Muscle State Analysis Based On Ultrasound Images

Skeletal muscle is the most abundant tissue in human,whose main function is to produce strength and maintain posture.Muscle loss will occur under various physiological and pathological conditions,severely depleting the body’s exercise capacity and health.Therefore,the dynamic detection of muscle state is of great significance for maintaining human athletic ability and health.Ultrasound imaging can provide clear images of human internal soft tissue and organ structure for researchers to analyze skeletal muscle structure.Muscle ultrasound monitoring has the advantages of noninvasive,harmless,convenient,and low cost,which has broad application prospects in muscle disease diagnosis,muscle loss detection and real-time monitoring of muscle state.The changes of myofascial characteristics reflect the health status of neuromuscle,so monitoring the muscle state parameters related to myofascial characteristics is of great significance for the early diagnosis of neuromuscular diseases and the evaluation of muscle health status.Muscle pennation angle which is a commonly parameters played an important role in quantitative evaluation of muscle state.Living myofascia presents a typical high echo band structure in ultrasound images.Early state measurement methods are inefficient and subjective by using image correlation software to measure muscle state manually in ultrasound images.Therefore,how to automatically measure muscle state parameters in ultrasound images based on image processing technology is one of the key problems to be solved in quantitative evaluation of muscle state.The existing methods of muscle ultrasonic testing are all focused on the detection of muscle state parameters,but the research of muscle structure and composition change detection based on muscle ultrasound image is still a challenging problem in the field of image processing and analysis.The main contents of this paper are as follows:（1）This dissertation analyzes the principles of several existing automatic measurement schemes of pennation angle,and finds that the existing automatic detection and measurement methods of pennation angle are based on line detection technology,which is not completely consistent with the real state of muscle,which affects their detection and measurement accuracy and reduces their robustness to noise interference.In this paper,we propose a new discrete transform,namely the BLS（Belt Linear Summation）discrete transform.Based on BLS transform,this dissertation design an automatic detection method of pennation angle,which is significantly different from the existing detection methods.Firstly,from the detection principle,the existing detection methods are based on the line detection in the ultrasonic image,and this dissertation proposed method is based on the band detection in the ultrasonic image,which is closer to the real situation of the muscle fiber bundle in the ultrasonic image;secondly,BLS transform is directly defined in discrete domain,and there is no integral approximation error and discrete sampling error in radon transform and Hough transform.（2）This dissertation compare the BLS transform based method with the existing automatic detection methods of pennation angle in artificial image and clinical image data respectively.Through the field Ⅱ simulation program to obtain the simulation muscle ultrasound image,the advantage of using artificial simulation image is that there is no subjective measurement deviation of pennation angle,which can verify the effectiveness of the algorithm.Clinical data is obtained by testing the subjects with laboratory ultrasound equipment,which can verify the feasibility of the algorithm in clinical application.The final experimental results show that this method has high detection accuracy,low calculation cost,no prior knowledge,high adaptability,and automation level.It also can be applied to real-time monitoring and dynamic measurement of muscle state.（3）Our research found that the local line structure features of muscle fiber pixel neighborhood in muscle ultrasound images are closely related to the changes of muscle structure and composition.Therefore,a statistical monitoring method of muscle structure and composition changes based on multi-directional,multi-scale analysis and local line feature descriptors is proposed.The results show that this method can monitor the changes of the ratio of muscle mass to tendon membrane in weightless muscle atrophy of mice caused by tail suspension,and can also monitor the changes of muscle atrophy in human bed rest experiment.