Assessment of leg muscle morphology by contrast-enhanced magnetic resonance imaging (CE-MRI) in patients with peripheral arterial disease

Abstract

Background Peripheral arterial disease (PAD) is associated with functional impairment in the lower extremities (LE). PAD related skeletal muscle loss is a complex and poorly understood process which has been associated with denervation and a decrease in type II muscle fiber mass. Previous studies suggest a potential link between change in muscle morphology and PAD. We hypothesize that imaging can be used to quantify differences in skeletal muscle morphology between PAD patients and controls. We utilized CE-MRI and machine learning using texture features to quantify muscle morphology in PAD patients and healthy controls. Methods Thirty-four individuals (20 PAD patients [median age=67 years, median ankle brachial index=0.62], 14 healthy controls [median age=35 years]) underwent CE-MRI of the LE using a 36-element bilateral LE coil with a 3.0T MRI system. Imaging was performed post reactive hyperemia induced with a bilateral blood-pressure cuff positioned above the knee. Cross-sectional images were acquired at the mid-calf level using a high resolution saturation recovery gradient echo pulse sequence (temporal resolution=409.57ms; slice thickness=10mm; echo time=1.23ms; in plane resolution=1.3x1.3mm). The anterior-, lateral-, and deep posterior muscle compartments and the soleus-, and gastrocnemius muscles were semiautomatically segmented with in-house developed software. The segmented muscle groups were utilized to automatically extract texture features that characterize muscle morphology in the CE-MRI data: 1) texture maps, 2) multi-directional Haralick gray level co-occurrence matrix features; 3) Gabor wavelet features; 4) scale-invariant feature transform features; and 5) intensity histogram feature. Subsequently, a support vector machine (SVM) based framework was developed to classify skeletal muscle tissue as ‘normal’ or ‘PAD’ by using the extracted features. Results