Automated quantification of coronal balance in spinal deformity: a safety-aware clinical workflow.

The assessment of coronal balance is critical for cosmetic appearance and quality of life in patients with adolescent idiopathic scoliosis (AIS), yet its manual quantification remains constrained by substantial workflow demands and inherent subjectivity. While artificial intelligence has shown promise in automated Cobb angle measurement, robust and clinically safety-aware solutions for coronal balance parameters are still lacking. In this study, we developed and validated a robust HRNet-based automated system for quantifying Shoulder Height Difference (SHD), Pelvic Obliquity (PO), and C7 Plumb Line Offset (C7 offset) using a dataset of 847 full-spine radiographs. Beyond anatomically driven topological constraints, we integrated a novel Clinical Safety Gating Mechanism that proactively suppresses unreliable outputs in cases of anatomical ambiguity, functioning as a safety gating by design. The system achieved millimeter-level precision (mean absolute error: 0.86-2.45 mm) and demonstrated strong agreement with manual reference measurements (Pearson r > 0.95), with overall performance approximating the inter-rater variability among senior spinal surgeons. Importantly, the deterministic algorithmic consistency of the system mitigates human fatigue and subjective fluctuation, offering theoretical advantages for future longitudinal assessments. Collectively, this work positions the automated coronal balance system not merely as a measurement tool, but as a safety-aware quality safety gating mechanism. These findings validate the technical feasibility and internal safety of the model, providing a robust foundation for large-scale clinical data auditing and future prospective studies on long-term patient outcomes.