Deep Learning for Cervical Spine Radiography: Automated Measurement of Intervertebral and Neural Foraminal Distances

Background/Objectives: The precise localization of cervical vertebrae in X-ray imaging was essential for effective diagnosis and treatment planning, particularly as the prevalence of cervical degenerative conditions increased with an aging population. Vertebrae from C2 to C7 were commonly affected by disorders such as ossification of the posterior longitudinal ligament (OPLL) and nerve compression caused by posterior osteophytes, necessitating thorough evaluation. However, manual annotation remained a major aspect of traditional clinical procedures, making it challenging to manage increasing patient volumes and large-scale medical imaging data. Methods: To address this issue, this study presented an automated approach for localizing cervical vertebrae and measuring neural foraminal distance. The proposed technique analyzed the neural foramen distance and intervertebral space using image enhancement to determine the degree of nerve compression. YOLOv8 was employed to detect and segment the cervical vertebrae. Moreover, by integrating automated cervical spine analysis with advanced imaging technologies, the system enabled rapid detection of abnormal intervertebral disc gaps, facilitating early identification of degenerative changes. Results: According to the results, the system achieved a spine localization accuracy of 99.5%, representing an 11.7% improvement over existing approaches. Notably, it outperformed previous methods by 66.67% in recognizing the C7 vertebra, achieving a perfect 100% accuracy. Conclusions: Furthermore, the system significantly streamlined the diagnostic workflow by processing each X-ray image in just 17.9 milliseconds. This approach markedly improved overall diagnostic efficiency.