Uncertainty-Aware Domain Adaptation for Vitiligo Segmentation in Clinical Photographs

Wentao Jiang, Vamsi Varra, Caitlin Perez-Stable, Harrison Zhu, Meredith Apicella, Nicole Nyamongo

Keywords: vitiligo segmentation, clinical photography, skin lesion segmentation, domain-adaptive transfer learning, hard negative mining, uncertainty quantification

Abstract: Accurately quantifying vitiligo extent in routine clinical photographs is crucial for longitudinal monitoring of treatment response. We propose a trustworthy, frequency-aware segmentation framework built on three synergistic pillars: (1) a domain-adaptive pre-training strategy on the ISIC 2019 dermoscopy dataset; (2) an architectural refinement via a ConvNeXt V2-based encoder enhanced with a novel High-Frequency Spectral Gating (HFSC) module and stem-skip connections to capture subtle textures; and (3) a clinical trust mechanism employing K-fold ensemble and Test-Time Augmentation (TTA) to generate pixel-wise uncertainty maps. Extensive validation on an expert-annotated clinical cohort demonstrates superior performance, achieving a Dice score of 85.05\% and significantly reducing boundary error (95\% Hausdorff Distance improved from 44.79 px to 29.95 px), consistently outperforming strong CNN (ResNet-50 and UNetPP) and Transformer (MiT-B5) baselines. Notably, our framework demonstrates high reliability with zero catastrophic failures and provides interpretable entropy maps to identify ambiguous regions for clinician review. This work establishes a robust and practical tool for objective vitiligo assessment. Our approach suggests that the proposed framework is a practical and reliable component for vitiligo management in routine clinical practice.

Primary Subject Area: Segmentation

Secondary Subject Area: Application: Dermatology

Registration Requirement: Yes

Visa & Travel: Yes

Read CFP & Author Instructions: Yes

Originality Policy: Yes

Single-blind & Not Under Review Elsewhere: Yes

LLM Policy: Yes

Submission Number: 250