Oops, looks like you need to register to access this feature.
Select “Keep Exploring” to look around Scholars in Medicine. You have full access to all videos and podcasts for a limited time. Registration is provided at no cost exclusively for healthcare providers.
Sign Up or Log In
SIM - Scholars in Medicine Logo
Login
SIM - Scholars in Medicine Logo
Founding Sponsors
Register

Dermatology

Seeing Smarter: How Explainable AI Boosts Melanoma Detection

Jun 02, 2025

newspaper-banner

AT A GLANCE

In a study of 76 dermatologists evaluating dermoscopic images, explainable AI (XAI)—which provides detailed, transparent reasoning—improved diagnostic accuracy over standard AI by 2.8 percentage points. Eye-tracking data revealed that XAI also influenced cognitive processing, with greater visual effort observed during diagnostic uncertainty and complex cases.

A new study reveals that AI designed to think like a dermatologist can do more than improve diagnostic accuracy—it can change how dermatologists interact with diagnostic tools. In a reader study involving 76 dermatologists, participants were asked to evaluate 16 dermoscopic images of melanomas and nevi using an explainable AI (XAI) system while eye-tracking technology captured their visual behavior. Unlike conventional AI, which provides a prediction without context, XAI offers transparent, domain-specific explanations that clarify how and why a diagnosis is made—enhancing clinical understanding and trust.

The results were compelling: XAI improved dermatologists’ diagnostic balanced accuracy by 2.8 percentage points compared to standard AI. Eye-tracking technology added another layer of insight, showing that when clinicians encountered diagnostic conflicts or complex lesions, they experienced higher cognitive load—indicated by increased eye fixations. These findings suggest that while XAI enhances performance, it also shifts the way clinicians process visual information.

Ultimately, this study highlights the value of pairing interpretability with precision. For dermatology—and potentially across other image-dependent specialties—XAI may be the key to deeper trust and more confident decision-making in AI-assisted diagnostics.

“These insights have significant implications for the design of AI/XAI tools for visual tasks in dermatology and the broader development of XAI in medical diagnostics,” according to the authors.

Reference: Chanda T, Haggenmueller S, Bucher TC, et al. Dermatologist-like explainable AI enhances melanoma diagnosis accuracy: eye-tracking study. Nat Commun. 2025;16(1):4739. Published 2025 May 21. doi:10.1038/s41467-025-59532-5