NIH ImageJ and Slice-O-Matic computed tomography imaging software to quantify soft tissue.

Abstract

To compare reliability and limits of agreement of soft tissue cross-sectional areas obtained using Slice-O-Matic and NIH ImageJ medical imaging software packages. Abdominal and midthigh images were obtained using single-slice computed tomography. Two trained investigators analyzed each computed tomography image in duplicate. Adipose tissue and skeletal muscle cross-sectional areas (centimeters squared) were calculated using standard Hounsfield unit ranges (adipose tissue: -190 to -30 and skeletal muscle: -29 to 150). Regions of interest included abdominal total area, total fat area, subcutaneous fat area, visceral fat area (AVF), and right and left thigh total area, fat area, and skeletal muscle area. For all images, intra-investigator coefficients of variation ranged from 0.2% to 3.4% and from 0.4% to 5.6% and inter-investigator coefficients of variation ranged from 0.9% to 4.8% and 0.2% to 2.6% for Slice-O-Matic and NIH ImageJ, respectively, with intra- and inter-investigator coefficients of reliability of R(2) = 0.99. Mean AVF values for investigators A and B ranged from 168 to 170 cm(2) using Slice-O-Matic and NIH ImageJ. Bland-Altman analyses revealed that Slice-O-Matic and NIH ImageJ results were comparable. The mean differences (95% confidence intervals) between the AVF cross-sectional areas obtained using the Slice-O-Matic and NIH ImageJ medical imaging software were +2.5 cm(2) (-5.7, +10.8 cm(2)) or +1.4% (-3.4%, +6.4%). These findings show that both the Slice-O-Matic and NIH ImageJ medical imaging software systems provide reliable measurements of adipose tissue and skeletal muscle cross-sectional areas.