When to Trust the Answer: Question-Aligned Semantic Nearest Neighbor Entropy for Safer Surgical VQA

TL;DR

This paper introduces QA-SNNE, a novel uncertainty estimation method for surgical VQA that explicitly accounts for question-answer alignment, improving safety by reducing overconfidence in irrelevant answers.

Contribution

QA-SNNE incorporates question relevance into semantic entropy, enhancing uncertainty estimation in surgical VQA models, especially under language variation and out-of-template questions.

Findings

QA-SNNE improves AUROC by up to 21% on EndoVis18-VQA for some models.

QA-SNNE enhances robustness to question rephrasing in surgical VQA datasets.

The method is model-agnostic and effective in zero-shot and PEFT settings.

Abstract

Safety and reliability are critical for deploying visual question answering (VQA) systems in surgery, where incorrect or ambiguous responses can cause patient harm. A key limitation of existing uncertainty estimation methods, such as Semantic Nearest Neighbor Entropy (SNNE), is that they do not explicitly account for the conditioning question. As a result, they may assign high confidence to answers that are semantically consistent yet misaligned with the clinical question, especially under variation in question phrasing. We propose Question-Aligned Semantic Nearest Neighbor Entropy (QA-SNNE), a black-box uncertainty estimator that incorporates question-answer alignment into semantic entropy through bilateral gating. QA-SNNE measures uncertainty by weighting pairwise semantic similarities among sampled answers according to their relevance to the question, using embedding-based, entailment-based, or cross-encoder alignment strategies. To assess robustness to language variation, we construct an out-of-template rephrased version of a benchmark surgical VQA dataset, where only the question wording is modified while images and ground-truth answers remain unchanged. We evaluate QA-SNNE on five VQA models across two benchmark surgical VQA datasets in both zero-shot and parameter-efficient fine-tuned (PEFT) settings, including out-of-template questions. QA-SNNE improves AUROC on EndoVis18-VQA for two of three zero-shot models in-template (e.g., +15% for Llama3.2 and +21% for Qwen2.5) and achieves up to +8% AUROC improvement under out-of-template rephrasing, with mixed results on external validation. Overall, QA-SNNE provides a practical, model-agnostic safeguard for surgical VQA by linking semantic uncertainty to question relevance.