VL-AcneSeg: A Vision-Language Framework for Region-Aware Acne Lesion Segmentation

2026-09-28

AI Summary

This study addresses the subjectivity of conventional acne assessment, its neglect of lesion size, and the lack of spatial priors in existing segmentation models by proposing a multimodal acne analysis framework based on vision-language models. The method leverages CLIP with regional text prompts to incorporate spatial prior knowledge, enabling precise localization and segmentation of facial lesions. Furthermore, it introduces a novel single global prompt protocol that operates without positional information, combined with area-based severity quantification to support cross-domain generalization on unconstrained smartphone images. Experimental results demonstrate that the proposed model achieves a Dice coefficient of 0.5082, outperforming existing methods. Its area measurements exhibit correlation with expert annotations comparable to established standards, while maintaining robust performance across diverse datasets.

Abstract

Acne assessment is crucial for clinical decision-making, yet traditional grading and counting are subjective and fail to account for lesion size. While area-based assessment has emerged as a promising alternative, acne segmentation has continued to rely on general-purpose architectures. To address this gap, we propose VL-AcneSeg, a multimodal framework for acne lesion segmentation that leverages CLIP and region-level text prompts to incorporate spatial priors, enabling lesions to be localized across the whole face. Because region-level prompts indicate which facial areas contain lesions, we report a single global prompt, which requires no such information, as our primary setting. On our internal clinical dataset, VL-AcneSeg achieves a Dice score of 0.5082 and an IoU of 0.3407 under this protocol, the highest among all compared methods, including recent vision-language segmentation methods that are themselves given region-level prompts; region-level prompting raises these to 0.5296 and 0.3602. Moreover, lesion area measurements derived from our segmentation correlate with IGA scores at a level comparable to expert annotations (Pearson r = 0.719 versus 0.658). Notably, our framework maintains consistent performance across external validation datasets, performing reliably even on uncontrolled smartphone images without requiring additional training or fine-tuning. By pairing a protocol that requires no lesion-location information with area-based severity estimation, this work provides a foundation for objective acne assessment outside the clinic. Our implementation is publicly available at: https://github.com/sukjuoh/VL-AcneSeg

Problem

• acne segmentation
• lesion assessment
• area-based evaluation
• region-aware

Innovation

• Vision-Language Framework
• Acne Lesion Segmentation
• CLIP
• Region-level Prompts
• Cross-domain Generalization