RESUMEN
Much clinical research on osteoporosis is aimed at documenting a reduction in vertebral fracture rates, but there is considerable disagreement about defining normality. Most methods for measuring vertebral body dimensions use lateral radiographs. In the present paper, we investigate the reliability of magnetic resonance imaging (MRI) for normal reference determination. A validation study was performed on a cadaver by comparing vertebral body volume measured both with MRI (sagittal acquisition in T1 weighted sequence) and with immersion. MRI was then performed with the same protocol from T4 to L5 in a standard population of 80 women with no history of vertebral fractures. Then all vertebral measurements were standardized relative to each other, and means and standard deviations were calculated using a statistical fitting procedure derived from volume and medial area. The validation study confirmed the reproducibility and accuracy of MRI (intraclass correlation coefficient 0.95). There was a strong correlation between volume and medial area of vertebral bodies (Pearson correlation coefficient 0.95) and a constant relationship between the medial area of vertebral bodies for each subject (coefficient of variation 5.6%). The variations in vertebral body dimensions will allow comparison with pathologic vertebral fractures in further studies. This could be useful for monitoring osteoporosis treatments.