From routine full-spine radiographs to decision-oriented Risser stratification: an interpretable deep-learning approach - Takeaways - MDSpire

From routine full-spine radiographs to decision-oriented Risser stratification: an interpretable deep-learning approach

  • By

  • Zexi Wang

  • Yuan Zhang

  • Yixi Wang

  • Feng Xue

  • Yi Yang

  • Wen Zhao

  • June 17, 2026

  • 0 min

Share

  • 1

    An interpretable deep-learning model was developed for automatic Risser stratification using routine full-spine radiographs in adolescent idiopathic scoliosis.

  • 2

    The model achieved an AUC of 0.938 and an accuracy of 0.875 for binary stratification of Risser stages 0–2 versus 3–5.

  • 3

    Model assistance significantly reduced reading time by 9.7–11.8 seconds per case and improved agreement with expert references among spine surgeons.

  • 4

    Grad-CAM visualizations indicated that the model focused on the iliac apophysis and adjacent ossification regions for predictions.

  • 5

    The study highlights the potential of deep learning to enhance standardized assessment of growth potential in adolescent idiopathic scoliosis management.

Original Source(s)

From routine full-spine radiographs to decision-oriented Risser stratification: an interpretable deep-learning approach

  • by Zexi Wang, Yuan Zhang, Yixi Wang, Feng Xue, Yi Yang, Wen Zhao

  • June 17, 2026

Related Content

Novel UHMWPE sleeve for stabilizing loose suture anchors in rotator cuff repair

Correction: Cable-Plate augmentation improves the therapeutic effect of intramedullary nailing for AO/OTA type A2.3 intertrochanteric fractures with large coronal fragments: a double-center retrospective study

Self-Management May Reduce Chronic Low Back Pain

In a randomized clinical trial of patients at increased risk for persistent symptoms, clinician-supported biopsychosocial self-management was associated with lower pain impact and fewer chronic pain outcomes than guideline-based medical care.