An Explainable Framework for Non-Invasive Scoliosis Screening Using Geometric Body Symmetry Analysis

— Early identification of scoliosis is essential for preventing the progression of spinal deformities and enabling timely clinical intervention. However, conventional scoliosis assessment primarily relies on radiographic imaging, exposing patients to repeated ionizing radiation during routine follow-up. This paper presents an explainable framework for non-invasive scoliosis screening using conventional posterior body images. The proposed methodology integrates image preprocessing, geometric feature extraction, and the Explainable Scoliosis Risk Score (ESRS) to evaluate external body symmetry and classify subjects into three clinically meaningful categories: Scoliosis Not Present, Mild Scoliosis Risk, and Scoliosis Present. Unlike computationally intensive learning-based approaches, the proposed framework employs explainable geometric analysis to provide transparent and clinically interpretable screening outcomes without requiring deep neural network training or manually annotated datasets. Experimental evaluation on 203 posterior body images demonstrates the effectiveness of the proposed framework in distinguishing different levels of postural asymmetry while maintaining computational efficiency. The generated explainable clinical visualization further improves result interpretability by presenting the extracted anatomical features together with the corresponding screening outcome. Owing to its radiation-free, cost-effective, and explainable design, the proposed framework is well suited for preliminary scoliosis screening in school-based programs, primary healthcare centers, and other resource-constrained clinical environments.