Human-interpretable image features derived from densely mapped cancer pathology slides predict diverse molecular phenotypes

TLDR

Computational methods have advanced pathology workflows, yet lack of interpretability remains a barrier to clinical integration. We present an approach that predicts clinically relevant molecular phenotypes from whole‑slide histopathology images using human‑interpretable image features (HIFs). Using >1.6 million pathologist annotations on >5 700 samples, we train deep‑learning models to classify cells and tissues at 2–4 µm resolution, then combine the outputs into 607 biologically relevant HIFs across five cancer types. The HIFs correlate with tumor‑microenvironment markers and predict diverse molecular signatures (AUROC 0.601–0.864), including four immune‑checkpoint proteins and homologous recombination deficiency, with performance comparable to black‑box methods, providing a quantitative, interpretable view of tumor composition and spatial architecture.

Abstract

Abstract Computational methods have made substantial progress in improving the accuracy and throughput of pathology workflows for diagnostic, prognostic, and genomic prediction. Still, lack of interpretability remains a significant barrier to clinical integration. We present an approach for predicting clinically-relevant molecular phenotypes from whole-slide histopathology images using human-interpretable image features (HIFs). Our method leverages >1.6 million annotations from board-certified pathologists across >5700 samples to train deep learning models for cell and tissue classification that can exhaustively map whole-slide images at two and four micron-resolution. Cell- and tissue-type model outputs are combined into 607 HIFs that quantify specific and biologically-relevant characteristics across five cancer types. We demonstrate that these HIFs correlate with well-known markers of the tumor microenvironment and can predict diverse molecular signatures (AUROC 0.601–0.864), including expression of four immune checkpoint proteins and homologous recombination deficiency, with performance comparable to ‘black-box’ methods. Our HIF-based approach provides a comprehensive, quantitative, and interpretable window into the composition and spatial architecture of the tumor microenvironment.