Comprehensive Summary
The goal of this study was to diagnose temporomandibular joint osteoarthritis using a neural network AI model trained to assess Cone Beam Computed Tomography Scans (CBCT). The "You Only Look Once" (YOLO) AI model was used, which features a human-based architecture framework with a spine, neck, and head, utilizing a CSPDarknet53 backbone. An oral and maxillofacial surgeon and an oral radiologist both assessed images to evaluate the AI model. The evaluated diagnostic metrics were Sensitivity, Specificity, Positive Predictive Values (PPV), Negative Predictive Values (NPV), and Accuracy. The overall accuracy of the model ranged from 92.7% to 98.74%, demonstrating statistically significant agreement with the expert diagnoses. Despite the study’s implications for clinical practice, the study specifically highlights the necessity of clinician input for further development of the model. There are also limitations within the parameters of the images used for testing the AI model, since uniform parameters cannot be expected within clinical settings.
Outcomes and Implications
This research is highly valuable in the field of diagnostic medicine. Traditionally, diagnoses using imaging have high interobserver variability and take more time for patients and clinicians. The AI model developed in this study has the potential to be developed into a clinical diagnostic tool that provides faster and more accurate interpretation of medical imaging. This could mean an impact on treatment times, financial investment in diagnoses, and more time for clinicians to focus on complex cases.