Learn
Follow detailed staged curricula through ML, AI, deep learning, computer vision, NLP, LLMs, natural-image Vision & Language, and the pathology-specific engineering they support. Begin with the learning routes.
Computational Pathology Reading Roadmap
Learn the foundations. Start reading the field in order. Build WSI research that survives careful evaluation.
Maintained · evidence-aware · openly reusable
One map from fundamentals to research practice
Section titled “One map from fundamentals to research practice”This project connects the material that is usually scattered across courses, papers, model cards, challenge sites, and reading groups. Students, researchers, and lab teams can start at an appropriate level and move toward reproducible whole-slide research without losing the assumptions between layers.
3 entry routesFoundational, graduate, and advanced computational-pathology research
WSI-first data guideNative slides, derived patches, reports, outcomes, and spatial data
Daily discoveryTerms-compliant arXiv radar with transparent reading-priority signals
Natural-image Vision & Language
Track image–text representation learning, captioning and VQA, grounding and open vocabulary, multimodal LLMs, generation, video, agents, and their evaluation. Open the field map.
Computational pathology
Follow nine connected chapters from WSI engineering and dense morphology through MIL, SSL, foundation encoders, pathology VLMs, spatial/molecular modeling, and clinical evidence. Open the field map.
WSI Data Atlas
Filter reviewed dataset records by research lane, data unit, organ, WSI relationship, access route, provenance, or task without confusing patches with native WSIs. Open the Atlas.
Paper Radar
Review new arXiv records by topic lane. Scores explain reading priority, never scientific quality or author prestige. Open the provisional Radar.
My Library
Keep paper status, comprehension, review history, and due dates in a reader-controlled library. Opt in to Google Drive for cross-device sync, or retain browser-only storage and portable JSON backup; calendar reminders can be exported separately. Open the reading tracker.
Built for trustworthy extension
Section titled “Built for trustworthy extension”Every curated record is expected to carry primary links, publication or release state, limitations, provenance, and a verification date. Automation can place a record in the provisional Radar; only human review can promote it into a trusted learning path.
“Comprehensive” means coverage across the maintained scientific taxonomy and its major lineages—not every paper ever published. Scope, evidence labels, privacy boundaries, and maintenance rules are documented throughout the resource so that readers and contributors can evaluate how material enters each route.