Layer-Specific Fine-Tuning for Improved Negation Handling in Medical Vision-Language Models

TL;DR

This paper introduces a layer-specific fine-tuning method called NAST that enhances medical vision-language models' ability to correctly interpret negation in clinical reports, using causal interpretability to guide targeted learning.

Contribution

It presents a novel interpretability-guided fine-tuning approach that modulates layer updates based on causal contributions to improve negation handling in medical VLMs.

Findings

NAST improves negation discrimination in clinical statements.

The method maintains overall vision-language alignment.

Causal interpretability effectively guides targeted model adaptation.

Abstract

Negation is a fundamental linguistic operation in clinical reporting, yet vision-language models (VLMs) frequently fail to distinguish affirmative from negated medical statements. To systematically characterize this limitation, we introduce a radiology-specific diagnostic benchmark that evaluates polarity sensitivity under controlled clinical conditions, revealing that common medical VLMs consistently confuse negated and non-negated findings. To enable learning beyond simple condition absence, we further construct a contextual clinical negation dataset that encodes structured claims and supports attribute-level negations involving location and severity. Building on these resources, we propose Negation-Aware Selective Training (NAST), an interpretability-guided adaptation method that uses causal tracing effects (CTEs) to modulate layer-wise gradient updates during fine-tuning. Rather than applying uniform learning rates, NAST scales each layer's update according to its causal contribution to negation processing, transforming mechanistic interpretability signals into a principled optimization rule. Experiments demonstrate improved discrimination of affirmative and negated clinical statements without degrading general vision-language alignment, highlighting the value of causal interpretability for targeted model adaptation in safety-critical medical settings. Code and resources are available at https://github.com/healthylaife/NAST.