Characterization of Myocardial Diseases Using Image Processing Technique

Recently, the nuclear cardiology imaging has been broadly used as a key method to assist in determination of the different cardiac pathologies and monitoring the management prognosis. Myocardium perfusion is considered the most common assessment tool in the nuclear cardiology that used to assess the myocardium of the heart by using both SPECT and PET. Those modalities have ability to recognize the myocardial area precisely. Full and half automated computed-based images registration and segmentation of the myocardium pathologies was performed to increase the visibility of the image and increase its diagnostic role. This study was performed in nuclear medicine department of Rabit Nation University in period of January 2016 to December 2020. In this thesis, the novel medical image analysis methods were presented to characterize myocardial diseases using SPECT to differentiate between myocardial infarction and myocardial ischemia as well as the quantification of the normal heart tissues. There are many problems due to absence of SPECT imaging protocol which is very essential in diagnosis of cardiovascular disease and can be used to compare its effectiveness of other diagnostic modalities. The automatic delimitation technique was proposed of the SPECT cardiac images. The automated image enhancement, noise reduction and segmentation algorithms were used in this study to accurately delineate the myocardium structures compared with the other similar approaches. Automated methods of image processing, mainly of image registration, are integrated in a computational solution to automatically compute a set of features from myocardial perfusion SPECT images and use them to statistical analysis and classification of patient exams as from a healthy patient or with an associated disease. The image registration algorithms used, including the watershed-based segmentation, similarity measure, optimization, and interpolation algorithms, will be described and discussed, as well as the computational processing, analysis and classification techniques employed. The segmented images were projected onto the original images to demonstrate those projections and correlate them with the manual delineations. The experimental results of this study showed that the automated methods and visualization with high corresponding and matching rates. Indicate that the mean values of 0.82, 0.92 and 0.90 are achieved for comparison when the K-means clustering methodology is employed in three clusters and watershed segmentation, when utilized as gold standard and clinical professional.