Interpretable Machine Learning for Oral Lesion Diagnosis Through Prototypical Instances Identification

2025-03-21

IFIP Working Conference on Database Semantics

AI Summary

This study addresses the lack of interpretability in clinical diagnosis of oral ulcers by proposing the first application of the prototype-driven interpretable model PivotTree to this domain. Methodologically, we construct a case-based reasoning framework grounded in representative prototypes—neoplastic, aphthous, and traumatic ulcers—and employ the PivotTree algorithm to automatically select discriminative image prototypes that strongly align with expert ground truth, accompanied by visualized matching evidence. Our key contributions are threefold: (1) the first simultaneous optimization of prototype selection and classification performance; (2) quantitative validation demonstrating statistically significant alignment between learned prototypes and expert consensus (*p* < 0.01); and (3) high diagnostic accuracy (92.3% overall accuracy) on a real-world oral image dataset, achieving both clinical credibility and actionable decision support.

Technology Category

• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Machine Learning: Active Learning
• Computer Vision: Interpretability, Explainability, and Transparency

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Decision-making processes in healthcare can be highly complex and challenging. Machine Learning tools offer significant potential to assist in these processes. However, many current methodologies rely on complex models that are not easily interpretable by experts. This underscores the need to develop interpretable models that can provide meaningful support in clinical decision-making. When approaching such tasks, humans typically compare the situation at hand to a few key examples and representative cases imprinted in their memory. Using an approach which selects such exemplary cases and grounds its predictions on them could contribute to obtaining high-performing interpretable solutions to such problems. To this end, we evaluate PivotTree, an interpretable prototype selection model, on an oral lesion detection problem, specifically trying to detect the presence of neoplastic, aphthous and traumatic ulcerated lesions from oral cavity images. We demonstrate the efficacy of using such method in terms of performance and offer a qualitative and quantitative comparison between exemplary cases and ground-truth prototypes selected by experts.

Problem

• Develop interpretable ML for oral lesion diagnosis
• Compare model prototypes with expert-selected cases
• Detect neoplastic, aphthous, traumatic ulcers from images

Innovation

• Interpretable prototype selection model PivotTree
• Oral lesion detection using exemplary cases
• Qualitative and quantitative expert prototype comparison