Reducing Hallucinations of Medical Multimodal Large Language Models with Visual Retrieval-Augmented Generation

TL;DR

This paper introduces Visual RAG, a retrieval-augmented generation framework that incorporates visual data to reduce hallucinations in medical multimodal large language models, improving accuracy in clinical report generation.

Contribution

It presents a novel Visual RAG approach that enhances medical MLLMs by integrating visual retrieval, significantly reducing hallucinations and improving entity grounding in clinical reports.

Findings

Improved entity grounding accuracy in medical reports.

Enhanced performance on both frequent and rare medical entities.

Higher RadGraph-F1 score in X-ray report generation.

Abstract

Multimodal Large Language Models (MLLMs) have shown impressive performance in vision and text tasks. However, hallucination remains a major challenge, especially in fields like healthcare where details are critical. In this work, we show how MLLMs may be enhanced to support Visual RAG (V-RAG), a retrieval-augmented generation framework that incorporates both text and visual data from retrieved images. On the MIMIC-CXR chest X-ray report generation and Multicare medical image caption generation datasets, we show that Visual RAG improves the accuracy of entity probing, which asks whether a medical entities is grounded by an image. We show that the improvements extend both to frequent and rare entities, the latter of which may have less positive training data. Downstream, we apply V-RAG with entity probing to correct hallucinations and generate more clinically accurate X-ray reports, obtaining a higher RadGraph-F1 score.