Paper

PHARAOH: A collaborative crowdsourcing platform for PHenotyping And Regional Analysis Of Histology

Done, Susan, Alrumeh, Assem Saleh, Duan, Xianpi, Liu, Mugeng, Cotau, Raul, Alyousef, Mohammed J., Echelard, Philippe, Marins, Lidiane Vieira, Yousef, George M., Chen, Min Li, Alsafwani, Noor Said, Zadeh, Parsa Babaei, Alwelaie, Yazeed, Aljohani, Safiyh S., Duan, Xianzhao, Richer, Maxime, Ahangari, Narges, Nikzad, Nikfar, Oreopoulos, Dimitrios, Kamski-Hennekam, Evelyn Rose, Sugden, Richard J., Castillo, Vincent Francis, Leon, Alberto J., Murphy, Patrick, Mikhail, Marly, Diamandis, Phedias, Faust, Kevin, Jakate, Kiran, Saleeb, Rola

2024

Overview

Deep learning has proven to be capable of automating key aspects of histopathologic analysis, but its continual reliance on large expert-annotated training datasets hinders widespread adoption. Here, we present an online collaborative portal that streamlines tissue image annotation to promote the development and sharing of custom computer vision models for PHenotyping And Regional Analysis Of Histology (PHARAOH; https://www.pathologyreports.ai/). PHARAOH uses a weakly supervised active learning framework whereby patch-level image features are leveraged to organize large swaths of tissue into morphologically-uniform clusters for batched human annotation. By providing cluster-level labels on only a handful of cases, we show how custom PHARAOH models can be developed and used to guide the quantification of cellular features that correlate with molecular, pathologic and patient outcome data. Both custom model design and feature extraction pipelines are amenable to crowdsourcing making PHARAOH a fully scalable systems-level solution for the systematic expansion and cataloging of computational pathology applications.

• Publisher: Cold Spring Harbor Laboratory
• Subject: Bioinformatics