Automatic Detection and Pennation Angle Measurement of Muscle Fascicles in Ultrasound Images Using Belt Linear Summation Transform

Abstract

Ultrasound images of muscle fascicles have been widely used to investigate muscle properties for diagnosis and rehabilitation assessment.The existing automatic fascicle detection and measure methods are based on line detection techniques which do not exactly coincide with the true state of the muscle images. This affects their detection and measure accuracy and depresses their robustness to background interference. In this work, a novel discrete transform namely Belt Linear Summation (BLS) transform is proposed. Unlike the line transform techniques which calculate the sum of pixels on a straight line in images, the BLS transform intends to determine the weighted summation of a belt of pixel values. Based on BLS transform, an automatic fascicle detection and measure method is designed. The performance of the proposed method is compared to the recent automatic fascicle detection and measure methods using both simulated images and clinical images. Experimental results show that the proposed method is robust to background and noise interference, accurate in terms of muscle pennation angle measurement, and feasible for analyzing clinical data.