Validating the Relationship Between Higher-Order Echotexture Features, Skeletal Muscle Fat Infiltration, and Glucose Metabolic Health

Abstract

Ultrasound is commonly used in muscle physiology research to assess tissue quantity, and its utility may extend to the evaluation of muscle composition (i.e. the amount of contractile versus non-contractile tissue within the muscle). However, more work is needed to refine this innovative application. Previous research has sought to determine whether echointensity (average pixel brightness) reflects muscle fat infiltration, but this simple first-order echotexture feature is highly influenced by external factors. Second- and higher-order echotexture features also consider the spatial arrangement of pixels (in addition to brightness) and may better characterize fat infiltration within muscle, but there is a paucity of data investigating this idea. Therefore, this thesis sought to determine: 1) whether higher-order ultrasound echotexture is predictive of muscle fat infiltration assessed using gold-standard magnetic resonance imaging (MRI), and 2) whether these ultrasound echotexture features also associate with glucose metabolic health. Ninety relatively healthy adults (50% female, age range: 18 – 82 years ) participated in this cross-sectional validation study. Participants underwent ultrasound, a thigh MRI, as well as a fasting blood sample to assess circulating features of metabolic health. We used stepwise multiple regression analysis to identify potential ultrasound and metabolic predictors of muscle fat infiltration, assessed using MRI. We found that higher-order (but not first- or second-order) ultrasound echotexture features and glycated hemoglobin (HbA1c) emerged as key potential predictors of muscle fat infiltration. Additionally, we found that HbA1c correlated significantly with certain first- (i.e. echointensity, histogram skew, kurtosis) and higher-order (i.e. blob size, number and local binary pattern energy) ultrasound echotexture features. These results suggest that ultrasound can successfully characterize muscle composition in a reference cohort of adults. Overall, this study has begun to fill in a critical research gap surrounding more accessible alternatives for evaluating skeletal muscle, which could lead to enhanced monitoring of disease progression and metabolic health outcomes.