Comparison of ultrasound speed-of-sound of the lower extremity and lumbar muscle assessed with computed tomography for muscle loss assessment

Abstract

To compare the speed of propagation of ultrasound (US) waves (SoS) of the lower leg with the clinical reference standard computed tomography (CT) at the level of lumbar vertebra 3 (L3) for muscle loss assessment. Both calf muscles of 50 patients scheduled for an abdominal CT were prospectively examined with ultrasound. A plexiglas-reflector located on the opposite side of the probe with the calf in between was used as a timing reference for SoS (m/s). CT measurements were performed at the level of L3 and included area (cm2) and attenuation (HU) of the psoas muscle, abdominal muscles, subcutaneous fat, visceral fat and abdominal area. Correlations between SoS, body mass index (BMI) and CT were determined using Pearson's correlation coefficient. Based on reported CT sarcopenia threshold values, receiver operating characteristic (ROC) analysis was performed for SoS. Inter-examiner agreement was assessed with the median difference, inter-quartile range (IQR) and intraclass correlation coefficients. SoS of the calf correlated moderately with abdominal muscle attenuation (r = 0.48; P < .001), psoas muscle attenuation (r = 0.40; P < .01), abdominal area (r = -0.44; P < .01) and weakly with subcutaneous fat area (r = -0.37; P < .01). BMI correlated weakly with psoas attenuation (r = -0.28; P < .05) and non-significantly with abdominal muscle attenuation. Normalization with abdominal area resulted in moderate correlations with abdominal muscle area for SoS (r = 0.43; P < .01) and BMI (r = -0.46; P < .001). Based on sarcopenia threshold values for skeletal muscle attenuation (SMRA), area under curve (AUC) for SoS was 0.724. Median difference between both examiners was -3.4 m/s with IQR = 15.1 m/s and intraclass correlation coefficient = 0.794. SoS measurements of the calf are moderately accurate based on CT sarcopenia threshold values, thus showing potential for muscle loss quantification.

Figure 1

L3 slice segmentation: The L3 slice (A) was manually selected. Using this slice, fat values were binarized with an X-ray absorption range between [−190 HU, −30 HU] and muscle values were binarized between [−29, 150 HU]. Out of the muscle map, viscerae, bone and skin were manually excluded and the psoas muscle (red) and further abdominal muscles (orange) were segmented (B). Out of the fat map, subcutaneous fat (yellow), visceral fat (green) and intramuscular fat (blue) were segmented (C).

Figure 2

Speed of Sound (SoS) principle and measurement setup: A) Principle: The speed of sound (m/s) is calculated by assessing the sound wave propagation time (s) and the distance d (m) between the probe and reflector. An automatic algorithm (red line) recognizes the backwall echo line and calculates the mean SoS. B) The reflector was located medial and the ultrasound probe lateral to the posterior calf. Both components are connected through a frame, from which the distance between the probe and the reflector can be read manually.

Figure 3

Exemplary Speed of Sound (SoS) images of a patient with high (mean = 1572 m/s) (A) and low (mean = 1503 m/s) (B) Speed of Sound (SoS) values are shown.

Figure 4

Exemplary Computed Tomography (CT) images of a patient with a large (24.26 cm²) (A) and a small psoas area (15.43 cm²) (B) are shown.

Figure 5

Correlation plots Speed of Sound (SoS) – CT parameter: Correlation plots of SoS and abdominal muscle (a) and psoas muscle (b) attenuation are shown.

Figure 6

Diagnostic accuracy: Receiver operating characteristic (ROC) curve (A) and boxplots (B) based on skeletal muscle radiation attenuation (SMRA) threshold values^[7] for sarcopenia are shown.