VGAT: A Cancer Survival Analysis Framework Transitioning from Generative Visual Question Answering to Genomic Reconstruction
This work addresses clinical implementation barriers for cancer survival prediction in resource-constrained settings by enabling genomic-informed inference without sequencing.
The paper tackles the problem of cancer survival analysis in under-resourced regions by proposing VGAT, a framework that uses only whole-slide images to predict survival, outperforming existing WSI-only methods across five TCGA datasets.
Multimodal learning combining pathology images and genomic sequences enhances cancer survival analysis but faces clinical implementation barriers due to limited access to genomic sequencing in under-resourced regions. To enable survival prediction using only whole-slide images (WSI), we propose the Visual-Genomic Answering-Guided Transformer (VGAT), a framework integrating Visual Question Answering (VQA) techniques for genomic modality reconstruction. By adapting VQA's text feature extraction approach, we derive stable genomic representations that circumvent dimensionality challenges in raw genomic data. Simultaneously, a cluster-based visual prompt module selectively enhances discriminative WSI patches, addressing noise from unfiltered image regions. Evaluated across five TCGA datasets, VGAT outperforms existing WSI-only methods, demonstrating the viability of genomic-informed inference without sequencing. This approach bridges multimodal research and clinical feasibility in resource-constrained settings. The code link is https://github.com/CZZZZZZZZZZZZZZZZZ/VGAT.