🤖 AI Summary
Current AI models for skin cancer diagnosis lack interpretability and fail to support clinical hypothesis validation. Method: This paper proposes a hypothesis-driven visual assessment AI system that takes dermoscopic images as input and enables clinicians to specify clinical hypotheses (e.g., “malignant melanoma”). Leveraging CLIP-style concept alignment, Concept Activation Mapping (CAM), and Bayesian evidence quantification, the system automatically identifies and quantifies high-level semantic evidence supporting or contradicting the hypothesis, outputting an interpretable Weight of Evidence (WoE). Contribution/Results: It pioneers the deep integration of hypothesis reasoning with concept-level interpretability, establishing a dynamic, verifiable visual decision-support paradigm—overcoming limitations of conventional black-box classification and static saliency maps. Experiments demonstrate significant improvements in clinicians’ diagnostic confidence and inter-rater consistency, while maintaining compatibility with diverse concept-based explanation mechanisms.
📝 Abstract
This paper presents Visual Evaluative AI, a decision aid that provides positive and negative evidence from image data for a given hypothesis. This tool finds high-level human concepts in an image and generates the Weight of Evidence (WoE) for each hypothesis in the decision-making process. We apply and evaluate this tool in the skin cancer domain by building a web-based application that allows users to upload a dermatoscopic image, select a hypothesis and analyse their decisions by evaluating the provided evidence. Further, we demonstrate the effectiveness of Visual Evaluative AI on different concept-based explanation approaches.