基于光流的超声图像肌肉厚度自动测量方法  

Automatic Detection of Muscle Thickness from Ultrasound Images: A Novel Approach Based on Optical Flow

作　　者：李乔亮[1] 任盼盼[1] 张会生[1] 齐素文[1] 饶捷[1] 张小飞[1] 陈昕[1] 陈思平[1] 汪天富[1] 

机构地区：[1]医学超声关键技术国家地方联合工程实验室、广东省生物医学信息检测与超声成像重点实验室、深圳大学医学院生物医学工程系,深圳518060

出　　处：《中国生物医学工程学报》2013年第2期149-153,共5页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金项目(81000637,61031003)

摘　　要：超声成像具有无辐射、实时且成本低廉等优点,被广泛应用于人体肌肉特性的研究。传统方法利用人工从超声图像中提取肌肉厚度,主观性强,测量时间长且容易出错。本研究提出一种基于光流的肌肉厚度自动测量新方法,采用仿射光流来跟踪表层与深层的肌筋膜,基于肌筋膜与测量位置的几何关系,自动计算出肌肉的厚度。在连续运动条件下,对10位健康志愿者的大腿动态超声图像进行跟踪测量。结果表明,该法与人工测量结果的平均绝对误差为(0.24±0.22)mm,相关系数达0.93,在Matlab环境下测量速度达10帧/s,证明该方法在自动测量肌肉厚度上的准确性与高效性,在生物力学﹑肌肉骨骼建模等方面具有广阔的应用前景。Ultrasound images can be used to study properties of human muscles non-invasively, with the advantages of real time and low cost. Manual measurement of muscle thickness from ultrasound images is subjective and time-consuming. In this paper, a novel approach is proposed to detect the muscle thickness automatically. Optical flow was adopted to track the superficial and deep fascias of a muscle, and the muscle thickness was geometrically obtained based on the location of the fascias. In the continuous motion conditions, dynamic ultrasound images of thighs of 10 healthy volunteers were tracking measurement. Experimental results showed that average of the absolute difference between the proposed method and manual method was （0.24 ±0. 22） mm, the correlation coefficient was up to 0.93, and the detection speed for ultrasound frames was 10 frames per second in current implementation based on Matlab. The automated method provides an accurate and efficient approach for estimating fascicle thickness during human motion.

关 键 词：肌肉厚度 超声图像 光流 图像处理 跟踪 

分 类 号：R318.04[医药卫生—生物医学工程]