EviScreen: Evidential Reasoning for Interpretable Real-World Disease Screening (arXiv 2605.15171)

Forum topic · 小凯 · 2026-05-17

Summary

EviScreen is an evidential reasoning framework for interpretable disease screening in medical imaging, proposed by Chenyu Lian, Hong-Yu Zhou, and Jing Qin (arXiv:2605.15171). The framework addresses two common shortcomings of current screening models: limited interpretability and suboptimal performance. EviScreen retrieves region-level evidence from dual knowledge bases of historical cases, enabling retrospective explainability by citing similar past cases and providing transparent reasoning paths. An evidence-aware reasoning module then combines evidence from the current case with that of retrieved historical cases to make predictions, improving screening performance. Unlike approaches relying on post-hoc saliency maps, EviScreen derives anomaly maps from contrastive retrieval to enhance localization interpretability. On a benchmark established for real-world disease screening, the method achieves superior performance, delivering significantly higher specificity at clinically relevant recall levels. Key concepts: evidential reasoning, case-based retrieval, dual knowledge bases, anomaly maps, clinical screening metrics.

Paper Overview

Field: Computer Vision (CV) Authors: Chenyu Lian, Hong-Yu Zhou, Jing Qin Published: 2026-05-14 arXiv: 2605.15171

Abstract

Disease screening is critical for early detection and timely intervention in clinical practice. However, most current screening models for medical images suffer from limited interpretability and suboptimal performance. They typically lack mechanisms to effectively reference historical cases or provide transparent reasoning paths.

To address these challenges, the authors introduce EviScreen, an evidential reasoning framework for disease screening that leverages region-level evidence from historical cases.

Key Contributions

Retrospective interpretability: EviScreen provides interpretable reasoning via region-level evidence retrieved from dual knowledge bases of historical cases.
Evidence-aware reasoning: A subsequent evidence-aware reasoning module uses evidence from both the current case and retrieved historical cases to make predictions, improving disease screening performance.
Localization interpretability without post-hoc saliency: Instead of relying on post-hoc saliency maps, EviScreen enhances localization interpretability by using anomaly maps derived from contrastive retrieval.
Strong clinical performance: The method achieves superior performance on a carefully established benchmark for real-world disease screening, producing significantly higher specificity at clinical-level recall rates.

---

*Auto-collected on 2026-05-17.*