Musculoskeletal (MSK)PediatricAI / InformaticsResearchTrainee

Deep Learning Framework Automates Lenke Classification on Multi-View Spine Radiographs

Global spine journalyesterday

On 76 test cases, the AI framework achieved Cobb angle MAE 2.4°, Lenke curve type accuracy 0.895, and reduced review time from 8.2 to 1.5 min/case.

  • Fully automated pipeline used vertebral detection, landmark estimation, parameter computation, and rule-based classification on 4-view full-spine radiographs.
  • Complete Lenke label accuracy was 0.816, with high surgeon agreement (κ=0.873 for curve type).
  • AI-assisted review reduced mean interpretation time from 8.2 to 1.5 minutes per case, with processing time of 12.3 seconds.
Read the source

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.