ARTIFICIAL INTELLIGENCE AVAILABLE TO THE DEVELOPMENT OF A VIRTUAL REALITY SOFTWARE FOR AN AUTOMATED CEPHALOMETRIC ANALYSIS OF ULTRA-REDUCED CBCT FOV

This thesis investigates the integration of artificial intelligence (AI), virtual reality (VR), and ultra-reduced cone-beam computed tomography (CBCT) fields of view (FOV) to develop a novel approach for automated cephalometric analysis in orthodontics. The study addresses key challenges such as enhancing diagnostic precision, ensuring radioprotection, and optimizing clinical workflows through innovative technological solutions. Cephalometric analysis, a cornerstone of orthodontic diagnosis and treatment planning, has traditionally relied on manual or semi-automated methods using 2D radiographs. However, these methods face limitations, including superimposition of structures and the inability to assess depth. While 3D CBCT imaging offers significant advantages, its associated higher radiation exposure raises safety concerns, particularly for pediatric patients and cases requiring multiple imaging sessions. This research responds to the need for safer, more efficient imaging solutions by focusing on AI-driven automation and ultra-reduced FOVs in CBCT imaging. The project developed an AI system based on the V-Net deep learning architecture to automate the detection of anatomical landmarks critical for cephalometric analysis. Using a robust training and validation process with expert-annotated CBCT datasets, the system achieved high accuracy, with a mean Euclidean distance error of less than 2 mm. Preprocessing techniques, such as intensity thresholding and data augmentation, enhanced the model’s reliability and robustness. The introduction of ultra-reduced FOV CBCT scans further minimized radiation exposure by up to 70% compared to conventional protocols, without compromising diagnostic quality. The integration of intraoral scans with CBCT data provided a comprehensive and accurate anatomical representation, enhancing the precision of automated analyses. The use of VR technology added a transformative dimension to the system, offering an immersive platform for clinicians to interact with 3D cephalometric data. Through dynamic manipulation and visualization of craniofacial structures, the VR interface enhanced diagnostic confidence and facilitated a more intuitive understanding of patient anatomy. This approach streamlined the diagnostic workflow, reducing the time required for cephalometric measurements by over 50%, allowing clinicians to focus more on treatment planning and patient care. Validation studies demonstrated the system's superior efficiency and reproducibility compared to traditional methods. Clinicians highlighted the VR platform’s user-friendly interface and its potential to revolutionize diagnostic workflows. By combining AI automation, VR interactivity, and radioprotection-focused imaging, the research provides a comprehensive diagnostic tool tailored to modern orthodontic needs. This thesis concludes that the integration of AI, VR, and ultra-reduced FOV CBCT imaging represents a significant advancement in orthodontic diagnostics. The proposed system offers a safer, faster, and more precise alternative to conventional methods, aligning with the principles of patient-centered care and radioprotection. Future research should aim to expand the dataset to ensure broader applicability and explore predictive modeling for personalized treatment strategies. The findings underscore the transformative potential of combining cutting-edge technologies to address the evolving challenges of orthodontics, setting a new standard for diagnostic excellence and patient safety.