MRI-Based Artificial Intelligence Prediction of Fluoroscopic C-Arm Angles for Lumbar Transforaminal Epidural Steroid Injections

Lumbar transforaminal epidural steroid injection (TFESI) is an interventional pain treatment commonly used for lumbar radicular pain. It requires appropriate fluoroscopic visualization of the targeted nerve root and neural foramen to safely advance the needle. During the procedure, the fluoroscopic C-arm must be positioned at appropriate oblique and craniocaudal angles to optimally visualize the target anatomical structures. The required C-arm angles vary among patients due to individual anatomical features such as vertebra and pedicle morphology, facet joint structure, vertebral rotation, and lumbar lordosis. Currently, these angles are determined based on operator experience and sequential fluoroscopic images, often requiring multiple repositioning attempts and additional radiation exposure. Since many patients undergo lumbar MR imaging prior to TFESI, utilizing anatomical information from these existing images could help predict patient-specific initial C-arm angles. This study will use an AI-based system (ChatGPT) to evaluate standard T2-weighted lumbar MR images and predict the optimal oblique and craniocaudal C-arm angles. The operating physician will be blinded to the AI predictions and will determine the C-arm positioning based solely on their clinical evaluation. The study aims to evaluate whether AI can accurately predict these angles and to determine if prediction accuracy varies based on the target lumbar level and patient characteristics.