Objective: To investigate the intra- and inter-observer agreement and correlation between CT and MR measurements of skeletal muscle area (SMA) in the abdomen.
Methods: CT and MR images from twelve patients were analyzed by two blinded observers using segmentation software (MITK-3M3, Mint Medical and Slice-O-Matic, Tomovision) to quantify SMA. MR images included T1w "in-phase", T1w "out-of-phase", and T2w sequences. Inter- and intra-observer agreement was assessed using the intraclass correlation coefficient (ICC). Pearson correlation coefficient (r) was used to correlate measurements obtained on MR with CT. CT and MR measurements were compared with Bland-Altman plots.
Results: Intra- and inter-observer agreement for SMA was high for CT and MR. For MR, the measurements on T2w images showed the highest inter-observer agreement (ICC=0.96). CT SMA correlated closely with MR, with T2w images showing the highest correlation (r=0.98; P<0.01). Bland-Altman plots showed a 1.7%-3.9% bias between CT and MR measurements, lowest for T2w images.
Conclusions: MR SMA measurements are reproducible and correlate closely with CT. The T2w sequence is recommended to quantify SMA on MR images.
Keywords: Image analysis; MR; Sarcopenia; Segmentation; Skeletal muscle area.
Copyright © 2016 Elsevier Ireland Ltd. All rights reserved.