NeRD: Neuro-Symbolic Rule Distillation for Efficient Ontology-Grounded Chain-of-Thought in Medical Image Diagnosis

2026-06-14

AI Summary

This work addresses the limitations of existing explainable methods in medical image analysis, which often rely on all predefined concepts or discriminative features misaligned with clinical ontologies, leading to redundant and medically implausible explanations. To overcome this, the authors propose NeRD, a neuro-symbolic rule distillation framework that, for the first time, integrates multimodal chain-of-thought reasoning with expert-in-the-loop intervention to automatically generate diagnosis pathways that are ontologically grounded, non-redundant, and amenable to human intervention—without requiring manually crafted rules. By synergizing neural and symbolic reasoning, NeRD achieves state-of-the-art diagnostic performance on two dermatological datasets. Blind expert evaluations confirm that its reasoning rationales are clinically sound and demonstrate the efficacy and efficiency of concept-level interventions.

Abstract

Interpretability is essential for trustworthy medical image diagnosis. However, existing concept-driven interpretable methods have key limitations: Concept Bottleneck Models (CBMs) require scoring all predefined concepts at inference time and for manual intervention, imposing a substantial burden on clinicians, while rationale-based generative approaches often select concepts by class discriminability, which can drift from diagnostic ontologies. To address these issues, we propose Neuro-Symbolic Rule Distillation (NeRD), a framework that produces efficient, ontology-grounded reasoning chains that are sufficient yet non-redundant, without manually crafting diagnostic rules. Experiments on two skin datasets demonstrate strong diagnostic performance and interpretability, and blinded expert evaluation confirms the clinical plausibility of NeRD rationales. Our method further enables a first expert-in-the-loop study for Multimodal Chain-of-Thought-based diagnosis, achieving efficient and effective concept-level intervention.

Problem

• interpretability
• medical image diagnosis
• concept-driven methods
• diagnostic ontologies
• clinical burden

Innovation

• Neuro-Symbolic
• Rule Distillation
• Ontology-Grounded Reasoning
• Chain-of-Thought
• Medical Image Diagnosis