Biomarker prediction from images of histological slides of cancerous tissues using modern AI techniques: comparisons of different neural architectures

Project Summary

Our project involves linking digital pathology and the genomic analysis of tumors, particularly for immunotherapy applications. These images are very large and have multiple resolutions. The Digital Pathology group at SIP/LIPADE has settled strong collaborations with the medical community and translational results are published in Lancet Oncology 2023 for the prediction of genomic signature out of tissue images, and ongoing 2024 on the positive result of an in silico test for stratifying MSS patients for immunotherapy in a clinical phase 2 trial. We are now exploring the new foundation models and self-supervised strategies for multi-modal learning.

Nicolas Loménie
nicolas.lomenie@u-paris.fr

• Chair In.CU.bator at Circle U. / Université Paris Cité
• Full Professor at Université Paris Cité (Faculté des Sciences)

Zhuxian Guo
zhuxian.guo@hotmail.com
Camille Kurtz
camille.kurtz@u-paris.fr
Julien Calderaro