This repository also configures its own agents. See what Pathology-Wiki tells them →
Skill Claude CodeCodex
Paige.AI's multimodal generative slide-level PFM. Virchow patch encoder + slide-level Transformer + generative text decoder. Sister to TITAN with explicit slide-level report generation.
Skill Claude CodeCodex
Transcriptomics-guided slide-level PFM. CLIP-style contrastive alignment between WSI and bulk RNA-seq. Sibling to report-guided multimodal slide PFMs (TITAN, PRISM).
Skill Claude CodeCodex
Extract technical-focus fields from an article that primarily proposes or evaluates an algorithm, model, training scheme, dataset, or benchmark.
Skill Claude CodeCodex
First widely-cited multimodal whole-slide foundation model. CONCH/UNI patch encoder + slide-level Transformer + report alignment + synthetic captions. Cite when slide-text retrieval / slide captioning / zero-shot slide-level classification are the topic.
Skill Claude CodeCodex
The standard reference for Transformer-based MIL in pathology. Cite when patch-patch correlations are explicitly modelled.
Skill Claude CodeCodex
Provide Claude with the rules for citing and using the UNI paper.
Skill Claude CodeCodex
Largest published patch-level pathology PFM at release: ViT-H/14, DINOv2, 1.5M MSKCC WSIs. Cite for PFM scaling claims and rare-cancer detection.
Skill Claude CodeCodex
ICIAR 2018 breast-histology classification challenge — 400 microscopy ROIs + 10 WSIs in 4 morphology classes (normal / benign / in-situ / invasive). Small-data stress test for stain robustness and few-shot transfer.
Skill Claude CodeCodex
The seed public benchmark for breast-cancer lymph-node metastasis detection on H&E WSIs (400 slides, two centres). Anchors most weakly-supervised MIL papers.
Skill Claude CodeCodex
NIH/NCI proteomics-focused sister cohort to TCGA. 4.5k FFPE WSIs, 2k patients, 10 cancer cohorts. The canonical external validation cohort for TCGA-trained pathology models, and the only large public cohort with paired H&E + bulk mass-spec proteomics.
Skill Claude CodeCodex
Canonical multi-domain mitotic-figure detection benchmark on H&E. Multiple scanners, institutions, tumour types, and species (human + canine). Purpose-built for domain generalisation.
Skill Claude CodeCodex
The standard public benchmark for prostate Gleason / ISUP grading on H&E biopsy WSIs.
Skill Claude CodeCodex
Standard public benchmark for nuclei instance segmentation and 5-class classification on H&E patches across 19 tissue types. PQ is the canonical metric.
Skill Claude CodeCodex
The largest public multi-omics cancer cohort: 30k WSIs, 33 cancer types, 11k patients, paired with RNA-seq / mutations / CNV / methylation / clinical records. Default training cohort for almost every pathology AI paper.
Skill Claude CodeCodex
Maintain knowledge/graph/{nodes.yaml, edges.yaml} and produce graph.json for the website. Source of truth for graph topology (CLAUDE.md §8).
Skill Claude CodeCodex
Update knowledge/taxonomies/.yaml when a new object introduces a method / task / disease / fusion-pair / clinical category not yet covered.
Skill Claude CodeCodex
Default attention-MIL aggregator. Permutation-invariant learned attention pooling over instance embeddings; the foundational pathology MIL aggregator.
Skill Claude CodeCodex
Modern ViT-based nuclei instance segmentation + classification. Keeps HoVer-Net's three-decoder head; replaces the encoder with a SAM-pretrained ViT.
Skill Claude CodeCodex
De-facto modern open-source weakly-supervised MIL aggregator. Per-class attention branches + instance-level clustering loss.
Skill Claude CodeCodex
Canonical CNN for joint nuclei instance segmentation + classification on H&E. Three-decoder architecture with horizontal-vertical distance maps for instance separation.
Skill Claude CodeCodex
Pretrain a patch encoder on millions of unlabelled pathology patches using SSL (contrastive / DINO / MIM). Frozen features feed downstream MIL aggregators.
Skill Claude CodeCodex
Umbrella method for any large pretrained pathology encoder. Use this node to position a new model / paper / benchmark in the PFM space.
Skill Claude CodeCodex
Umbrella method for pathology vision-language models. Aligns image features with text via contrastive (CLIP-style) or generative pretraining; enables zero-shot classification, retrieval, captioning, and VQA.
Skill Claude CodeCodex
Transformer-based MIL aggregator with Nyström attention + PPEG positional encoding. Models patch-patch correlations at gigapixel scale.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: