DocumentCode :
241039
Title :
Automatic segmentation of left ventricle in cardiac MRI using Maximally Stable Extremal Regions
Author :
Abdelfadeel, Mohammed A. ; ElShehaby, Saleh ; Abougabal, Mohammed S.
Author_Institution :
Biomed. Eng. Dept., Alexandria Univ., Alexandria, Egypt
fYear :
2014
fDate :
11-13 Dec. 2014
Firstpage :
145
Lastpage :
148
Abstract :
Short-axis Cine Cardiac Magnetic Resonance Images (MRI) provide a good contrast between left ventricle (LV) blood pool and myocardium. Thus, they are suitable for segmenting left ventricle cavity and computing left ventricle areas, volumes and some important clinical parameters like ejection fraction and myocardial mass. Many recent endocardium segmentation methods utilize optimal thresholding as simple and fast technique to segment left ventricle cavity. Unfortunately, intensity inhomogeneity decreases the efficiency of optimal thresholding. In this paper, we propose an automatic technique to overcome the aforementioned problems. The proposed technique is based mainly on Maximally Stable Extremal Regions (MSER) detector. The proposed technique consists of two main steps: First, generating of MSERs and Second, these regions were analyzed to select the region representing the LV cavity. The proposed technique was validated using 15 test sets from MICCAI 2009 LV segmentation challenge. Results were compared with thresholding-based method applied to the same dataset. Results show better performance in terms of segmentation accuracy and computation time.
Keywords :
biomedical MRI; blood vessels; image segmentation; medical image processing; LV cavity; MICCAI 2009 LV segmentation challenge; MSER; Maximally Stable Extremal Region detector; Short-axis Cine Cardiac Magnetic Resonance Images; automatic segmentation; automatic technique; cardiac MRI; clinical parameters; computation time; contrast; dataset; ejection fraction; endocardium segmentation method; intensity inhomogeneity; left ventricle areas; left ventricle blood pool; left ventricle cavity segmentation; left ventricle volumes; maximally stable extremal region; myocardial mass; myocardium; optimal thresholding; segmentation accuracy; thresholding-based method; Biomedical imaging; Blood; Image segmentation; MATLAB; Myocardium; Sensitivity; MSER; cardiac MRI; image segmentation;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Biomedical Engineering Conference (CIBEC), 2014 Cairo International
Conference_Location :
Giza
ISSN :
2156-6097
Print_ISBN :
978-1-4799-4413-2
Type :
conf
DOI :
10.1109/CIBEC.2014.7020940
Filename :
7020940
Link To Document :
بازگشت