Levi M Teigen1, Adam J Kuchnia1, Emily Nagel1, Christopher Deuth1, David M Vock2, Urvashi Mulasi3, Carrie P Earthman1. 1. Department of Food Science and Nutrition, University of Minnesota-Twin Cities, Minneapolis, Minnesota, USA. 2. Division of Biostatistics, University of Minnesota-Twin Cities, Minneapolis, Minnesota, USA. 3. Department of Family and Consumer Sciences (Nutrition and Food/Dietetics), California State University, Sacramento, California, USA.
Abstract
BACKGROUND: There is growing interest in computed tomography (CT) measures of skeletal muscle cross-sectional area (CSA) for nutrition assessment. Multiple software programs are available, but little work has been done comparing programs. We aimed to determine if CT-derived measures of skeletal muscle CSA at the level of the L3 are influenced by the software program used. We also demonstrate the importance of the ImageJ corrigendum published in this journal. METHODS: Two software programs, National Institutes of Health ImageJ and Tomovision sliceOmatic, were compared. ImageJ measures were obtained using both the original tutorial and corrigendum instructions. Skeletal muscle CSA at the level of the L3 was measured in advanced heart failure and head and neck cancer populations by 3 different investigators. Intraclass correlation coefficients were used to calculate intrarater and interrater reliability. Bland-Altman analysis was used to assess agreement. RESULTS: Both software programs yielded excellent intrarater and interrater reliability scores (intraclass correlation coefficients, 0.985-1.000). The overall mean difference (ImageJ tutorial with corrigendum - sliceOmatic) for the entire sample (N = 51) was found to be 1.53 cm2 (95% CI, 0.59-2.47 cm2 ). The overall mean difference (ImageJ corrected - original) for the entire sample (N = 51) was found to be -11.35 cm2 (95% CI, -12.75 to -9.95 cm2 ). CONCLUSION: Measures of skeletal muscle CSA at the L3 were found to be ∼1.53 cm2 higher with ImageJ than sliceOmatic. This difference was not found to affect interpretation against a published cut point. The importance of accounting for the ImageJ tutorial corrigendum was shown to be clinically significant when applied to published cut points.
BACKGROUND: There is growing interest in computed tomography (CT) measures of skeletal muscle cross-sectional area (CSA) for nutrition assessment. Multiple software programs are available, but little work has been done comparing programs. We aimed to determine if CT-derived measures of skeletal muscle CSA at the level of the L3 are influenced by the software program used. We also demonstrate the importance of the ImageJ corrigendum published in this journal. METHODS: Two software programs, National Institutes of Health ImageJ and Tomovision sliceOmatic, were compared. ImageJ measures were obtained using both the original tutorial and corrigendum instructions. Skeletal muscle CSA at the level of the L3 was measured in advanced heart failure and head and neck cancer populations by 3 different investigators. Intraclass correlation coefficients were used to calculate intrarater and interrater reliability. Bland-Altman analysis was used to assess agreement. RESULTS: Both software programs yielded excellent intrarater and interrater reliability scores (intraclass correlation coefficients, 0.985-1.000). The overall mean difference (ImageJ tutorial with corrigendum - sliceOmatic) for the entire sample (N = 51) was found to be 1.53 cm2 (95% CI, 0.59-2.47 cm2 ). The overall mean difference (ImageJ corrected - original) for the entire sample (N = 51) was found to be -11.35 cm2 (95% CI, -12.75 to -9.95 cm2 ). CONCLUSION: Measures of skeletal muscle CSA at the L3 were found to be ∼1.53 cm2 higher with ImageJ than sliceOmatic. This difference was not found to affect interpretation against a published cut point. The importance of accounting for the ImageJ tutorial corrigendum was shown to be clinically significant when applied to published cut points.
Authors: Alexandra B Schroeder; Ellen T A Dobson; Curtis T Rueden; Pavel Tomancak; Florian Jug; Kevin W Eliceiri Journal: Protein Sci Date: 2020-11-20 Impact factor: 6.725