What question did this study set out to answer?

This research aims to enhance interpretability in clinical AI by minimizing concept leakage in skin lesion diagnosis models.

August 5, 2026Discover Computing0 citationsOpen Access

A vision transformer based leakage free concept bottleneck model for clinical skin lesion diagnosis

KKKailash Chandra Kandpal PVPrabhat Verma

Key Points

This research aims to enhance interpretability in clinical AI by minimizing concept leakage in skin lesion diagnosis models.
Developed LCBM, a leakage-constrained architecture with two subspaces: concept-aligned and residual.
Utilized orthogonality regularization and adversarial Gradient Reversal Layer to decouple latent features.
Evaluated the model on Derm7pt and HAM10000 benchmarks using ResNet and Vision Transformer architectures.
Improved semantic reliability of CBMs, decreasing leakage gap from 0.4487 to 0.0539 with ground-truth concept intervention.
Achieved AUC scores of 0.8358 for ResNet-50 and 0.8107 for ViT-B/16 on Derm7pt.
Reached an AUC of 0.94 on the HAM10000 benchmark.

Abstract

Concept Bottleneck Models (CBMs) aim to improve interpretability by forcing predictions to pass through human-interpretable concepts. However, many CBMs achieve high predictive accuracy by bypassing the bottleneck through latent shortcut features, a phenomenon known as concept leakage. Such behavior weakens the reliability of concept-based explanations, particularly in high-stakes clinical applications. To address this, we propose LCBM (Leakage-Constrained Bottleneck Model), a purified concept bottleneck architecture that decomposes latent representations into two structurally decorrelated subspaces: a concept-aligned semantic space (zₒ₄₌) containing clinically meaningful features and a residual space (zₑ₄ₒ) capturing non-diagnostic artifacts. Through orthogonality regularization and an adversarial Gradient Reversal Layer (GRL), the framework minimizes the propagation of diagnostic information through non-semantic pathways. The framework is evaluated on the Derm7pt and HAM10000 dermatology benchmarks using ResNet and Vision Transformer backbones. Our results demonstrate that LCBM significantly improves the semantic reliability of CBMs, reducing the leakage gap from 0. 4487 to 0. 0539 under ground-truth concept intervention. The model achieves competitive AUC scores of 0. 8358 (ResNet-50) and 0. 8107 (ViT-B/16) on Derm7pt, while reaching an AUC of 0. 94 on HAM10000. These results suggest that structured latent purification through decorrelation can enhance faithful concept-based reasoning in clinical AI systems, providing a more transparent foundation for diagnostic decision support.

Ask AI

Helpful

Bookmark

View Full Paper