Annals of anatomy = Anatomischer Anzeiger : official organ of the Anatomische Gesellschaft · 2026
Deep-learning-based assessment of postoperative nasolabial morphology in unilateral cleft lip repair.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Šimánek Petr, Keil Christiane, Schröder Tom Alexander, Lauer Günter, Buckova Michaela
Аннотация
BACKGROUND: Objective assessment of postoperative facial appearance following unilateral cleft lip and palate (UCLP) repair remains limited by subjective evaluation and operator-dependent morphometric methods. This study developed and clinically validated a fully automated deep-learning-based framework for objective assessment of postoperative nasolabial morphology.
METHODS: Standardized frontal photographs of postoperative UCLP patients (n = 39) and healthy controls (n = 57) were analyzed. Anatomically defined regions of interest of the nose, upper lip, and combined nasolabial complex were extracted using the MediaPipe 468-point facial landmark model. Diffusion-based generative inpainting was applied to generate individualized healthy reference appearances. Original and generated images were compared using semantic, perceptual, and geometric metrics, including DINOv2 feature distances, LPIPS, SSIM, and landmark-derived symmetry parameters.
RESULTS: Postoperative UCLP patients demonstrated significantly greater deviation from healthy facial appearance patterns compared with controls across multiple metrics. DINOv2 feature distances consistently differentiated postoperative patients from controls, indicating persistent semantic deviations in nasolabial morphology following cleft repair. Landmark-based geometric analysis additionally demonstrated significantly increased residual lip and nasal asymmetry in the postoperative cohort. The differential behavior of the generative model between healthy and postoperative faces further suggested that the magnitude of generative correction reflects residual postoperative deformity.
CONCLUSION: The proposed framework enables automated and observer-independent assessment of postoperative nasolabial morphology following UCLP repair. By integrating anatomy-informed landmark analysis with AI-driven morphometric evaluation, the method provides a scalable tool for retrospective outcome analysis and standardized comparison of surgical results in craniofacial surgery.
METHODS: Standardized frontal photographs of postoperative UCLP patients (n = 39) and healthy controls (n = 57) were analyzed. Anatomically defined regions of interest of the nose, upper lip, and combined nasolabial complex were extracted using the MediaPipe 468-point facial landmark model. Diffusion-based generative inpainting was applied to generate individualized healthy reference appearances. Original and generated images were compared using semantic, perceptual, and geometric metrics, including DINOv2 feature distances, LPIPS, SSIM, and landmark-derived symmetry parameters.
RESULTS: Postoperative UCLP patients demonstrated significantly greater deviation from healthy facial appearance patterns compared with controls across multiple metrics. DINOv2 feature distances consistently differentiated postoperative patients from controls, indicating persistent semantic deviations in nasolabial morphology following cleft repair. Landmark-based geometric analysis additionally demonstrated significantly increased residual lip and nasal asymmetry in the postoperative cohort. The differential behavior of the generative model between healthy and postoperative faces further suggested that the magnitude of generative correction reflects residual postoperative deformity.
CONCLUSION: The proposed framework enables automated and observer-independent assessment of postoperative nasolabial morphology following UCLP repair. By integrating anatomy-informed landmark analysis with AI-driven morphometric evaluation, the method provides a scalable tool for retrospective outcome analysis and standardized comparison of surgical results in craniofacial surgery.