Cobb Angle Measurement using U-Net for Scoliosis Treatment Plans

Scoliosis, a condition characterized by abnormal spinal curvature, remains a critical challenge in paediatric healthcare, especially in the accurate measurement of the Cobb angle, which is essential for assessing its severity. This study presents ATT UNet 3+, a deep learning model designed for automatic Cobb angle measurement, improving upon existing models with its hybrid attention mechanism. By training and validating on 155 X-ray images, ATT UNet 3+ demonstrated superior performance in segmentation accuracy compared to traditional methods. With an error margin of just 1.6°, it offers a reliable tool for efficient scoliosis diagnosis and treatment planning, reducing dependence on manual evaluation. This research highlights the potential of deep learning in enhancing medical imaging techniques, facilitating faster and more accurate assessments in clinical environments.