Opportunistic bone mineral density assessment in pediatric scoliosis: validating Hounsfield unit measurement from lumbar CT images.
BACKGROUND Hounsfield Units (HU) derived from CT scans are a promising bone mineral density surrogate in pediatrics, yet their reliability in patients with scoliosis and significant lumbar tilt remains unknown. We investigated the reliability of HU measurements in adolescent idiopathic scoliosis (AIS), particularly addressing the impact of spinal tilt. METHODS We retrospectively reviewed lumbar CT images of 11- to 19-year-old AIS patients who underwent posterior spinal fusion (July 2007-March 2024). HU values and lumbar tilt were measured at L1, L3, and L5. Reliability was assessed using intra-rater (N = 36), inter-rater (N = 30), and equivalence datasets (N = 62) comparing standard pelvis-parallel images versus images reconstructed parallel to the vertebral body. RESULTS Both intra-rater (ICC(1,1) > 0.95) and inter-rater (ICC(2,1) > 0.80) reliability were excellent. Measurements from standard pelvis-parallel images and reconstructed images showed stable overall equivalence (ICC(3,1) = 0.96). However, the percentage of cases with an HU value difference of 20 or more was 11.3% when the lumbar tilt angle was 10 degrees or greater, compared to 5.6% when the tilt angle was below 9 degrees. CONCLUSION Our analyses for AIS patients showed that vertebral HU measurements are reliable when using vertebral-parallel reconstructed images but standard pelvis-parallel images lose reliability with significant tilt. While our findings do not promote routine CT usage, our validated methodology provides a baseline technique that could potentially be applied to complex, high-risk cases where osteoporosis is prevalent and critical for surgical planning.