Less Is More? Selective Visual Attention to High-Importance Regions for Multimodal Radiology Summarization

TL;DR

This paper shows that selectively focusing on high-importance visual regions improves multimodal radiology report summarization, challenging the belief that more visual data is always better.

Contribution

The authors introduce ViTAS, a novel multi-stage pipeline that effectively fuses visual and textual data, achieving state-of-the-art results in radiology summarization.

Findings

ViTAS achieves 29.25% BLEU-4 and 69.83% ROUGE-L scores.

Selective visual attention improves factual alignment and human evaluation scores.

Focusing on relevant visual patches outperforms using full images.

Abstract

Automated radiology report summarization aims to distill verbose findings into concise clinical impressions, but existing multimodal models often struggle with visual noise and fail to meaningfully improve over strong text-only baselines in the FINDINGS $\to$ IMPRESSION transformation. We challenge two prevailing assumptions: (1) that more visual input is always better, and (2) that multimodal models add limited value when findings already contain rich image-derived detail. Through controlled ablations on MIMIC-CXR benchmark, we show that selectively focusing on pathology-relevant visual patches rather than full images yields substantially better performance. We introduce ViTAS, Visual-Text Attention Summarizer, a multi-stage pipeline that combines ensemble-guided MedSAM2 lung segmentation, bidirectional cross-attention for multi-view fusion, Shapley-guided adaptive patch clustering, and hierarchical visual tokenization feeding a ViT. ViTAS achieves SOTA results with 29.25% BLEU-4 and 69.83% ROUGE-L, improved factual alignment in qualitative analysis, and the highest expert-rated human evaluation scores. Our findings demonstrate that less but more relevant visual input is not only sufficient but superior for multimodal radiology summarization.