Tissue Tracking in Transverse and Longitudinal Ultrasound Image Sequences for Hand Diseases

Abstract

In clinical diagnosis ultrasound is an important technique and has been widely used for many common hand diseases such as carpal tunnel syndrome (CTS) and trigger finger Recent studies show that the displacement and deformation of the median nerve and the tendon between healthy subjects and patients have significant difference Moreover CTS may occur with trigger finger patients more often However some problems such as speckle noise out-of-plane etc make it hard to track and measure manually in the ultrasound images This study presents two novel tracking strategies for the median nerve and the tendon in transverse and longitudinal view respectively To track the contour of the median nerve in the traverse ultrasound image sequence the proposed method adopts the machine learning method for localization; then optical flow and active contour model are used to track and refine the contour in the ultrasound image sequences To track the motion of the tendon the proposed method integrates optical flow and block matching method to calculate the optimal tendon motion between ultrasound image frames In median nerve tracking the accuracy of the proposed method is about 0 88 in average Dice similarity coefficient 4 46 pixels in average mean of absolute difference and 3 52 pixel for average center difference In tendon tracking the proposed method is validated by the phantom ultrasound sequence and compared with some classical tracking methods The experimental results reveal that the proposed method is better and more stable than the comparative methods in most cases In the future the proposed methods can further be applied in patient data to obtain clinical parameters such as the area and velocity of the tissues By comparing the parameters between patients and normal subjects the indexes use to distinguish the symptomatic and asymptomatic can then be defined

Date of Award	2017 Sept 6
Original language	English
Supervisor	Yung-Nien Sun (Supervisor)