Ultrasound diagnosis of skeletal muscle promoted by artificial intelligence: A quantitative evaluation of injured skeletal muscle by computer-aided ultrasonographic texture analysis / 第二军医大学学报

ABSTRACT

Skeletal muscle injuries are common in clinic, and the rate of disability caused by frequent natural disasters is increasing in recent years. Prompt diagnosis and accurate assessment of the damage range and severity are critical when a muscular injury occurs. According to our emergency ultrasound experiences, structural disorder and textural changes of damaged thigh muscle of the crush syndrome patients can be dynamically observed by high resolution ultrasound, which has similar texture imaging effect compared with magnetic resonance imaging and has unique imaging advantages. However, visual evaluation of ultrasonograms depends more on the experiences of the ultrasonographers, and thus is quite subjecitve. Besides, there is no unified standard for quantitative diagnosis of injuried muscle by ultrasound. In order to provide references for quantitative evaluation of injured muscle pathology, we performed a series of in vitroexperiments, animal experiments and preliminary clinical studies, and used computer-aided ultrasonsographic texture analysis to quantitatively evaluate the texture changes of skeletal muscle in different periods. We used contrast-enhanced ultrasound to observe the vascular distribution characteristics of skeletal muscles, and compared the muscle fiber structure, collagen distribution, and the degree of vascularization and vascular distribution in different periods, hoping to make quantitative muscle ultrasound as a practical and reliable tool for the muscle injury diagnosis.