subcategories": [], "tasks": [ "wsi-io", "data-management" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "Open Microscopy Environment" ], "related": [ "dicom-wsi", "openslide", "omero" ], "licence": "BSD-2-Clause", "licence_notes": null, "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "Matters if slides are going to the cloud — proprietary scanner formats were not designed for that.", "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "developer", "researcher" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "ome-ngff", "name": "OME-Zarr / NGFF", "tagline": "Cloud-native chunked format for bioimaging, including whole slides.", "links": { "homepage": "https://ngff.openmicroscopy.org/", "repo": "https://github.com/ome/ngff" }, "summary": "Next-generation file format storing large multi-resolution images as chunked arrays, designed to be read efficiently from object storage rather than a local disk.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "data-management", "annotation" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [ "wsi", "fluorescence", "brightfield" ], "granularity": [ "G2", "G3" ], "platforms": [ "web", "linux", "docker" ], "maintainers": [ "Open Microscopy Environment" ], "related": [ "digital-slide-archive" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "omero", "name": "OMERO", "tagline": "Image data management server for microscopy, including whole-slide images.", "category": "software-viewer", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "GPL-2.0", "cost": "free", "links": { "homepage": "https://www.openmicroscopy.org/omero/", "repo": "https://github.com/ome/openmicroscopy" }, "origin": "academic", "summary": "Long-established server and client suite for storing, annotating and sharing microscopy image data across many proprietary formats.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "telepathology", "education", "annotation" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "linux", "web" ], "maintainers": [ "OpenFlexure Project", "University of Bath" ], "related": [ "grundium-ocus", "qupath" ], "licence": "CERN-OHL-S-2.0", "licence_notes": "Open hardware licence; the software is separately open-source.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": false, "min_ram_gb": 4, "clinician_note": "The genuinely low-cost option. You print the body, add optics and a Raspberry Pi, and get automated scanning for telepathology or teaching.", "caveats": "Not a diagnostic device, and image quality does not match a commercial scanner. Appropriate for education, documentation and second opinions rather than primary diagnosis.", "featured": true, "showcase": false, "origin": "community", "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "audience": [ "clinician", "educator", "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "hardware-lowresource", "id": "openflexure-microscope", "name": "OpenFlexure Microscope", "tagline": "Open-source 3D-printed automated microscope for low-resource settings.", "links": { "homepage": "https://openflexure.org/about/pathology", "repo": "https://gitlab.com/openflexure/openflexure-microscope" }, "summary": "3D-printable motorised microscope that automates scanning and stitches composite images viewable on a laptop, tablet or phone. Designed explicitly for settings where a commercial scanner is out of reach.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [ "Carnegie Mellon University" ], "related": [ "qupath", "tiatoolbox" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": true, "min_ram_gb": 4, "clinician_note": null, "caveats": null, "featured": true, "showcase": false, "id": "openslide", "name": "OpenSlide", "tagline": "C library with Python bindings for reading proprietary whole-slide formats.", "category": "library-framework", "audience": [ "developer", "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "LGPL-2.1", "cost": "free", "links": { "homepage": "https://openslide.org/", "repo": "https://github.com/openslide/openslide" }, "origin": "academic", "summary": "The foundational library for reading vendor whole-slide formats (Aperio, Hamamatsu, Leica, MIRAX and others) through one common interface. Most Python pathology tooling depends on it.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "linux" ], "maintainers": [], "related": [ "openflexure-microscope" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": "A research design rather than a product; building it requires hardware skills.", "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher", "developer" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "hardware-lowresource", "id": "openwsi", "name": "OpenWSI", "tagline": "Low-cost high-throughput whole slide imaging from open-source hardware.", "links": { "paper": "https://arxiv.org/abs/1912.03446" }, "summary": "Whole slide imaging system built from off-the-shelf and open-source components, using single-frame autofocusing to keep cost and complexity down.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "tissue-segmentation", "survival-prediction", "retrieval-search" ], "subspecialty": [ "histopathology", "molecular-pathology" ], "organs": [ "any" ], "modality": [ "wsi", "he" ], "granularity": [ "G2", "G3" ], "platforms": [ "web" ], "maintainers": [ "Owkin" ], "related": [ "pathagent", "phikon-v2", "pathgpt" ], "licence": "proprietary", "licence_notes": "Check the repository before any reuse.", "cost": "paid", "self_hostable": false, "sends_data_offsite": true, "offline_capable": false, "hardware_floor": "cloud-only", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": null, "clinician_note": "Notable as an early example of pathology analysis reachable from a general assistant rather than a dedicated application.", "caveats": "Hosted: slide-derived data leaves your environment. Research use — check the terms and your data-protection obligations before using patient material.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "audience": [ "researcher" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "owkin-pathology-explorer", "category": "agent-mcp", "name": "Owkin Pathology Explorer", "tagline": "Pathology analysis agent reachable over the Model Context Protocol.", "links": { "homepage": "https://www.owkin.com/newsfeed/owkins-specialized-biological-ai-agent-pathology-explorer-launches-with-anthropics-claude-for-healthcare-and-life-sciences" }, "summary": "Agent that locates cell and tissue types in digitised slides for spatially aware analysis of tumours and their microenvironment, exposed to assistants through MCP. Aimed at research and drug development rather than diagnosis.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "generative" ], "tasks": [ "vqa", "classification" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "plip", "llava-med" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pa-llava", "name": "PA-LLaVA", "tagline": "Pathology language-vision assistant built on PLIP with two-stage training.", "links": { "repo": "https://github.com/ddw2AIGROUP2CQUPT/PA-LLaVA", "paper": "https://doi.org/10.1109/BIBM62325.2024.10821785" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Domain-specific assistant using a PLIP-derived encoder, trained first for domain alignment and then for instruction following on pathology question answering.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "data-management" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Paige.AI, Inc." ], "related": [ "paige-prostate", "pathai-aisight-dx" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": null, "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "paige-fullfocus", "name": "Paige FullFocus", "tagline": "Whole-slide image viewer for primary diagnosis.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2025-01-09 (K241273); earlier clearance K201005 in 2020. Product code QKQ — digital pathology image viewing and management software.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K241273", "verified_on": "2026-08-02" }, "links": { "homepage": "https://www.paige.ai/" }, "summary": "Viewer for reading whole-slide images in a diagnostic workflow, cleared separately from the company's analysis algorithms.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "grading" ], "subspecialty": [ "histopathology" ], "organs": [ "prostate" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Paige.AI, Inc." ], "related": [ "paige-fullfocus", "ibex-galen-second-read" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": "An assistive second look at prostate biopsies. It flags suspicious areas — it does not make the diagnosis, and you remain responsible for the report.", "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": true, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician", "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "paige-prostate", "name": "Paige Prostate", "tagline": "AI aid for detecting prostate cancer in core needle biopsies.", "regulatory": { "status": "fda-de-novo", "detail": "De Novo granted 2021-09-21 (DEN200080). Product code QPN — software algorithm device to assist users in digital pathology. The first AI device authorised by FDA for use in pathology.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/denovo.cfm?ID=DEN200080", "verified_on": "2026-08-02" }, "links": { "homepage": "https://www.paige.ai/" }, "summary": "Software intended to assist pathologists in detecting areas suspicious for cancer in prostate core needle biopsy whole-slide images. Notable as the first AI device FDA authorised for pathology, creating the QPN product code.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "grading", "classification", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "prostate" ], "modality": [ "wsi", "he" ], "granularity": [ "G3" ], "platforms": [ "web" ], "maintainers": [ "Radboud UMC", "Karolinska Institutet" ], "related": [ "tcga" ], "licence_notes": "Kaggle competition terms apply; check before redistribution.", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "panda", "name": "PANDA", "tagline": "Prostate biopsy Gleason grading challenge dataset.", "category": "dataset", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "homepage": "https://www.kaggle.com/competitions/prostate-cancer-grade-assessment", "paper": "https://doi.org/10.1038/s41591-021-01620-2" }, "origin": "academic", "summary": "The largest public prostate biopsy collection with Gleason grade labels, from two institutions, used to benchmark automated grading against pathologist variability.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "id": "pannuke", "name": "PanNuke", "tagline": "Pan-cancer nucleus instance segmentation and classification dataset.", "category": "dataset", "subcategories": [], "tasks": [ "nuclei-segmentation", "cell-detection", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [ "University of Warwick TIA Centre" ], "related": [ "hovernet", "cellvit", "stardist" ], "licence_notes": "Non-commercial research licence — check the terms before any commercial use.", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": "Nucleus class definitions are coarse and were produced semi-automatically then refined; treat per-class performance figures with corresponding caution.", "featured": false, "showcase": false, "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "homepage": "https://warwick.ac.uk/fac/cross_fac/tia/data/pannuke", "paper": "https://arxiv.org/abs/2003.10778" }, "origin": "academic", "summary": "Nucleus instance segmentation and classification dataset spanning 19 tissue types with five nucleus classes, semi-automatically annotated. One of the standard training and evaluation sets for nucleus models.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "agent" ], "tasks": [ "vqa", "report-writing" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "cpathagent", "patho-agenticrag" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": true, "offline_capable": false, "hardware_floor": "cloud-only", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": null, "caveats": "Depends on a hosted LLM, so slide-derived content leaves the local machine — check this against your data-protection obligations before using patient material.", "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "agent-mcp", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathagent", "name": "PathAgent", "tagline": "Training-free LLM agent that explores whole slides and reasons over findings.", "links": { "repo": "https://github.com/G14nTDo4/PathAgent", "paper": "https://arxiv.org/abs/2511.17052" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Uses an LLM to plan slide exploration, locate informative regions, extract visual cues and assemble them into a reasoning trajectory, without task-specific training.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "data-management" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "PathAI, Inc." ], "related": [ "paige-fullfocus", "proscia-concentriq-dx" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": null, "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use. Note that a clearance for viewing and management software is not a clearance of any analysis algorithm that may run alongside it.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathai-aisight-dx", "name": "PathAI AISight Dx", "tagline": "Digital pathology image management and viewing platform.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2025-06-26 (K243391). Product code QKQ — digital pathology image viewing and management software.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K243391", "verified_on": "2026-08-02" }, "links": { "homepage": "https://www.pathai.com/" }, "summary": "Platform for viewing, managing and reviewing whole-slide images in a clinical workflow, cleared as viewing and management software.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "generative" ], "tasks": [ "vqa", "classification" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "pathcap", "pathinstruct" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathasst", "name": "PathAsst", "tagline": "Generative pathology assistant with a domain-adapted CLIP encoder.", "links": { "repo": "https://github.com/superjamessyx/Generative-Foundation-AI-Assistant-for-Pathology", "paper": "https://arxiv.org/abs/2305.15072" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Combines a pathology-adapted CLIP vision encoder with a language model, trained on the PathCap caption corpus and PathInstruct instruction data released alongside it.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he", "text" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [], "licence": "unknown", "licence_notes": "Check the dataset card before redistribution.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "category": "dataset", "audience": [ "researcher" ], "tasks": [ "model-training" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathcap", "name": "PathCap", "tagline": "Pathology image-caption corpus released with PathAsst.", "links": { "dataset": "https://huggingface.co/datasets/jamessyx/PathCap", "paper": "https://arxiv.org/abs/2305.15072" }, "summary": "Roughly 207,000 pathology image-text pairs curated for vision-language pretraining, released alongside the PathAsst assistant.", "related": [ "pathasst", "pathinstruct" ], "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "generative" ], "tasks": [ "vqa", "report-writing", "education" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1", "G2" ], "platforms": [ "linux" ], "maintainers": [ "Mahmood Lab, Harvard Medical School" ], "related": [ "conch", "uni" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": "The published work is research; downstream commercial versions are separate products with their own regulatory status, which is not covered by this entry.", "featured": false, "showcase": false, "origin": "academic", "stage": "production", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathchat", "name": "PathChat", "tagline": "Multimodal generative copilot for human pathology.", "links": { "repo": "https://github.com/fedshyvana/pathology_mllm_training", "paper": "https://doi.org/10.1038/s41586-024-07618-3" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Vision-language assistant built on a pathology foundation encoder, able to take image and text input together and hold a diagnostic conversation about a case.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "education" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [], "related": [ "webpathology", "pathologyoutlines" ], "licence": "unknown", "licence_notes": "Free to use; content is copyrighted by its publisher.", "cost": "freemium", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "clinician", "educator" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathdojo", "name": "PathDojo", "tagline": "Practice cases and self-testing for pathologists and trainees.", "category": "education", "links": { "homepage": "https://pathdojo.com/" }, "summary": "Case-based practice environment built by pathologists for training and self-testing, rather than a passive reference.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "contrastive" ], "tasks": [ "classification", "retrieval-search", "tissue-segmentation" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "conch", "pathreasoner-r1" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathflip", "name": "PathFLIP", "tagline": "Fine-grained language-image pretraining grounding captions to regions.", "links": { "repo": "https://github.com/cyclexfy/PathFLIP", "paper": "https://doi.org/10.1609/aaai.v40i9.37649" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Aligns slide captions to specific image regions rather than whole images, aiming to improve slide classification, retrieval and lesion localisation.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "vqa", "education", "report-writing" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [ "text" ], "granularity": [], "platforms": [ "web" ], "maintainers": [], "related": [ "known-failure-modes", "owkin-pathology-explorer" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "The most useful part is the structure, not the individual prompts: it shows how to build a prompt that can be evaluated rather than just admired.", "caveats": "A research study rather than a maintained prompt library. Treat the prompts as a starting template, and never as validated clinical tooling.", "featured": true, "showcase": false, "origin": "academic", "stage": "research", "audience": [ "clinician", "researcher", "educator" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathgpt", "category": "prompt-skill", "name": "PathGPT", "tagline": "Structured pathology prompt library built on explicit prompt patterns.", "links": { "repo": "https://github.com/slrenne/PathGPT" }, "summary": "Clinico-pathological scenarios written across ten subspecialties, each reviewed by a subspecialty pathologist, and composed from named prompt modules — persona, clinical scenario, question type, reflection and references — then evaluated open-ended and multiple-choice.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he", "text" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [], "licence": "unknown", "licence_notes": "Check the dataset card before redistribution.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "category": "dataset", "audience": [ "researcher" ], "tasks": [ "model-training", "vqa" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathinstruct", "name": "PathInstruct", "tagline": "Pathology instruction-tuning dataset released with PathAsst.", "links": { "dataset": "https://huggingface.co/datasets/jamessyx/PathInstruct", "paper": "https://arxiv.org/abs/2305.15072" }, "summary": "Around 180,000 instruction-response pairs for pathology, used to give a vision-language model conversational and task-following behaviour.", "related": [ "pathasst", "pathcap" ], "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io", "nuclei-segmentation", "stain-normalisation", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "ihc", "multiplex" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux", "macos" ], "maintainers": [ "Dana-Farber Cancer Institute AI Operations & Data Science" ], "related": [ "tiatoolbox" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "pathml", "name": "PathML", "tagline": "Python library for computational pathology preprocessing and modelling.", "category": "library-framework", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "GPL-2.0", "cost": "free", "links": { "repo": "https://github.com/Dana-Farber-AIOS/pathml", "docs": "https://pathml.readthedocs.io/" }, "origin": "academic", "summary": "Framework for building reproducible pathology pipelines, with support for brightfield, multiparametric and volumetric imaging and a preprocessing DSL.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "vqa" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [], "related": [ "pathvqa", "cleanslide" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "benchmark", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathmmu", "name": "PathMMU", "tagline": "Expert-validated multimodal pathology understanding benchmark.", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "links": { "repo": "https://github.com/PathMMU-Benchmark/PathMMU", "dataset": "https://huggingface.co/datasets/jamessyx/PathMMU", "homepage": "https://pathmmu-benchmark.github.io/", "paper": "https://arxiv.org/abs/2401.16355" }, "summary": "Multiple-choice benchmark of roughly 33,000 questions over 24,000 images, assembled from several sources and checked by pathologists.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "agent" ], "tasks": [ "vqa", "retrieval-search", "education" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "patho-r1", "pathagent" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": "The interesting idea here is grounding: answers point back to a textbook page instead of being asserted from nowhere.", "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "agent-mcp", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "patho-agenticrag", "name": "Patho-AgenticRAG", "tagline": "Multimodal retrieval-augmented generation over pathology textbooks.", "links": { "repo": "https://github.com/Wenchuan-Zhang/Patho-AgenticRAG", "paper": "https://doi.org/10.1609/aaai.v40i35.40239" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Builds a searchable index over authoritative pathology textbook content supporting joint text and image queries, so answers can be grounded in a citable source rather than model memory.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [ "linux" ], "maintainers": [ "Mahmood Lab, Harvard Medical School" ], "related": [ "titan", "uni", "trident" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "patho-bench", "name": "Patho-Bench", "tagline": "Standardised evaluation suite for pathology slide-level foundation models.", "category": "benchmark", "audience": [ "researcher" ], "stage": "beta", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "repo": "https://github.com/mahmoodlab/Patho-Bench" }, "origin": "academic", "summary": "Harness for comparing slide-level foundation models across many public tasks under consistent splits and protocols, reducing the incomparability of self-reported numbers.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "reasoning" ], "tasks": [ "vqa", "classification" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "pathvlm-r1", "patho-agenticrag", "dr-llava" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "patho-r1", "name": "Patho-R1", "tagline": "Reinforcement-learning pathology reasoner trained on chain-of-thought data.", "links": { "repo": "https://github.com/Wenchuan-Zhang/Patho-R1", "paper": "https://doi.org/10.1609/aaai.v40i33.40071" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Applies continued pretraining, supervised fine-tuning on reasoning traces, then reinforcement learning, releasing the PathCoT reasoning dataset alongside.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "education" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [], "related": [ "pathologyoutlines", "webpathology" ], "licence": "unknown", "licence_notes": "Free to use; content is copyrighted by its publisher.", "cost": "freemium", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "clinician", "educator" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "app", "id": "pathologyapps", "name": "Apps for Pathologists", "tagline": "Differential diagnosis outlines organised by organ system and bench.", "links": { "homepage": "https://pathologyapps.com/" }, "summary": "Online and mobile reference presenting differential diagnoses in outline form, organised the way a pathologist actually works — by bench and by organ system.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "education" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "PathologyOutlines.com" ], "related": [], "licence_notes": "Free to read; content is copyrighted by the publisher.", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "Not an AI tool, but the reference most pathologists already use — included because AI outputs should always be checked against a trusted source.", "caveats": null, "featured": false, "showcase": false, "id": "pathologyoutlines", "name": "PathologyOutlines", "tagline": "Free comprehensive online pathology reference.", "category": "education", "audience": [ "clinician", "educator" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "homepage": "https://www.pathologyoutlines.com/" }, "origin": "community", "summary": "Widely used free reference covering diagnostic criteria, immunohistochemistry and differentials across pathology, maintained by a large contributor network.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "education", "telepathology" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "PathPresenter Corporation" ], "related": [ "pathai-aisight-dx" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": "Many pathologists already know the teaching platform; the clinical viewer is a separate, cleared product.", "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician", "educator" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathpresenter-clinical-viewer", "name": "PathPresenter Clinical Viewer", "tagline": "Whole-slide viewer with teaching and conferencing roots.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2025-05-14 (K250968). Product code QKQ — digital pathology image viewing and management software.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K250968", "verified_on": "2026-08-02" }, "links": { "homepage": "https://pathpresenter.com/" }, "summary": "Clinical viewer from a platform widely used for pathology education, tumour boards and case sharing, with the clinical module cleared for diagnostic viewing.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "reasoning" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he", "text" ], "granularity": [ "G1", "G3" ], "platforms": [ "linux" ], "maintainers": [], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "tasks": [ "vqa", "classification" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathreasoner-r1", "name": "PathReasoner-R1", "tagline": "Knowledge-guided structured reasoning for whole-slide pathology.", "links": { "repo": "https://github.com/cyclexfy/PathReasoner-R1", "paper": "https://arxiv.org/abs/2601.21617" }, "summary": "Uses a knowledge graph to tie pathological findings to clinical reasoning steps and diagnoses, with a slide-level reasoning dataset and a multi-granular reward function.", "related": [ "patho-r1", "pathflip" ], "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "reasoning" ], "tasks": [ "vqa" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "patho-r1", "smartpath-r1" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathvlm-r1", "name": "PathVLM-R1", "tagline": "GRPO-trained reasoning model for pathology visual questions.", "links": { "paper": "https://arxiv.org/abs/2504.09258" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Injects domain knowledge by supervised fine-tuning, then applies policy optimisation with format and accuracy rewards on a general-purpose vision-language backbone.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "vqa" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [], "related": [ "pathmmu", "quilt-llava" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": "Questions were largely derived automatically from textbook captions, so some are answerable from language alone without looking at the image.", "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "benchmark", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "pathvqa", "name": "PathVQA", "tagline": "Pathology visual question answering benchmark.", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "links": { "repo": "https://github.com/KaveeshaSilva/PathVQA", "dataset": "https://huggingface.co/datasets/flaviagiammarino/path-vqa", "paper": "https://arxiv.org/abs/2003.10286" }, "summary": "Early and still widely cited pathology VQA set of roughly 5,000 images and 32,000 question-answer pairs, used as a standard comparison point.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he" ], "granularity": [ "G1" ], "platforms": [ "linux", "macos" ], "maintainers": [ "Owkin" ], "related": [ "uni", "virchow2" ], "licence_notes": "Openly licensed — unusual among pathology foundation models and worth preferring where it performs comparably.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": "Notable because it is genuinely open — no access request, no non-commercial restriction.", "caveats": null, "featured": false, "showcase": false, "id": "phikon-v2", "name": "Phikon-v2", "tagline": "Openly licensed pathology tile encoder trained on public cohorts.", "category": "foundation-model", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "MIT", "cost": "free", "links": { "model": "https://huggingface.co/owkin/phikon-v2", "paper": "https://arxiv.org/abs/2409.09173" }, "origin": "industry", "summary": "ViT trained with DINOv2 on public histopathology cohorts and released under a permissive licence, making it one of the few pathology encoders usable without gated access.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io", "data-management" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Philips Medical Systems Nederland B.V." ], "related": [ "leica-aperio-gt450-dx", "roche-ventana-dp200" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": null, "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "philips-intellisite", "name": "Philips IntelliSite Pathology Solution", "tagline": "Whole-slide imaging system for primary diagnosis.", "regulatory": { "status": "fda-510k", "detail": "Most recent 510(k) clearance 2025-03-06 (K243871), with several earlier clearances in the same family. Product code PSY — whole slide imaging system.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K243871", "verified_on": "2026-08-02" }, "links": { "homepage": "https://www.philips.com/healthcare" }, "summary": "Scanner, storage and viewing system for digital pathology, one of the earliest whole slide imaging systems authorised in the United States for primary diagnosis.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "retrieval-search" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux", "macos" ], "maintainers": [ "Stanford University" ], "related": [ "conch", "quilt-llava" ], "licence_notes": "Check the model card and the OpenPath data terms before reuse.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": "Training data provenance is social media; coverage and label quality vary by subspecialty.", "featured": false, "showcase": false, "id": "plip", "name": "PLIP", "tagline": "Pathology CLIP model trained on image-text pairs mined from social media.", "category": "vision-language-model", "audience": [ "researcher" ], "stage": "research", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "repo": "https://github.com/PathologyFoundation/plip", "demo": "https://huggingface.co/spaces/vinid/webplip", "paper": "https://doi.org/10.1038/s41591-023-02504-3" }, "origin": "academic", "summary": "CLIP fine-tuned on the OpenPath corpus of pathology image-text pairs curated from public medical Twitter, enabling zero-shot classification and image-text retrieval.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "contrastive", "generative" ], "tasks": [ "classification", "report-writing" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [ "Paige AI" ], "related": [ "virchow2", "titan" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "industry", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "prism", "name": "PRISM", "tagline": "Slide-level multimodal model generating reports from tile embeddings.", "links": { "paper": "https://arxiv.org/abs/2405.10254" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Aggregates Virchow tile embeddings into a slide-level representation supervised by clinical reports, supporting report generation, cancer detection and subtyping.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "data-management" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Proscia, Inc." ], "related": [ "pathai-aisight-dx", "indica-halo-ap-dx" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": null, "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "proscia-concentriq-dx", "name": "Proscia Concentriq Dx", "tagline": "Digital pathology platform for diagnostic viewing and workflow.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2024-02-08 (K230839). Product code PSY — whole slide imaging system.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K230839", "verified_on": "2026-08-02" }, "links": { "homepage": "https://proscia.com/" }, "summary": "Enterprise platform for storing, viewing and routing whole-slide images through a diagnostic workflow, with an application layer for third-party algorithms.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "molecular-prediction", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he" ], "granularity": [ "G1", "G3" ], "platforms": [ "linux" ], "maintainers": [ "Microsoft Research", "Providence Health" ], "related": [ "uni", "titan" ], "licence_notes": "Gated on Hugging Face; non-commercial research terms apply.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "prov-gigapath", "name": "Prov-GigaPath", "tagline": "Slide-level foundation model with a tile encoder and long-context aggregator.", "category": "foundation-model", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "gated", "cost": "free-for-academic", "links": { "model": "https://huggingface.co/prov-gigapath/prov-gigapath", "repo": "https://github.com/prov-gigapath/prov-gigapath", "paper": "https://doi.org/10.1038/s41586-024-07441-w" }, "origin": "industry", "summary": "Pairs a tile-level encoder with a LongNet-based slide aggregator so whole slides can be represented in one pass, evaluated on subtyping and mutation prediction.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "model-training", "retrieval-search" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he", "text" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [ "Quilt-1M authors" ], "related": [ "quilt-llava", "plip" ], "licence_notes": "Derived from third-party video content; review the terms before redistribution.", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "quilt-1m", "name": "Quilt-1M", "tagline": "Histopathology image-text pairs mined from educational videos.", "category": "dataset", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "dataset": "https://zenodo.org/records/8239942", "repo": "https://github.com/wisdomikezogwo/quilt1m", "paper": "https://arxiv.org/abs/2306.11207" }, "origin": "academic", "summary": "Large open histopathology image-caption corpus assembled from educational videos, used to train pathology vision-language models.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "vqa", "education" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he", "text" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [ "Quilt-LLaVA authors" ], "related": [ "plip", "quilt-1m" ], "licence_notes": "Check model and dataset cards; derived from third-party video content.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "quilt-llava", "name": "Quilt-LLaVA", "tagline": "Instruction-tuned pathology assistant trained from educational video narration.", "category": "vision-language-model", "audience": [ "researcher" ], "stage": "research", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "repo": "https://github.com/aldraus/quilt-llava", "homepage": "https://quilt-llava.github.io/", "paper": "https://arxiv.org/abs/2312.04746" }, "origin": "academic", "summary": "Multimodal assistant instruction-tuned on localised narratives extracted from open histopathology teaching videos, aimed at diagnostic reasoning across patches.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "cell-detection", "ihc-quantification", "tissue-segmentation" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [ "wsi", "brightfield", "fluorescence", "ihc", "he" ], "granularity": [ "G2", "G3" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [ "University of Edinburgh" ], "related": [ "qupath-stardist", "qupath-cellpose", "qupath-instanseg", "openslide" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": "Free, installs on an ordinary laptop, and does not upload your slides anywhere. The usual first stop if you have never used digital pathology software.", "caveats": "Not a diagnostic device. Very large slides benefit from more RAM.", "featured": true, "showcase": false, "id": "qupath", "name": "QuPath", "tagline": "Open-source desktop software for whole-slide image analysis and annotation.", "category": "software-viewer", "audience": [ "clinician", "researcher", "developer", "educator" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "GPL-3.0", "cost": "free", "links": { "homepage": "https://qupath.github.io/", "repo": "https://github.com/qupath/qupath", "docs": "https://qupath.readthedocs.io/", "paper": "https://doi.org/10.1038/s41598-017-17204-5" }, "origin": "academic", "summary": "Cross-platform desktop application for whole-slide and microscopy image analysis, with scripting, an extension system, and strong annotation tooling. The de facto open-source starting point for digital pathology.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "nuclei-segmentation", "cell-detection" ], "subspecialty": [ "histopathology", "cytopathology" ], "organs": [ "any" ], "modality": [ "he", "ihc", "fluorescence" ], "granularity": [ "G1", "G2" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [ "EPFL BioImaging & Optics Platform" ], "related": [ "qupath", "qupath-stardist" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "none", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": "Another automatic cell-detection option inside QuPath. Needs a one-time Python setup.", "caveats": "Requires a working local Cellpose environment; setup is the fiddly part.", "featured": false, "showcase": false, "id": "qupath-cellpose", "name": "QuPath Cellpose Extension", "tagline": "Brings Cellpose and Omnipose segmentation into QuPath.", "category": "qupath-extension", "audience": [ "clinician", "researcher" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "GPL-3.0", "cost": "free", "links": { "repo": "https://github.com/BIOP/qupath-extension-cellpose" }, "origin": "academic", "summary": "Community extension that calls a local Cellpose/Omnipose install from QuPath, adding generalist cell segmentation models to the QuPath workflow.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "nuclei-segmentation", "cell-detection" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he", "ihc", "fluorescence" ], "granularity": [ "G1", "G2" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [ "QuPath team" ], "related": [ "qupath", "qupath-stardist" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": "Newer segmentation option that runs without a graphics card.", "caveats": null, "featured": false, "showcase": false, "id": "qupath-instanseg", "name": "QuPath InstanSeg Extension", "tagline": "InstanSeg nucleus and cell segmentation inside QuPath.", "category": "qupath-extension", "audience": [ "clinician", "researcher" ], "stage": "beta", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "Apache-2.0", "cost": "free", "links": { "repo": "https://github.com/qupath/qupath-extension-instanseg" }, "origin": "academic", "summary": "Extension exposing InstanSeg, an embedding-based instance segmentation method, with bundled models runnable on CPU directly from QuPath.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "nuclei-segmentation", "cell-detection" ], "subspecialty": [ "histopathology", "cytopathology" ], "organs": [ "any" ], "modality": [ "he", "ihc", "fluorescence" ], "granularity": [ "G1", "G2" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [ "QuPath team" ], "related": [ "qupath", "stardist" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": "Adds automatic nucleus detection to QuPath. Runs on a normal laptop, no coding needed.", "caveats": null, "featured": true, "showcase": false, "id": "qupath-stardist", "name": "QuPath StarDist Extension", "tagline": "Runs StarDist star-convex nucleus detection inside QuPath.", "category": "qupath-extension", "audience": [ "clinician", "researcher" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "Apache-2.0", "cost": "free", "links": { "repo": "https://github.com/qupath/qupath-extension-stardist", "docs": "https://qupath.readthedocs.io/en/stable/docs/deep/stardist.html" }, "origin": "academic", "summary": "Official QuPath extension wrapping StarDist, giving point-and-click nucleus segmentation on H&E, IHC and fluorescence images without writing code.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io", "data-management" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Ventana Medical Systems, Inc." ], "related": [ "leica-aperio-gt450-dx", "philips-intellisite" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": null, "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "roche-ventana-dp200", "name": "Roche Digital Pathology Dx (VENTANA DP 200)", "tagline": "Whole-slide scanner and viewing software for diagnostic use.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2024-12-17 (K242783); related clearance K232879 for VENTANA DP 200 in 2024. Product code PSY — whole slide imaging system.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K242783", "verified_on": "2026-08-02" }, "links": { "homepage": "https://diagnostics.roche.com/" }, "summary": "Slide scanner with accompanying image management and viewing software, cleared as a whole slide imaging system.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "task-specific-model", "subcategories": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he" ], "granularity": [ "G1", "G2" ], "platforms": [ "linux" ], "maintainers": [ "Meta AI" ], "related": [ "qupath", "cellpose", "stardist" ], "licence": "Apache-2.0", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": "Its practical value in pathology today is speeding up manual annotation, not replacing it.", "caveats": "Not trained on pathology. Performance on nuclei is well below purpose-built models; treat pathology-specific fine-tunes separately.", "featured": false, "showcase": false, "origin": "industry", "stage": "research", "audience": [ "researcher", "clinician" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "sam-pathology", "name": "Segment Anything for Pathology", "tagline": "Applying foundation segmentation models to histology.", "tasks": [ "annotation", "nuclei-segmentation", "tissue-segmentation" ], "links": { "repo": "https://github.com/facebookresearch/segment-anything", "paper": "https://arxiv.org/abs/2304.02643" }, "summary": "General-purpose promptable segmentation model that has been widely adapted for histology, most usefully as an interactive annotation accelerator rather than as an unattended segmenter.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "annotation", "data-management", "telepathology" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G3" ], "platforms": [], "maintainers": [ "Sectra AB" ], "related": [ "proscia-concentriq-dx", "dicom-wsi" ], "licence": "proprietary", "licence_notes": null, "cost": "paid", "self_hostable": null, "sends_data_offsite": null, "offline_capable": null, "hardware_floor": null, "bandwidth": null, "needs_scanner": true, "min_ram_gb": null, "clinician_note": "Worth knowing about if your hospital already uses Sectra for radiology — the pathology module reuses that infrastructure.", "caveats": "Clearance is specific to the indication for use stated in the FDA record and to the United States. It implies nothing about availability or approval elsewhere. Verify the current indication and your own jurisdiction's status before clinical use.", "featured": false, "showcase": false, "origin": "industry", "stage": "production", "category": "commercial-product", "audience": [ "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "sectra-digital-pathology", "name": "Sectra Digital Pathology Module", "tagline": "Digital pathology module within an enterprise imaging platform.", "regulatory": { "status": "fda-510k", "detail": "510(k) cleared 2024-04-16 (K232208) for version 3.3; earlier clearance K193054 in 2020. Product code PSY — whole slide imaging system.", "reference": "https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPMN/pmn.cfm?ID=K232208", "verified_on": "2026-08-02" }, "links": { "homepage": "https://sectra.com/medical/digital-pathology/" }, "summary": "Pathology module of an enterprise imaging platform, notable for sharing infrastructure with radiology PACS in institutions that already run it.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [], "related": [ "tcga", "cleanslide", "batch-effect-acquisition-site" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "The most important cautionary result in computational pathology. It explains why a model reporting excellent accuracy on TCGA may fail completely on your slides.", "caveats": null, "featured": true, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher", "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "site-specific-signatures", "category": "ethics-safety", "name": "Site-Specific Digital Histology Signatures", "tagline": "Evidence that models can learn the submitting site instead of the biology.", "links": { "paper": "https://doi.org/10.1038/s41467-021-24698-1" }, "summary": "Shows that images from different TCGA submitting sites are distinguishable by deep learning, that stain normalisation and augmentation do not remove the signal, and that this produces biased accuracy for survival, mutation and stage prediction.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "generative" ], "tasks": [ "vqa", "report-writing" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "wsi-llava", "cpath-omni" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "slidechat", "name": "SlideChat", "tagline": "Vision-language assistant operating on whole slides rather than patches.", "links": { "repo": "https://github.com/uni-medical/SlideChat", "homepage": "https://uni-medical.github.io/SlideChat.github.io/", "paper": "https://arxiv.org/abs/2410.11761" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Assistant designed for gigapixel whole-slide input, released with the Slide-Instruction training set and the SlideBench-VQA evaluation set built from TCGA and BCNB.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "model-training", "survival-prediction", "molecular-prediction" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he" ], "granularity": [ "G1", "G3" ], "platforms": [ "linux" ], "maintainers": [ "Slideflow contributors" ], "related": [ "clam", "tiatoolbox" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "slideflow", "name": "Slideflow", "tagline": "End-to-end deep learning pipeline for whole-slide images.", "category": "library-framework", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "GPL-3.0", "cost": "free", "links": { "repo": "https://github.com/slideflow/slideflow", "docs": "https://slideflow.dev/" }, "origin": "academic", "summary": "Opinionated pipeline covering tessellation, training, MIL, explainability and deployment, with both TensorFlow and PyTorch backends and a slide viewer.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "vlm-mil" ], "tasks": [ "classification" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "vila-mil", "clam" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "slip-mil", "name": "SLIP", "tagline": "Slide-level prompt learning for few-shot MIL.", "links": { "repo": "https://github.com/LTS5/SLIP", "paper": "https://arxiv.org/abs/2503.17238" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Draws prior knowledge from a language model about which tissue types matter for a given task, then uses it to guide slide classification when labelled data is scarce.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "reasoning" ], "tasks": [ "vqa", "classification" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "patho-r1", "cpath-omni" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "smartpath-r1", "name": "SmartPath-R1", "tagline": "Mixture-of-experts reasoner unifying ROI and whole-slide tasks.", "links": { "paper": "https://arxiv.org/abs/2507.17303" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Combines scale-dependent fine-tuning with task-aware reinforcement learning and expert routing so one model can serve both region-level and slide-level questions.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "standard-interop", "subcategories": [], "tasks": [ "data-management" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "SNOMED International" ], "related": [ "cap-ecc", "hl7-fhir", "loinc" ], "licence": "unknown", "licence_notes": "Free in member countries; licensing applies elsewhere. Check your national release centre.", "cost": "free-for-academic", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "The difference between a report a computer can count and one it can only store.", "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "developer", "clinician" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "snomed-ct", "name": "SNOMED CT", "tagline": "Clinical terminology underpinning coded pathology reporting.", "links": { "homepage": "https://www.snomed.org/", "docs": "https://docs.snomed.org/implementation-guides/cancer-synoptic-reporting-implementation-guide/1-introduction" }, "summary": "The terminology that makes a diagnosis machine-comparable rather than free text, with a dedicated implementation guide for cancer synoptic reporting.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "spatial-transcriptomics" ], "subspecialty": [ "molecular-pathology" ], "organs": [ "any" ], "modality": [ "he", "omics" ], "granularity": [ "G1" ], "platforms": [ "web" ], "maintainers": [], "related": [ "istar", "site-specific-signatures", "known-failure-modes" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": "A useful corrective: impressive correlation numbers in this field are sensitive to how the test set was built.", "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "audience": [ "researcher" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "st-data-quality", "name": "Data Quality in Histology-to-Expression Prediction", "category": "ethics-safety", "tagline": "How much reported performance is an artefact of the evaluation setup.", "links": { "paper": "https://pmc.ncbi.nlm.nih.gov/articles/PMC12439975/" }, "summary": "Examines how data quality and evaluation choices shape reported accuracy for predicting spatial transcriptomics from histology, and finds that some apparent performance does not survive stricter setups.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "spatial-transcriptomics", "molecular-prediction" ], "subspecialty": [ "molecular-pathology", "histopathology" ], "organs": [ "breast" ], "modality": [ "he", "omics" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [ "Stanford University" ], "related": [ "istar", "hest" ], "licence": "unknown", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "audience": [ "researcher" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "category": "spatial-omics", "id": "st-net", "name": "ST-Net", "tagline": "Early demonstration that gene expression can be predicted from H&E alone.", "links": { "repo": "https://github.com/bryanhe/ST-Net", "paper": "https://doi.org/10.1038/s41551-020-0578-x" }, "summary": "Convolutional model trained on matched breast histology and spatial transcriptomics to predict local expression from image tiles. The reference point most later work compares against.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [], "related": [ "tiatoolbox", "site-specific-signatures", "histoqc" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher", "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "stain-scanner-variability", "category": "ethics-safety", "name": "Stain and Scanner Variability", "tagline": "Why a model trained in one laboratory degrades in another.", "links": { "paper": "https://doi.org/10.1038/s43856-022-00186-5" }, "summary": "Collected evidence and tooling around colour, stain protocol and scanner differences as a source of performance loss, and the normalisation approaches used to mitigate it — none of which fully solve the problem.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "STARD-AI Steering Group" ], "related": [ "tripod-ai", "claim-checklist" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher", "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "stard-ai", "category": "validation-regulatory", "name": "STARD-AI", "tagline": "Reporting guideline for AI diagnostic accuracy studies.", "links": { "paper": "https://www.nature.com/articles/s41591-025-03953-8" }, "summary": "Extension of the STARD statement for studies reporting the diagnostic accuracy of an AI-based test, developed through a multi-stakeholder consensus process.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "nuclei-segmentation", "cell-detection" ], "subspecialty": [ "histopathology", "cytopathology" ], "organs": [ "any" ], "modality": [ "he", "ihc", "fluorescence" ], "granularity": [ "G1", "G2" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [ "StarDist authors" ], "related": [ "qupath-stardist", "cellvit" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "none", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": null, "featured": true, "showcase": false, "id": "stardist", "name": "StarDist", "tagline": "Star-convex polygon nucleus detection.", "category": "task-specific-model", "audience": [ "researcher", "developer", "clinician" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "BSD-3-Clause", "cost": "free", "links": { "repo": "https://github.com/stardist/stardist", "paper": "https://arxiv.org/abs/1806.03535" }, "origin": "academic", "summary": "Instance segmentation method representing nuclei as star-convex polygons, robust on crowded nuclei and the basis of the most-used QuPath segmentation extension.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "model-training", "molecular-prediction", "survival-prediction" ], "subspecialty": [ "histopathology", "molecular-pathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "omics" ], "granularity": [ "G3" ], "platforms": [ "web" ], "maintainers": [ "National Cancer Institute" ], "related": [ "camelyon16", "clam" ], "licence_notes": "Open and controlled-access tiers; controlled data requires dbGaP authorisation.", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": null, "clinician_note": null, "caveats": "Tissue-source-site signal is a well-documented confounder — split by site, not just by patient, or results inflate.", "featured": true, "showcase": false, "id": "tcga", "name": "TCGA", "tagline": "Pan-cancer whole-slide and molecular data archive.", "category": "dataset", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-01" }, "licence": "unknown", "cost": "free", "links": { "homepage": "https://www.cancer.gov/ccg/research/genome-sequencing/tcga", "dataset": "https://portal.gdc.cancer.gov/" }, "origin": "academic", "summary": "The most widely used public source of paired whole-slide images and molecular data across cancer types, underpinning most published WSI classification, survival and biomarker work.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io", "tissue-segmentation", "nuclei-segmentation", "classification", "stain-normalisation" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "ihc" ], "granularity": [ "G1", "G2", "G3" ], "platforms": [ "linux", "macos", "windows" ], "maintainers": [ "University of Warwick TIA Centre" ], "related": [ "openslide", "pathml" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "tiatoolbox", "name": "TIAToolbox", "tagline": "Python toolbox for computational pathology pipelines.", "category": "library-framework", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "BSD-3-Clause", "cost": "free", "links": { "repo": "https://github.com/TissueImageAnalytics/tiatoolbox", "docs": "https://tia-toolbox.readthedocs.io/", "paper": "https://doi.org/10.1038/s43856-022-00186-5" }, "origin": "academic", "summary": "Batteries-included Python library covering slide reading, stain normalisation, patch extraction, pretrained models and end-to-end inference pipelines.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "retrieval-search", "report-writing" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G3" ], "platforms": [ "linux" ], "maintainers": [ "Mahmood Lab, Harvard Medical School" ], "related": [ "uni", "clam", "conch" ], "licence_notes": "Gated on Hugging Face; research terms apply.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "titan", "name": "TITAN", "tagline": "Multimodal whole-slide foundation model producing slide-level embeddings.", "category": "foundation-model", "audience": [ "researcher" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "gated", "cost": "free-for-academic", "links": { "model": "https://huggingface.co/MahmoodLab/TITAN", "repo": "https://github.com/mahmoodlab/TITAN", "paper": "https://doi.org/10.1038/s41591-025-03982-3" }, "origin": "academic", "summary": "Slide-level model combining visual self-supervision with vision-language alignment, giving whole-slide embeddings usable without task-specific training.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "category": "task-specific-model", "subcategories": [], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "he" ], "granularity": [ "G1", "G2" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [], "related": [ "stain-scanner-variability", "tiatoolbox" ], "licence": "MIT", "licence_notes": "Check the repository before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": 8, "clinician_note": null, "caveats": "Normalisation reduces but does not eliminate inter-laboratory variation — see the stain and scanner variability entry before assuming it solves domain shift.", "featured": false, "showcase": false, "origin": "academic", "stage": "production", "audience": [ "researcher", "clinician" ], "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "torchstain", "name": "torchstain", "tagline": "Stain normalisation as a differentiable, GPU-capable operation.", "tasks": [ "stain-normalisation" ], "links": { "repo": "https://github.com/EIDOSLAB/torchstain" }, "summary": "Implements the standard stain normalisation methods (Macenko, Reinhard, Vahadane) in PyTorch, TensorFlow and NumPy so normalisation can sit inside a training pipeline rather than as a preprocessing step.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "wsi-io", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi" ], "granularity": [ "G1", "G3" ], "platforms": [ "linux" ], "maintainers": [ "Mahmood Lab, Harvard Medical School" ], "related": [ "clam", "uni", "titan" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "trident", "name": "TRIDENT", "tagline": "Toolkit for whole-slide preprocessing and foundation-model feature extraction.", "category": "library-framework", "audience": [ "researcher", "developer" ], "stage": "beta", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "GPL-3.0", "cost": "free", "links": { "repo": "https://github.com/mahmoodlab/TRIDENT" }, "origin": "academic", "summary": "Pipeline for tissue segmentation, patching and batch feature extraction across many pathology foundation models behind one interface.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [ "TRIPOD+AI Group" ], "related": [ "stard-ai", "claim-checklist", "decide-ai" ], "licence": "unknown", "licence_notes": "Published document; check the publisher's terms for reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "A quick way to judge a paper: if it does not report these items, you cannot tell whether the model would work in your laboratory.", "caveats": null, "featured": true, "showcase": false, "origin": "academic", "stage": "production", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "audience": [ "researcher", "clinician" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "tripod-ai", "category": "validation-regulatory", "name": "TRIPOD+AI", "tagline": "Reporting guideline for clinical prediction models using machine learning.", "links": { "paper": "https://doi.org/10.1136/bmj-2023-078378" }, "summary": "Updated TRIPOD statement covering prediction models built with regression or machine learning. Published in the BMJ in 2024, it sets out what a paper must report for its model to be assessable by anyone else.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "model-training", "retrieval-search" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [ "Mahmood Lab, Harvard Medical School" ], "related": [ "clam", "trident", "titan" ], "licence_notes": "Access is gated on Hugging Face and the licence restricts commercial use. Read the terms before any non-research application.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 16, "clinician_note": null, "caveats": "Gated weights with a non-commercial licence. Trained on a corpus that is not public, so independent replication is limited.", "featured": true, "showcase": false, "id": "uni", "name": "UNI", "tagline": "Self-supervised vision foundation model for pathology tile encoding.", "category": "foundation-model", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "gated", "cost": "free-for-academic", "links": { "model": "https://huggingface.co/MahmoodLab/UNI", "repo": "https://github.com/mahmoodlab/UNI", "paper": "https://doi.org/10.1038/s41591-024-02857-3" }, "origin": "academic", "summary": "ViT encoder trained with DINOv2 on a large private histopathology corpus, used as a frozen tile-level feature extractor across a wide range of downstream tasks.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "vlm-mil" ], "tasks": [ "classification" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "clam", "slip-mil" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "vila-mil", "name": "ViLa-MIL", "tagline": "Dual-scale vision-language MIL for whole-slide classification.", "links": { "repo": "https://github.com/Jiangbo-Shi/ViLa-MIL", "paper": "https://doi.org/10.1109/CVPR52733.2024.01069" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Uses text descriptions at two magnifications as prompts to steer multiple-instance learning, transferring vision-language priors into slide classification.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "model-training" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he" ], "granularity": [ "G1" ], "platforms": [ "linux" ], "maintainers": [ "Paige AI" ], "related": [ "uni", "prov-gigapath" ], "licence_notes": "Gated on Hugging Face; check the model card for permitted use.", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": false, "min_ram_gb": 32, "clinician_note": null, "caveats": "Industry-released with gated access; the training corpus is not public.", "featured": false, "showcase": false, "id": "virchow2", "name": "Virchow2", "tagline": "Large pathology vision foundation model trained on a very large slide corpus.", "category": "foundation-model", "audience": [ "researcher", "developer" ], "stage": "production", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "gated", "cost": "free-for-academic", "links": { "model": "https://huggingface.co/paige-ai/Virchow2", "paper": "https://arxiv.org/abs/2408.00738" }, "origin": "industry", "summary": "Vision transformer trained self-supervised on a very large multi-institutional slide collection, released for research as a general-purpose tile encoder.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "education" ], "subspecialty": [ "any" ], "organs": [ "any" ], "modality": [], "granularity": [], "platforms": [ "web" ], "maintainers": [], "related": [ "pathologyoutlines", "pathdojo" ], "licence": "unknown", "licence_notes": "Free to use; content is copyrighted by its publisher.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": false, "hardware_floor": "cpu", "bandwidth": "low", "needs_scanner": false, "min_ram_gb": null, "clinician_note": "Not AI, but included deliberately — AI output should always be checked against a trusted image reference.", "caveats": null, "featured": false, "showcase": false, "origin": "community", "stage": "production", "audience": [ "clinician", "educator" ], "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "webpathology", "name": "WebPathology", "tagline": "Free image atlas of benign and malignant entities.", "category": "education", "links": { "homepage": "https://www.webpathology.com/" }, "summary": "Large free collection of annotated pathology images organised by organ system, widely used for teaching and quick visual reference.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "vqa" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G3" ], "platforms": [ "web" ], "maintainers": [], "related": [ "wsi-llava", "cleanslide" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": false, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "cpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 16, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "benchmark", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "wsi-bench", "name": "WSI-Bench", "tagline": "Whole-slide VQA benchmark organised around morphological description.", "regulatory": { "status": "not-applicable", "detail": null, "reference": null, "verified_on": "2026-08-02" }, "links": { "homepage": "https://wsi-llava.github.io/", "paper": "https://arxiv.org/abs/2412.02141" }, "summary": "Large slide-level question set spanning morphological description, diagnosis and clinical reasoning categories, released with WSI-LLaVA.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [ "generative" ], "tasks": [ "vqa", "report-writing" ], "subspecialty": [ "histopathology" ], "organs": [ "any" ], "modality": [ "wsi", "he", "text" ], "granularity": [ "G2", "G3" ], "platforms": [ "linux" ], "maintainers": [], "related": [ "slidechat", "wsi-bench" ], "licence": "unknown", "licence_notes": "Check the repository and model card before any reuse.", "cost": "free", "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "workstation-gpu", "bandwidth": "high", "needs_scanner": true, "min_ram_gb": 32, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "origin": "academic", "stage": "research", "category": "vision-language-model", "audience": [ "researcher" ], "added_on": "2026-08-02", "last_verified": "2026-08-02", "id": "wsi-llava", "name": "WSI-LLaVA", "tagline": "Multimodal model for whole-slide images with a morphology-aware benchmark.", "links": { "homepage": "https://wsi-llava.github.io/", "paper": "https://arxiv.org/abs/2412.02141" }, "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-02" }, "summary": "Slide-level assistant released together with WSI-Bench, an evaluation set organised around morphological description as well as diagnosis.", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } }, { "subcategories": [], "tasks": [ "classification", "tissue-segmentation" ], "subspecialty": [ "histopathology" ], "organs": [ "breast", "prostate", "lung", "colon" ], "modality": [ "wsi", "he" ], "granularity": [ "G1", "G3" ], "platforms": [ "windows", "macos", "linux" ], "maintainers": [ "Stony Brook University" ], "related": [ "qupath" ], "licence_notes": null, "self_hostable": true, "sends_data_offsite": false, "offline_capable": true, "hardware_floor": "consumer-gpu", "bandwidth": "low", "needs_scanner": true, "min_ram_gb": 8, "clinician_note": null, "caveats": null, "featured": false, "showcase": false, "id": "wsinfer", "name": "WSInfer", "tagline": "Runs pretrained patch classification models across whole slides, with a QuPath front end.", "category": "qupath-extension", "audience": [ "researcher", "clinician" ], "stage": "beta", "regulatory": { "status": "ruo", "detail": "Research use only. Not for primary diagnosis.", "reference": null, "verified_on": "2026-08-01" }, "licence": "Apache-2.0", "cost": "free", "links": { "repo": "https://github.com/SBU-BMI/wsinfer", "docs": "https://wsinfer.readthedocs.io/" }, "origin": "academic", "summary": "Toolkit and model zoo for running patch-level deep learning inference over whole-slide images, usable from the command line or through a QuPath extension.", "added_on": "2026-08-01", "last_verified": "2026-08-01", "metrics": { "github_stars": null, "last_commit": null, "hf_downloads": null, "refreshed_on": null } } ] }