Hair Segmentation and Skin Characterization on Macroscopic Dermatological Images for Laser Treatment Parameterization

TL;DR

The paper introduces a deep learning system combining YOLOv9-based instance segmentation with classical image processing to extract objective skin and hair features from macroscopic dermatological images. It enables accurate phototype classification, hair density, color, and thickness estimation under varying lighting using spectrophotometric calibration.

Why It Matters

Automated parameterization of laser treatments reduces reliance on clinician subjectivity and uncalibrated tools, improving safety and consistency in dermatological procedures. This approach supports broader adoption of AI-driven precision medicine in light-based therapies.

Key Findings

• The system achieves high segmentation accuracy (mAP@50 ≈ 0.80) on short, shaved hairs typical in pre-treatment images.
• Spectrophotometric calibration enables robust feature estimation across diverse illumination conditions.
• The unified pipeline extracts clinically relevant parameters suitable for automated laser treatment planning.