Detection of cervical vertebrae maturation with artificial intelligence algorithms improved by deep learning method on lateral cephalometric images obtained from cone-beam computed tomography images

Aim: The aim of this study is to increase the success of the physician in the evaluation of cervical vertebra maturation by using a special artificial intelligence algorithm on lateral cephalometric images obtained from cone beam computed tomography images. Material and Method: The data set of our study consists of drawing and classifying vertebrae with polygonal drawing method using a special artificial intelligence algorithm on lateral cephalometric radiographs obtained from cone beam computed tomography images taken before orthodontic treatment of 1077 patients. Labeling on images was done using CranioCatch labeling software (CranioCatch, Eskişehir, Turkey). The training of the artificial intelligence model was carried out using 400 Epochs with the CNN-based deep learning method applied to PyTorch. The learning rate of the model was determined as 0.01. Results: In the artificial intelligence model trained in the study, the highest F1 score of 0.9984 was found in the detection of cervical vertebrae. Considering the data imbalance within the classes, the fact that the vertebral maturation and the growth development classification F1 score is 0.6946, the ROC curve and the AUC value close to 1 are promising for the growth-development detection of the individual with artificial intelligence algorithms. The lowest F1 value was observed in the maturation classifications of the vertebrae themselves. Conclusion: Our study is very important for the development of deep learning- based CBCT reporting systems to be made in the future. It is thought that these systems will play a role as a decision support mechanism by saving time for physicians in their clinical routine. At the same time, it is estimated that it will be very useful in minimizing the differences between observers in the evaluation of growth-development and the inconsistencies that may occur in the evaluations made by the observers at different times.