aiqbench analysis
WILSON is a vision-language foundation model designed to process whole-slide pathology images and multi-slide patient cases as unified composite inputs, addressing the gap between tile-based encoding and holistic clinical reasoning. Trained on nearly 190,000 slides across 42 organs, it demonstrates superior case-level diagnostic accuracy while requiring significantly less compute than existing slide-level models. This approach offers a compact, clinically aligned computational unit that improves histologic subtyping and diagnostic text generation without needing task-specific fine-tuning for baseline performance.
WILSON - a pathology foundation model framework for patient-level analysis and diagnostic text generation
arXiv:2609.25123v1 Announce Type: cross Abstract: Pathologists integrate morphology across magnifications and across the slides of a patient case, whereas pathology foundation models encode thousands of tiles from single slides and aggregate their features. Here we present WILSON, a vision--language foundation model that represents whole-slide images and multi-slide cases as single multi-magnification composite images, trained on approximately 189k slides from Mayo Clinic spanning 42 organs and 829 diagnostic entities using pathology reports as supervision. Without task-specific training, WILSON exceeded a dedicated case-level model on all internal cohorts (macro-F1 0.52 versus 0.38) and matched slide-level models up to 9.4 times larger at 272- to 2,155-fold lower compute. End-to-end fine-tuning on 508 triple-negative breast cancer cases improved histologic subtyping and stromal tumor-infiltrating lymphocyte grading by 0.16 and 0.11 macro-F1. WILSON retrieved matching diagnostic text at 75.6% recall@1 (PRISM, 58.1%) and generated captions closer to report-derived references than PRISM and PRISM2 on the internal cohort and on most external comparisons. Composite images thus offer a compact, clinically aligned computational unit for pathology.
Authors:Saghir Alfasly, Wataru Uegami, Sobhan Hemati, Wenchao Han, Xiaojia Tang, Kevin Thompson, Daniel Stone, Ghazal Alabtah, Saba Yasir, Michael R. Lucas, Eric W. Klee, Cheryl L. Willman, Judy C. Boughey, Matthew P. Goetz, Krishna R. Kalari, H.R. Tizhoosh
Abstract:Pathologists integrate morphology across magnifications and across the slides of a patient case, whereas pathology foundation models encode thousands of tiles from single slides and aggregate their features. Here we present WILSON, a vision--language foundation model that represents whole-slide images and multi-slide cases as single multi-magnification composite images, trained on approximately 189k slides from Mayo Clinic spanning 42 organs and 829 diagnostic entities using pathology reports as supervision. Without task-specific training, WILSON exceeded a dedicated case-level model on all internal cohorts (macro-F1 0.52 versus 0.38) and matched slide-level models up to 9.4 times larger at 272- to 2,155-fold lower compute. End-to-end fine-tuning on 508 triple-negative breast cancer cases improved histologic subtyping and stromal tumor-infiltrating lymphocyte grading by 0.16 and 0.11 macro-F1. WILSON retrieved matching diagnostic text at 75.6% recall@1 (PRISM, 58.1%) and generated captions closer to report-derived references than PRISM and PRISM2 on the internal cohort and on most external comparisons. Composite images thus offer a compact, clinically aligned computational unit for pathology.
Quantitative Methods (q-bio.QM); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
arXiv:2609.25123 [q-bio.QM]
(or arXiv:2609.25123v1 [q-bio.QM] for this version)
https://doi.org/10.48550/arXiv.2609.25123 arXiv-issued DOI via DataCite (pending registration)
Full text under the publisher's open licence · 236 words
Related coverage
News
Quantum Computing Report · Jun 7, 2026
A Deployment-Friendly Foundational Framework for Efficient Computational Pathology
arXiv · Aug 5, 2026
FedHisto-PAST: Parameter-Efficient Stain-Aware Federated Learning for Cross-Site Lung Histopathology Classification
arXiv · Sep 28, 2026
A multi-agent system for spine MRI report generation from multi-sequence imaging
arXiv · Jun 9, 2026