Quantifying morphological changes in middle trapezius with ultrasound scanning and a novel histogram matching algorithm

Abstract

Facioscapulohumeral dystrophy (FSHD) can affect upper-limb function through muscle degeneration, fatty infiltration and oedema. Muscle echogenicity, measured using ultrasound, could be used as a biomarker for muscular changes and disease progression [1–3]. Histogram-matching may be instrumental in overcoming existing shortcomings that prevent us from using such images to extract clinically useful information [4,5]. Can histogram-matching of muscle ultrasound images be used to extract clinically relevant measures to quantify the muscle morphological changes in people with FSHD (pwFSHD)? Participants attended a single motion analysis session for upper-limb 2D-ultrasound imaging and 3D-movement analysis. Stratified sampling by arm function was used for pwFSHD. Controls were age and sex matched. Middle trapezius measurement was taken at the midpoint of a line between C7 and ACJ. Six total measurements were taken (3-longitudinal and 3-transverse views) using an Esoate MyLab-Gamma device and linear probe (3-13 MHz). Muscle thickness measurements were carried using ImageJ 1.53t. Histogram-matching was carried out as described by Bottenus et al. [4]. All images were matched to a single reference image from a control group participant. Manual segmentation of the subcutaneous fat layer was carried out and used as the region-of-interest for histogram-matching across all images. Using full histogram-matching, the monotonic transformation was applied across the entire image. The trapezius muscle was segmented to determine mean grayscale values (echogenicity). The student t-test was used for evaluating between group differences and the relationship between echogenicity values and muscle thickness was investigated. Data was collected for 14 participants (7 pwFSHD (2 F:5 M) and 7 sex- and age-matched controls (2 F:5 M). PwFSHD had mean (SD) age, height and weight values of 41.9-years (17.1), 176 cm (8.8) and 90.6 kg (24.8) respectively. The control group had age, height and weight values of 41.4-years (15.5), 176.4 cm (5.7) and 77.1 kg (11.2) respectively. Group echogenicity values are presented in Fig. 1. Download : Download high-res image (106KB)Download : Download full-size image Mean (SD) echogenicity values for pwFSHD were higher than the control group (96.5 (30.3) vs 32.2 (11.2) respectively) with statistically significant differences (p<0.001). Mean (SD) trapezius muscle thickness was higher in the control group 1.48 cm (0.27) vs 0.74 cm (0.45) respectively. Mean echogenicity scores accounted for 82% of the variance in mean muscle thickness values (R2=0.824). PwFSHD demonstrated higher echogenicity values and smaller muscle thicknesses indicative of degenerative muscle structure changes associated with the disease. Preliminary results suggest that post capture processing of ultrasound images using histogram matching can provide quantifiable differences in people with and without FSHD. This could facilitate clinically feasible bedside methods for assessing and monitoring disease progression in pwFSHD. Further work is needed to recruit a larger sample of pwFSHD and varying levels of arm function, carry out longitudinal measurements and evaluate the sensitivity of these measures on the basis of variable reference images.