WEDNESDAY, Aug. 25, 2026 (HealthDay News) -- Combined two-dimensional (2D) segmentation with deep-learning-based three-dimensional (3D) reconstruction improves assessment of alopecia, according to a study published online Aug. 22 in the Journal of the American Academy of Dermatology.Yerin Lee, from Yonsei University Wonju College of Medicine in South Korea, and colleagues developed a 3D model to improve the accuracy and reliability of the Severity of Alopecia Tool (SALT) score by measuring the alopecic area based on surface area and volume. Data were obtained from 679 patients with alopecia areata who received treatment between 2012 and 2018. The scalp and hair loss areas were marked on images taken from four views of the patient.Segmentation performance was assessed by model accuracy for detecting scalp and hair loss regions in clinical photographs. The researchers found that the model achieved an average area under the receiver operating characteristic curve of 0.95 and 0.89 for scalp mask detection and hair loss detection, respectively. The 3D head mesh was reconstructed from the photographs, in alignment with 2D detections, allowing consistent evaluation of hair loss regions across the scalp. Using the aligned mesh, texture maps were generated by projecting the clinical images and detected hair loss areas onto a standardized 2D surface representation of the reconstructed 3D scalp. Alopecia distribution was visualized by applying these maps to 3D mesh. AutoSALT-3D provided individualized assessments using scalp surface area compared to 2D SALT, correcting underestimation."Although 3D reconstruction may introduce inaccuracies, it preserves relative comparisons beyond 2D assessment," the authors write. "Future studies should optimize segmentation and reconstruction accuracy and validate this approach across diverse populations."Abstract/Full Text (subscription or payment may be required).Sign up for our weekly HealthDay newsletter