Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Unsupervised Causal Prototypical Networks for De-biased Interpretable Dermoscopy Diagnosis

Junhao Jia, Yueyi Wu, Huangwei Chen, Haodong Jing, Haishuai Wang, Jiajun Bu, Lei Wu

TL;DR

CausalProto is proposed, an Unsupervised Causal Prototypical Network that fundamentally purifies the visual evidence chain, and achieves superior diagnostic performance and consistently outperforms standard black box models, while simultaneously providing transparent and high purity visual interpretability without suffering from the traditional accuracy compromise.

Abstract

Despite the success of deep learning in dermoscopy image analysis, its inherent black-box nature hinders clinical trust, motivating the use of prototypical networks for case-based visual transparency. However, inevitable selection bias in clinical data often drives these models toward shortcut learning, where environmental confounders are erroneously encoded as predictive prototypes, generating spurious visual evidence that misleads medical decision-making. To mitigate these confounding effects, we propose CausalProto, an Unsupervised Causal Prototypical Network that fundamentally purifies the visual evidence chain. Framed within a Structural Causal Model, we employ an Information Bottleneck-constrained encoder to enforce strict unsupervised orthogonal disentanglement between pathological features and environmental confounders. By mapping these decoupled representations into independent prototypical spaces, we leverage the learned spurious dictionary to perform backdoor adjustment via do-calculus, transforming complex causal interventions into efficient expectation pooling to marginalize environmental noise. Extensive experiments on multiple dermoscopy datasets demonstrate that CausalProto achieves superior diagnostic performance and consistently outperforms standard black box models, while simultaneously providing transparent and high purity visual interpretability without suffering from the traditional accuracy compromise.

Unsupervised Causal Prototypical Networks for De-biased Interpretable Dermoscopy Diagnosis

TL;DR

CausalProto is proposed, an Unsupervised Causal Prototypical Network that fundamentally purifies the visual evidence chain, and achieves superior diagnostic performance and consistently outperforms standard black box models, while simultaneously providing transparent and high purity visual interpretability without suffering from the traditional accuracy compromise.

Abstract

Despite the success of deep learning in dermoscopy image analysis, its inherent black-box nature hinders clinical trust, motivating the use of prototypical networks for case-based visual transparency. However, inevitable selection bias in clinical data often drives these models toward shortcut learning, where environmental confounders are erroneously encoded as predictive prototypes, generating spurious visual evidence that misleads medical decision-making. To mitigate these confounding effects, we propose CausalProto, an Unsupervised Causal Prototypical Network that fundamentally purifies the visual evidence chain. Framed within a Structural Causal Model, we employ an Information Bottleneck-constrained encoder to enforce strict unsupervised orthogonal disentanglement between pathological features and environmental confounders. By mapping these decoupled representations into independent prototypical spaces, we leverage the learned spurious dictionary to perform backdoor adjustment via do-calculus, transforming complex causal interventions into efficient expectation pooling to marginalize environmental noise. Extensive experiments on multiple dermoscopy datasets demonstrate that CausalProto achieves superior diagnostic performance and consistently outperforms standard black box models, while simultaneously providing transparent and high purity visual interpretability without suffering from the traditional accuracy compromise.
Paper Structure (17 sections, 8 equations, 3 figures, 2 tables)

This paper contains 17 sections, 8 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. SCM Formulation and Framework Architecture
  4. Unsupervised Disentanglement via Information Bottleneck
  5. Causal Prototypical Metric for Interpretable Reasoning
  6. Backdoor Adjustment via do-calculus
  7. Overall Objective
  8. Experiments
  9. Experimental Setup
  10. Datasets and metrics:
  11. Implementation details:
  12. Baselines:
  13. Results and Analysis
  14. Comparative analysis:
  15. Ablation study:
  16. ...and 2 more sections

Figures (3)

  • Figure 1: The underlying Structural Causal Model and performance overview.
  • Figure 2: The framework of CausalProto. (a) Training Phase: A dual-branch network disentangles causal ($Z_C$) and spurious ($Z_S$) features via mutual information minimization. Distinct causal ($P_C$) and spurious ($P_S$) prototype spaces are optimized for interpretable representation learning. (b) Inference Phase: Causal backdoor adjustment is performed by marginalizing over the learned spurious prototype dictionary ($P_S$) via NWGM pooling to achieve de-biased interventional prediction $P(Y|do(X))$.
  • Figure 3: Qualitative visualization of CausalProto's de-confounded reasoning.