Lateral cephalometry has been widely used for skeletal classification in orthodontic diagnosis and treatment planning. However, this conventional system, requiring manual tracing of individual landmarks, contains possible errors of inter- and intravariability and is highly time-consuming. This study aims to provide an accurate and robust skeletal diagnostic system by incorporating a convolutional neural network (CNN) into a 1-step, end-to-end diagnostic system with lateral cephalograms. A multimodal CNN model was constructed on the basis of 5,890 lateral cephalograms and demographic data as an input. The model was optimized with transfer learning and data augmentation techniques. Diagnostic performance was evaluated with statistical analysis. The proposed system exhibited >90% sensitivity, specificity, and accuracy for vertical and sagittal skeletal diagnosis. Clinical performance of the vertical classification showed the highest accuracy at 96.40 (95% CI, 93.06 to 98.39; model III). The receiver operating characteristic curve and the area under the curve both demonstrated the excellent performance of the system, with a mean area under the curve >95%. The heat maps of cephalograms were also provided for deeper understanding of the quality of the learned model by visually representing the region of the cephalogram that is most informative in distinguishing skeletal classes. In addition, we present broad applicability of this system through subtasks. The proposed CNN-incorporated system showed potential for skeletal orthodontic diagnosis without the need for intermediary steps requiring complicated diagnostic procedures.

Background Despite the integral role of cephalometric analysis in orthodontics, there have been limitations regarding the reliability, accuracy, etc. of cephalometric landmarks tracing. Attempts on developing automatic plotting systems have continuously been made but they are insufficient for clinical applications due to low reliability of specific landmarks. In this study, we aimed to develop a novel framework for locating cephalometric landmarks with confidence regions using Bayesian Convolutional Neural Networks (BCNN). Methods We have trained our model with the dataset from the ISBI 2015 grand challenge in dental X-ray image analysis. The overall algorithm consisted of a region of interest (ROI) extraction of landmarks and landmarks estimation considering uncertainty. Prediction data produced from the Bayesian model has been dealt with post-processing methods with respect to pixel probabilities and uncertainties. Results Our framework showed a mean landmark error (LE) of 1.53 ± 1.74 mm and achieved a successful detection rate (SDR) of 82.11, 92.28 and 95.95%, respectively, in the 2, 3, and 4 mm range. Especially, the most erroneous point in preceding studies, Gonion, reduced nearly halves of its error compared to the others. Additionally, our results demonstrated significantly higher performance in identifying anatomical abnormalities. By providing confidence regions (95%) that consider uncertainty, our framework can provide clinical convenience and contribute to making better decisions. Conclusion Our framework provides cephalometric landmarks and their confidence regions, which could be used as a computer-aided diagnosis tool and education.