INTRODUCTION: Assessment of spine alignment is crucial in the management of scoliosis, but current auto-analysis of spine alignment suffers from low accuracy. We aim to develop and validate a hybrid model named SpineHRNet+, which integrates AI and rule-based methods to improve auto-alignment reliability and interpretability.

METHODS: 1,542 consecutive patients attending two scoliosis clinics were recruited with written informed consent. Their radiographs were recaptured using smartphones or screenshots, with deidentified images securely stored. Manually labelled landmarks and alignment parameters by a spine surgeon were considered as ground truth (GT). The data were split 8:2 to train and internally test SpineHRNet+ respectively. This was followed by a prospective validation on another 337 patients. Quantitative analyses of landmark predictions were conducted, and reliabilities of auto-alignment were assessed using linear regression and Bland-Altman plots. Deformity severity and sagittal abnormality classifications were evaluated by confusion matrices.

RESULTS: SpineHRNet+ achieved accurate landmark detection with mean Euclidean distance errors of 2.78 and 5.52 pixels on posteroanterior and lateral radiographs, respectively. The mean angle errors between predictions and GT were 3.18° and 6.32° coronally and sagittally. All predicted alignments were strongly correlated with GT (p<0.001, R²>0.97), with minimal overall difference visualised via Bland-Altman plots. For curve detections, 95.7% sensitivity and 88.1% specificity was achieved, and for severity classification, 88.6-90.8% sensitivity was obtained. For sagittal abnormalities, greater than 85.2-88.9% specificity and sensitivity were achieved.

DISCUSSION: The auto-analysis provided by SpineHRNet+ was reliable, fast, and continuous. It offers the potential to assist clinical work and facilitate large-scale clinical studies (https://www.aimed.hku.hk/alignprocare; password: alignprocare; also available on App Store and Google Play as AlignProCARE).