Discovery of molecular features underlying morphological landscape by integrating spatial transcriptomic data with deep features of tissue image

Abstract

Profiling molecular features associated with the morphological landscape of tissue is crucial to interrogate structural and spatial patterns that underlie biological function of tissues. Here, we present a new method, SPADE, to identify important genes associated with morphological contexts by combining spatial transcriptomic data with co-registered images. SPADE incorporates deep learning-derived image patterns with spatially resolved gene expression data to extract morphological context markers. Morphological features that correspond to spatial maps of transcriptome were extracted by image patches surrounding each spot, and then, represented by image latent features. The molecular profiles correlated with the image latent features were identified. The extracted genes could be further analyzed to discover functional terms and also exploited to extract clusters maintaining morphological contexts. We apply our approach to spatial transcriptomic data of three different tissues to suggest an unbiased method capable of offering image-integrated gene expression trends.