Currently, CT muscle density is measured in Hounsfield units and is not converted to g/cm 3 . Muscle mass and fat fraction (FF) cannot be measured from CT images. We propose a phantomless calibration method to calculate muscle tissue and muscle density in g/cm 3 , as well as muscle FF in percent, based on the measurement of water-offset corrected CT values of muscle and subcutaneous adipose tissue (SAT). Calibration requires the mass density and FF of standard muscle tissue (SMT), as well as the FF of SAT. In this study, the International Commission on Radiation Units (ICRU) definition of muscle tissue with a mass density of 1.05 g/cm 3 was used as standard muscle tissue (SMT). The CT values of SMT were measured using a phantom scanned with seven different CT scanners employing tube voltages ranging from 80 to 140 kV. Existing CT scans of 42 men aged 72 years or over were used to estimate the accuracy errors of the calibrated parameters of the paraspinal muscles. MR Dixon data from 31 subjects were available for comparing CT and MRI FF. The mean SMT CT value of the phantom measurements was 47.7 ± 1.9 HU. In line with previous publications, the SMT fat fraction was set to 3%, and the subcutaneous adipose tissue (SAT) FF to 85%. Using an SMT CT value of 48 HU, the mean muscle tissue density, muscle density, and muscle fat fraction (FF) of the 41 subjects were 0.94 ± 0.08 g/cm 3 , 1.04 ± 0.01 g/cm 3 , and 11.7 ± 6.4%, respectively. Simulated variations in SMT density of ±0.02 g/cm 3 , SMT CT value of ±2 HU, SMT FF of 5%, and SAT FF of ±3 HU resulted in relative changes in muscle and muscle tissue densities and in absolute changes in muscle FF of below 5%. Simulated accuracy errors were comparable to those caused by an error of ±5 HU in the water offset corrections of the measured CT values of muscle and SAT. The mean MRI fat fraction values were 4.1% higher than the mean CT FF values. The two measurements exhibited a high degree of linear correlation (r 2 = 0.88, p < 0.001). The proposed methodology for muscle calibration in single-energy CT images showed a high degree of agreement with MR Dixon FF measurements. The simulated accuracy errors were comparable to those caused by missing water offset corrections of the measured CT values. This is a proof-of-concept study, further validation in subjects with higher muscle FF and in other muscle groups is required. Quantitative assessments of muscle properties such as density and fat infiltration are important biomarkers for myopathies, sarcopenia, obesity and potentially for osteoporosis. While MRI techniques are state-of-the-art, the opportunistic use of existing CT scans can support screening strategies and may help to identify more subjects at early risk for muscle and perhaps even bone loss. New CT technology, such as photon counting CT, which reduces radiation exposure by around 50% compared to standard CT, may also make CT attractive for dedicated muscle imaging.
Bammessel et al. (Fri,) studied this question.