AI can predict parts of your genome from just a histology image

Deep learning methods can allow us to predict molecular features from histological images alone. Owkin have developed a model called HE2RNA that provides virtual spatialization of gene expression on a digital pathology slide - it predicts the transcriptome of a piece of tissue on a slide, without the need for expensive spatial transcriptomics. Owkin’s MISO is a more recent, even more powerful version of this tool.

Predicting gene expression and genomic mutation in tumors from whole slide images (WSIs) using technology like this would greatly facilitate patient diagnosis, prediction of response to treatment, and survival outcome. It can also help pharmaceutical research by identifying novel biomarkers in images of specific tumoral regions that are important to better understand disease evolution and differentiated outcomes.