Three-dimensional Depth Imaging for Early Scoliosis Automated Screening

Study Design. Diagnostic study. Objective. To systematically evaluate the reliability and validity of an automated screening system based on three-dimensional (3D) depth imaging for adolescent idiopathic scoliosis (AIS). Summary of Background Data. Traditional scoliometry is time-consuming and operator-dependent, while existing non-radiographic imaging techniques often lack portability. Despite the potential of 3D depth imaging in spinal assessment, its application in the coronal plane for AIS screening remains insufficiently evaluated. Methods. A total of 238 participants (8–18 y) were enrolled, including 160 AIS patients and 78 non-scoliotic subjects. All underwent standing radiography (gold standard), 3D depth imaging, and scoliometer examination. The depth-based spinal (DS) angle was computed by the 3D system. Intra- and inter-observer reliability were assessed via intraclass correlation coefficient (ICC). Correlation between DS angle and Cobb angle was analyzed, with diagnostic performance evaluated using receiver operating characteristic (ROC) curves. Operational time and diagnostic metrics were compared between the 3D system and scoliometer using paired statistical tests. Results. The DS angle demonstrated excellent reliability (intra- and inter-observer ICC >0.99) and moderate-to-high correlation with Cobb angle (r=0.70–0.77, P <0.001). For AIS diagnosis (Cobb >10°) and bracing indication (Cobb >20°), the AUCs were 0.91 and 0.94, with sensitivities of 0.92/0.93 and specificities of 0.85/0.89, respectively. Single curves showed superior performance to double curves. The 3D system had shorter operational time (35 s vs. 150 s, P <0.001) and better diagnostic metrics (AUC: 0.91 vs. 0.70; sensitivity: 0.92 vs. 0.60; specificity: 0.85 vs. 0.72, all P <0.001) than the scoliometer. Conclusion. The 3D depth imaging system is a radiation-free AIS screening tool with excellent reliability, high diagnostic value, and superior operational efficiency compared to traditional scoliometry, supporting its clinical translation and widespread application.