Scoliosis Reporter AI
Bao Do MD
Scoliosis affects 3.1% of children globally, impacting females at 4.06% versus 2.58% in males. Most cases involve mild curvatures of 10 to 19 degrees, with severe curves over 40 degrees in ~0.05% of the population. ~3.8% of diagnosed patients required surgery, with 97% utilizing fusion [1], [2].
Automatic quantitative and qualitative analysis of scoliosis radiographs using deep learning is feasible and holds potential in improving radiologist workflow. This app is an update of the original end-to-end system and publication to analyze scoliosis radiographs and generate report measurements. Post-processing now uses confidence-aware anatomical sequence reconstruction, scale-normalized duplicate and outlier rejection, explicit inference of likely missed vertebral centers, robust local centerline fitting, joint apex and Cobb endpoint selection, and perturbation-based stability checks. Inferred centers are marked with an asterisk. Cobb lines remain interactive and all image decoding and analysis occurs locally.
Use DICOM, JPG or PNG images. Images are not uploaded, remaining strictly on your computer to ensure complete data privacy.
Try sample images from the internet 1, 2.
Move the AI Cobb lines to update the report. Drag and drop next images without reloading.
Education or research, not for clinical use.
Loading
Model loading...