Magnetic resonance imaging vs. computed tomography image concordance for linear measurements and the quantification of abdominal skeletal muscle

Background

Linear measurement analysis using computed tomography (CT) scans to quantify abdominal muscle mass has been validated as a clinically practical approach for screening individuals with low muscle mass. However, there is still a need to validate such analysis using magnetic resonance imaging (MRI) imaging. The aim of this study is to assess the reproducibility and concordance of CT and MRI imaging for linear measurement analyses of skeletal muscle at mid-L3.

Methods

We retrospectively analysed 66 patients with available CT and MRI images within 30 days of one other to evaluate linear measurement CT and MRI concordance. Linear measurement analysis for abdominal/pelvic CT and MRI scans for eight patients was conducted independently three times by the same person separated by at least 1 week to assess intra-rater variability. The intra-observer variability for both CT and MRI was assessed using the intraclass correlation coefficient (ICC). The concordance and correlation of CT and MRI mid-L3 for linear measurements were assessed using Pearson correlation coefficients and Bland–Altman plots.

Results

The intra-rater reliability of linear measurements for both CT and MRI was high, as measured by the ICC (CT range: 0.788–0.992; MRI range: 0.766–0.984). CT and MRI linear measurements were found to be significantly positively correlated for all psoas (total psoas r = 0.98; P < 0.0001) and paraspinal muscle measurements (total paraspinal r = 0.99; P < 0.0001). Bland–Altman analysis revealed a mean bias of 0.83 (range: 0.03–5.56) for MRI over CT linear measurements.

Conclusions

CT and MRI images were shown to be concordant for linear measurement analysis of abdominal muscle mass. T2-weighted MRI sequences can be used interchangeably with CT in the assessment of sarcopenia using linear measurement analysis.