Deep Learning (DL) can predict biomarkers from cancer histopathology. Several clinically approved applications use this technology. Most approaches, however, predict categorical labels, whereas biomarkers are often continuous measurements. We hypothesized that regression-based DL outperforms classification-based DL. Therefore, we developed and evaluated a new self-supervised attention-based weakly supervised regression method that predicts continuous biomarkers directly from images in 11,671 patients across nine cancer types. We tested our method for multiple clinically and biologically relevant biomarkers: homologous repair deficiency (HRD) score, a clinically used pan-cancer biomarker, as well as markers of key biological processes in the tumor microenvironment. Using regression significantly enhances the accuracy of biomarker prediction, while also improving the interpretability of the results over classification. In a large cohort of colorectal cancer patients, regression-based prediction scores provide a higher prognostic value than classification-based scores. Our open-source regression approach offers a promising alternative for continuous biomarker analysis in computational pathology.
@article{arxiv.2304.05153,
title = {Regression-based Deep-Learning predicts molecular biomarkers from pathology slides},
author = {Omar S. M. El Nahhas and Chiara M. L. Loeffler and Zunamys I. Carrero and Marko van Treeck and Fiona R. Kolbinger and Katherine J. Hewitt and Hannah S. Muti and Mara Graziani and Qinghe Zeng and Julien Calderaro and Nadina Ortiz-Brüchle and Tanwei Yuan and Michael Hoffmeister and Hermann Brenner and Alexander Brobeil and Jorge S. Reis-Filho and Jakob Nikolas Kather},
journal= {arXiv preprint arXiv:2304.05153},
year = {2023}
}