AI and the World's Hard Problems · Medicine & drug discovery · Nature (computational biology) · 2026-10-23
A deep learning framework for automated cervical spinal cord vertebral labeling and morphometric extraction from brain MRI with large-scale multicenter validation
The study aimed to build and validate a fully automated pipeline that identifies the C1–C3 vertebral levels of the upper cervical spinal cord in T1-weighted brain MRI and includes quality control. Deep learning models were trained on data from 998 controls and 879 patients with hereditary ataxias across 32 sites, with nnU-Net performing best (Dice of 0.85 for C1 and 0.89 for C2 and C3). External validation on 2,737 MRIs showed labeling accuracy above 92% in four of five datasets, and the method also detected longitudinal spinal cord atrophy in patients with multiple sclerosis.