Automatic Estimation Of Skeletal Muscle Thickness Based On Ultrasound

Ultrasound imaging is a popular technique for diagnosis of multiple diseases such as abdominal,obstetrical and cordis.It can provide detailed information for imaging organs and soft tissue structures in the human body,and it is an efficient tool for studying the morphological features in skeletal muscle.Ultrasound imaging has been introduced to qualify the muscle changes during contractions,such as fascicle length,pennation angle,muscle thickness and other properties.As these architectural parameters are directly related to the mechanical properties of the muscle.They are able to provide a valid method for revealing the intrinsic muscle characteristic.Muscle thickness is defined as the distance between the contours of superficial and deep aponeuroses.It is probably the most direct determinant to quantify muscle activity among these morphological parameters.However,muscle thickness was conventionally detected manually in ultrasound images,it is time-consuming with the increasing amount of ultrasound images.Subsequently this greatly affects the wider applications of these parameters.To improve this,in this paper we analyzed the structure of the gastrocnemius muscle and quadriceps femoris muscle,put forward two kinds of effective method to achieve the quantitative and continuous measurement of muscle thickness in ultrasound images.The follows are the main contents in this paper :1.Analyzed the deficiency of the Hough transform(HT)method,we proposed a novel method based on modified Hough transform(HT)to achieve the quantitative measurement of muscle thickness in ultrasound images.In the part of image pre-processing,we use the Multiscale Vessel Enhancement Filtering(MVEF)method,which is based on the Hessian matrix,to enhance the aponeuroses in original images,this will facilitate the boundary extraction in the next step;then,the Soble operator was proceeded to replace the traditional Canny operator on edge detection,increased the edge gradient on the classical Hough transform to reduce the computational complexity.At last,the muscle thickness was achieved by calculating the distance between the contours of superficial and deep aponeuroses.2.Analyzed several common methods of target tracking,as compressed sensing technology is the frontier in image restoration,object detection and other fields.According to the special structure of the quadriceps,we take the lead in using compressed sensing theory to measure the muscle thickness.We applied an automatic coarse-to-fine tracking algorithm to achieve the continuous measurement.The method includes two steps: first we extract coarse location using the compressive sensing algorithm,and then the exact distance will be calculated by two different ways: Center-based distance and Edge-based distance.And muscle thickness can be obtained.In the end,the measurement results from the two proposed muscle thickness measurement methods were compared with those obtained from manual method.The comparing results demonstrate that the two algorithms are effective and can adapt to the real-time detection of muscle thickness,and it will have some practical significance and application prospects.