Advancing multi-modal research on the Cancer Genomics Cloud: deep learning applications of a pan-cancer PDX histology image repository

For our September webinar, Velsera scientist Dr. Phil Webster will present a case study in conducting multimodal research on the Cancer Genomics Cloud recently submitted for publication by the PDXNet consortium. The consortium demonstrated joint imaging and genomic analysis on the CGC by developing a genomics data repository, developing a haemotoxylin-eosin stain (H&E) image repository, and collaborating to bring new imaging tools to the CGC. Extensive studies of clinical H&E images demonstrating that deep learning can identify inter-cellular and morphological signals correlated with disease phenotype and therapeutic response motivated the Patient-Derived Xenograft (PDX) case study.  

In this webinar, we will present results demonstrating that deep learning can be applied to PDX H&E images to distinguish neoplastic, stromal, and necrotic regions and predict xenograft-transplant lymphoproliferative disorder. Lastly, we will present the strategies for making genomics and imaging repositories publicly available and the analytical tools used in the analysis. 

About the speaker

Phil Webster, Ph.D.

Dr. Phil Webster received his B.S. in Cell and Molecular Biology from San Diego State University, and completed his Ph.D. in Cell & Structural Biology at the University of Texas Health Science Center in San Antonio, TX. His research interests include functional genomics, metabolic dysfunction, cell signalling pathways, and the biology of aging. As a Data Scientist at Velsera, Dr. Webster has worked with various consortia including PDXnet to develop and implement deep learning algorithms for the analysis of data from patient-derived xenografts.