The aim of this systematic review was to assess the accuracy and reliability of automatic landmarking for cephalometric analysis of 3D craniofacial images. We searched for studies that reported results of automatic landmarking and/or measurements of human head CT or CBCT scans in MEDLINE, EMBASE and Web of Science until march 2019.Two authors independently screened articles for eligibility. Risk of bias and applicability concerns for each included study were assessed using the QUADAS-2 tool. Eleven studies with test dataset sample sizes ranging from 18 to 77 images were included. They used knowledge-, atlas-or learning-based algorithms to landmark 2 to 33 points of cephalometric interest. Ten studies measured mean localization errors between manually-and automatically-detected landmarks. Depending on the studies and the landmarks, mean errors ranged from <0.50 mm to >5 mm. The two best-performing algorithms used a deep learning method and reported mean errors <2 mm for every landmark, approximating results of operator variability in manual landmarking. Risk of bias regarding patient selection and implementation of the reference standard were found, therefore the studies might have yielded overoptimistic results. The robustness of these algorithms needs to be more thoroughly tested in challenging clinical settings. PROSPERO registration number: CRD42019119637.

The increasing use of 3-dimensional (3D) imaging by orthodontists and maxillofacial surgeons to assess complex dentofacial deformities and plan orthognathic surgeries implies a critical need for 3D cephalometric analysis. Although promising methods were suggested to localize 3D landmarks automatically, concerns about robustness and generalizability restrain their clinical use. Consequently, highly trained operators remain needed to perform manual landmarking. In this retrospective diagnostic study, we aimed to train and evaluate a deep learning (DL) pipeline based on SpatialConfiguration-Net for automatic localization of 3D cephalometric landmarks on computed tomography (CT) scans. A retrospective sample of consecutive presurgical CT scans was randomly distributed between a training/validation set ( n = 160) and a test set ( n = 38). The reference data consisted of 33 landmarks, manually localized once by 1 operator( n = 178) or twice by 3 operators ( n = 20, test set only). After inference on the test set, 1 CT scan showed “very low” confidence level predictions; we excluded it from the overall analysis but still assessed and discussed the corresponding results. The model performance was evaluated by comparing the predictions with the reference data; the outcome set included localization accuracy, cephalometric measurements, and comparison to manual landmarking reproducibility. On the hold-out test set, the mean localization error was 1.0 ± 1.3 mm, while success detection rates for 2.0, 2.5, and 3.0 mm were 90.4%, 93.6%, and 95.4%, respectively. Mean errors were −0.3 ± 1.3° and −0.1 ± 0.7 mm for angular and linear measurements, respectively. When compared to manual reproducibility, the measurements were within the Bland–Altman 95% limits of agreement for 91.9% and 71.8% of skeletal and dentoalveolar variables, respectively. To conclude, while our DL method still requires improvement, it provided highly accurate 3D landmark localization on a challenging test set, with a reliability for skeletal evaluation on par with what clinicians obtain.

To design a quasi-automated three-dimensional reconstruction method of the spine from biplanar X-rays as the daily used method in clinical routine is based on manual adjustments of a trained operator and the reconstruction time is more than 10 minutes per patient. Methods: The proposed method of 3D reconstruction of the spine (C3-L5) relies first on a new manual input strategy designed to fit clinicians' skills. Then, a parametric model of the spine is computed using statistical inferences, image analysis techniques and fast manual rigid registration. Results: An agreement study with the clinically used method on a cohort of 57 adolescent scoliotic subjects has shown that both methods have similar performance on vertebral body position and axial rotation (null bias in both cases and standard deviation of signed differences of 1mm and 3.5° around respectively). In average, the solution could be computed in less than 5 minutes of operator time, even for severe scoliosis. Conclusions: The proposed method allows fast and accurate 3D reconstruction of the spine for wide clinical applications and represents a significant step toward full automatization of 3D reconstruction of the spine. Moreover, it is to the best of our knowledge the first method including also the cervical spine.