IMPROVEMENT OF SEGMENTATION OF UPPER AND LOWER JAW COMPUTED TOMOGRAPHY IN PATIENTS WITH MAXILLOFACIAL INJURIES AT THE ORTHOPEDIC STAGE OF REHABILITATION

The relevance of this work was due to the increasing number of patients with maxillofacial injuries requiring orthopedic rehabilitation. At the same time, specialists are increasingly relying on artificial intelligence (AI) technology. One of its most important advantages is its ability to quickly and accurately analyze huge amounts of data, providing specialists with valuable information to improve decision-making processes for planning orthopedic rehabilitation for patients with maxillofacial injuries. The synergy between AI workflows and computed tomography (CT) segmentation has the potential to improve the accuracy and efficiency of further treatment planning and patient management. Objective. The aim of the study is to evaluate an improved method of CT image segmentation for patients with maxillofacial injuries, combining an automatic algorithm and manual post-processing, in order to improve segmentation accuracy and reduce processing time compared to traditional methods. Materials and Methods. The study was conducted in 30 patients with maxillofacial injuries at the orthopedic stage of rehabilitation. In the course of the study, we compared the methods of CT segmentation of the upper and lower jaws: a step-by-step method (reference), automatic segmentation with AI, and an innovative method (own development). This method of CT image segmentation for patients with maxillofacial injuries combines an automatic AI algorithm and manual post-processing. It is this combination that helps to improve segmentation accuracy, which has been proven by the results of the IoU and Dice metrics. Results. The improved method demonstrated higher localization accuracy and was much faster than Stepwise segmentation. The innovative segmentation method has proven to be a new solution for improving CT segmentation diagnostics, namely for localizing images with different resolutions and reducing processing time compared to conventional methods. Conclusion. Our study proved the effectiveness of the improved method for patients with maxillofacial injuries and substantiated the practical application of this improved method of automatic segmentation with manual post-processing in clinical practice. Thus, an improved method for segmenting CT images for patients with maxillofacial injuries, combining an automatic algorithm and manual post-processing, improves segmentation accuracy and reduces processing time compared to traditional methods.