A multi-task deep learning framework for simultaneous prediction of microsatellite instability and tumor mutational burden in gastric cancer from histopathological images

1

A multi-task deep learning framework was developed to predict microsatellite instability (MSI) and tumor mutational burden (TMB) from histopathological images.
2

The model integrated whole slide images and clinical data, achieving AUC values of 0.828 for MSI and 0.836 for TMB on the internal test set.
3

External validation showed moderate performance decrease, with AUCs of 0.78 for MSI and 0.74 for TMB, indicating challenges in generalizability.
4

Attention heatmaps provided interpretability, revealing spatial concordance between predictive regions for MSI and TMB.
5

This framework offers a cost-effective screening tool for MSI and TMB, potentially enhancing precision oncology in gastric cancer management.

Frontiers In Oncology

by Yazhou Chang, Haoyue Chang, Yaping Lv, Shuxue Xi, Jialiang Yang, Bingzhi Wang, Xiaohao Zheng, Yibin Xie
June 8, 2026

1