Quantitative measures of modic changes in lumbar spine magnetic resonance imaging: intra- and inter-rater reliability.

摘要

STUDY DESIGN: A measurement reliability study. OBJECTIVE: To develop quantitative measures for Modic changes (MCs) on magnetic resonance (MR) images and evaluate measurement reliability. SUMMARY OF BACKGROUND DATA: MCs have been studied for more than 20 years but the clinical significance remains controversial. Little effort has been made to improve the measurement of MCs. METHODS: The study was approved by the responsible institutional review board. On the basis of Modic classification, a series of quantitative dimension and signal intensity measures were developed for assessing MCs. Midsagittal T1- and T2-weighted MR images from 83 lumbar spines were then qualitatively and quantitatively assessed by two observers independently. Kappa and intraclass correlation coefficient (ICC) were used to examine intra- and inter-rater reliability. Pearson correlation coefficient was used to assess the relationships between the quantitative measurements of MCs. Mean absolute deviation (MAD) and Bland-Altman plots also were used to evaluate measurement errors and limits of agreement for selected measures. RESULTS: For Modic classification, intrarater agreement was excellent (kappa = 0.88) and inter-rater agreement was substantial (kappa = 0.79). Intrarater agreement also was excellent when obtaining dimension measurements (ICC = 0.82-0.96) from T1- or T2-weighted images and inter-rater agreement was slightly greater using T1-weighted images (ICC = 0.73-0.88) than T2-weighted images (ICC = 0.66-0.82). Signal intensity measurements on T2-weighted images were found to have almost perfect intra- and inter-rater reliability (ICC = 0.92-0.99). The correlation analysis demonstrated that the quantitative measures represent different constructs. The MAD and Bland-Altman Plots further confirmed the high reliability of the area ratio, MCs mean signal intensity and MCs total signal intensity measurements. CONCLUSION: Three quantitative measures are suggested to assess the severity of MCs, which provide reliable, precise measurements for research on the etiology, pathogenesis, and clinical relevance of MCs.