Rapid Prediction of NSCLC Genomic Alterations from Routine H&E Images
Authors:
Inbal Gazy, Eran Shmuel, Jonathan Zalach. Imagene AI
Leica Biosystems Pathology Forward Summit, 2026
This poster presents Imagene AI’s NSCLC multi-gene biomarker panel, developed to predict eight genomic alterations directly from routine H&E whole-slide images. The panel was evaluated using approximately 8,000 samples from 10 source sites, with a nested cross-validation framework that held out source sites for testing. The poster details the panel’s performance across mutations, fusions, and exon skipping, and explores the potential of H&E-based biomarker prediction to support research sample prioritization, retrospective biomarker analysis, and investigational clinical trial prescreening without requiring additional tissue.
Fill out this form to have the publication sent to you