SpaRED benchmark: Enhancing Gene Expression Prediction from Histology Images with Spatial Transcriptomics Completion
This work provides a comprehensive benchmark for gene expression prediction from histology images, addressing inconsistencies in databases and methods, which is crucial for advancing spatial transcriptomics research in biomedical applications.
The paper tackles the challenge of fairly comparing methods for predicting gene expression from histology images by introducing a systematically curated database from 26 public sources, which is 8.6 times larger than previous works, and a transformer-based completion technique that significantly boosts prediction performance across all datasets.
Spatial Transcriptomics is a novel technology that aligns histology images with spatially resolved gene expression profiles. Although groundbreaking, it struggles with gene capture yielding high corruption in acquired data. Given potential applications, recent efforts have focused on predicting transcriptomic profiles solely from histology images. However, differences in databases, preprocessing techniques, and training hyperparameters hinder a fair comparison between methods. To address these challenges, we present a systematically curated and processed database collected from 26 public sources, representing an 8.6-fold increase compared to previous works. Additionally, we propose a state-of-the-art transformer based completion technique for inferring missing gene expression, which significantly boosts the performance of transcriptomic profile predictions across all datasets. Altogether, our contributions constitute the most comprehensive benchmark of gene expression prediction from histology images to date and a stepping stone for future research on spatial transcriptomics.