Automated X-Ray Image Analysis for Lumbar Spondylolisthesis Detection and Severity Grading

Abstract

Spondylolisthesis, characterized by the anterior displacement of a vertebra, significantly impacts spinal health diagnosis and treatment. This study introduces a groundbreaking machine learning strategy for automated detection and grading of lumbar spondylolisthesis from X-ray images, utilizing Roboflow for data management and a customized convolutional neural network (CNN). This CNN accurately identifies lumbar vertebral segments and objectively grades vertebral slippage. The evaluations show a mean average precision (mAP) of 98.5%, with precision at 96.8% and recall at 97.2%, underscoring the model's accuracy and reliability. Additionally, we developed a user-friendly interface for healthcare professionals, enhancing the tool's clinical applicability. The method offers a significant improvement over existing diagnostic approaches, providing a reliable, efficient solution for the early detection and management of lumbar spondylolisthesis.