IViT: A Novel Interpretable Visual Transformer for Skin Disease Detection

2026-06-22

AI Summary

This work addresses the challenges in skin disease diagnosis caused by high inter-class similarity among lesions, where existing Vision Transformer approaches suffer from poor interpretability, limited few-shot adaptability, and a trade-off between accuracy and explainability. To overcome these limitations, we propose an Interpretable Vision Transformer (IViT) constrained by quadratic programming. Our method leverages pre-trained transfer learning to enhance feature extraction under few-shot settings and employs discrete quadratic programming to select clinically plausible discriminative features. A multi-objective loss function is designed to reduce feature redundancy and optimize activation distributions. Evaluated on six benchmark datasets, IViT achieves an accuracy of 93.80%—only 0.21% lower than the baseline—while reducing feature redundancy by 29.5% and producing core activation regions that closely align with clinically relevant lesion areas, thereby significantly improving model transparency with minimal performance cost.

Abstract

The clinical diagnosis of skin diseases is susceptible to interference from inter-class similarity of skin lesions, and over-reliance on clinicians'experience easily leads to subjective bias. Although existing deep learning aided diagnosis methods achieve competitive accuracy, they suffer from the black-box opacity of Vision Transformer (ViT) and poor adaptability to medical few-shot scenarios. Moreover, mainstream explainable algorithms generally face the bottleneck of significant accuracy degradation when improving interpretability. This paper proposes an interpretable ViT (IViT) constrained by Quadratic Programming (QP). The introduced pre-trained transfer learning adapts to few-shot feature extraction. A discrete QP feature selection framework is constructed to screen generic and discriminative features consistent with clinical diagnostic logic. A multi-objective loss function is designed to reduce feature redundancy and optimize activation distribution while preserving classification performance. Experimental results on six standard skin disease datasets show that IViT achieves an accuracy of 93.80%, only 0.21% lower than the baseline, with feature redundancy reduced by 29.5%. Its core activation regions are consistent with clinically concerned lesion areas. The proposed model balances accuracy and interpretability, providing a reliable solution for the clinical deployment of few-shot intelligent skin disease diagnosis.

Problem

• skin disease diagnosis
• interpretability
• few-shot learning
• Vision Transformer
• accuracy-degradation bottleneck

Innovation

• Interpretable Vision Transformer
• Quadratic Programming
• Few-shot Learning
• Feature Selection
• Multi-objective Loss