An Automatically Partial Volume Correction-Based Method for the Measurement of Adipose Tissue

Abstract

In the research of obesity the volume of adipose tissue surrounding the aorta plays an essential role in evaluation of atherosclerotic cardiovascular disease (CVD) In many clinical studies the volume measurement of adipose tissue often uses 3D medical images Additionally image processing techniques such as segmentation are used to retrieve the aorta and adipose tissue; the region of interest (ROI) is then acquired from the segmented results However the results calculated using conventional binary segmentation and voxel-count method usually differ from the actual volume of the object because of the medical instrument scanning process In order to solve this problem we provide an automatic algorithm for measuring the volume of adipose tissue in ROI The algorithm resolves the problem of the partial volume effect (PVE) which emerges in the region around the boundary of the segmented results thus improving accuracy The proposed method contains active contour models (ACMs) genetic algorithm (GA) and fuzzy C-means (FCM) clustering algorithm for segmentation and utilizes partial volume correction-based method for quantification of volume from the previously segmented results We provide computed tomography (CT) images from 30 patients and the ground truth of adipose tissue which were checked by a radiologist to evaluate the measurement of adipose tissue volume by the proposed and other methods The experimental results show this method improved accuracy significantly and has high reproducibility

Date of Award	2015 Aug 4
Original language	English
Supervisor	Shu-Mei Guo (Supervisor)