Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/50888
Title: Automated detection and segmentation of vertebral bodies in volumetric abdominal magnetic resonance images
Authors: Lin, Dehui.
Keywords: DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
Issue Date: 2012
Abstract: Detection and segmentation of the column of vertebral bodies are intermediate steps required to identify the bone marrow which is considered as the wrongly labeled visceral adipose tissue in the assessment of abdominal obesity. Motivated by the necessity, a fully automated algorithm is designed to detect and segment the column of vertebral bodies in the volumetric abdominal magnetic resonance images. In addition, the feasibility of the use of frequency domain to detect the vertebral bodies is also evaluated through the performance of the algorithm. In the development of the algorithm, common image processing techniques such as discrete Fourier transform and thresholding are used to detect and segment the vertebral bodies. In total, 21 data sets of abdominal magnetic resonance images are used to test the performance of the algorithm. Accuracy rates of 98.6% in detection and 93.7% in segmentation are achieved. In spite of different resolutions, equally good performance of the algorithm is observed. The efficient and effective automated algorithm proves the usefulness of frequency domain in detecting the column of vertebral bodies and the accuracy of finding the volumes of vertebral bodies.
URI: http://hdl.handle.net/10356/50888
Rights: Nanyang Technological University
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:SCSE Student Reports (FYP/IA/PA/PI)

Files in This Item:
File Description SizeFormat 
SCE11_0497.pdf
  Restricted Access
Main article1.27 MBAdobe PDFView/Open

Page view(s) 50

187
checked on Oct 28, 2020

Download(s) 50

7
checked on Oct 28, 2020

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.