Journal of dentistry · 2026
Facial surface tracing approach to reduce the observation errors in 3D curvilinear landmark annotations on cone-beam computed tomography images.
СтатьяВалидационное исследованиеНазвание 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.
Mangal Utkarsh, Jinn Hyeonseok, Hyun Chang Min, Kim Harim, Cha Jung Yul, Lee Kee-Joon, Choi Sung-Hwan
Аннотация
OBJECTIVES: To propose and validate a 3D facial surface tracing (3D-FAST) algorithm as a robust tracing method to reduce observation errors in identifying maxillofacial curvilinear landmarks.
METHODS: The 3D-FAST algorithm comprises four sequential steps: (1) extraction of bone structures from cone-beam computed tomography (CBCT) volumes utilizing Otsu's thresholding method; (2) conversion of segmented volumetric data into a mesh representation through the marching-cubes algorithm; (3) initial landmark identification employing a data-driven approach, followed by selective manual refinement; and (4) optimization of landmark positions by identifying maximal or minimal points within defined curvilinear regions. The algorithm's efficiency was validated using 55 CBCT datasets, with particular emphasis on the Orbitale landmark. Annotation efficiency was assessed by comparing landmark deviations, dihedral angles between reference planes, and asymmetry measurements with independent manually annotated datasets.
RESULTS: Compared to orthodontically trained CBCT users, 3D-FAST demonstrated the highest consistency with the centroid derived from user annotations. In constructing horizontal and vertical reference planes and conducting asymmetry analyses, reference planes generated by 3D-FAST showed significantly improved dihedral angle variation and asymmetry than those generated by an experienced CBCT human observer, the most consistent and experienced manual user. In addition, 3D-FAST achieved improved consistency when bilateral Orbitale landmarks were used for reference plane construction, compared to when unilateral Orbitale landmarks were used.
CONCLUSIONS: 3D-FAST is a reliable method for annotating curvilinear landmarks in CBCT images.
CLINICAL SIGNIFICANCE: 3D-FAST has the potential to enhance clinical workflows and outcomes by reducing the learning curve, minimizing subjectivity, and saving time.
METHODS: The 3D-FAST algorithm comprises four sequential steps: (1) extraction of bone structures from cone-beam computed tomography (CBCT) volumes utilizing Otsu's thresholding method; (2) conversion of segmented volumetric data into a mesh representation through the marching-cubes algorithm; (3) initial landmark identification employing a data-driven approach, followed by selective manual refinement; and (4) optimization of landmark positions by identifying maximal or minimal points within defined curvilinear regions. The algorithm's efficiency was validated using 55 CBCT datasets, with particular emphasis on the Orbitale landmark. Annotation efficiency was assessed by comparing landmark deviations, dihedral angles between reference planes, and asymmetry measurements with independent manually annotated datasets.
RESULTS: Compared to orthodontically trained CBCT users, 3D-FAST demonstrated the highest consistency with the centroid derived from user annotations. In constructing horizontal and vertical reference planes and conducting asymmetry analyses, reference planes generated by 3D-FAST showed significantly improved dihedral angle variation and asymmetry than those generated by an experienced CBCT human observer, the most consistent and experienced manual user. In addition, 3D-FAST achieved improved consistency when bilateral Orbitale landmarks were used for reference plane construction, compared to when unilateral Orbitale landmarks were used.
CONCLUSIONS: 3D-FAST is a reliable method for annotating curvilinear landmarks in CBCT images.
CLINICAL SIGNIFICANCE: 3D-FAST has the potential to enhance clinical workflows and outcomes by reducing the learning curve, minimizing subjectivity, and saving time.