Cardio Minute Segmentation Paper

Cardio vascular magnetic resonance imaging (CMRI) is one of the most important imaging modalities for the detection of heart diseases. Heart diseases are the result of plaque build up in the coronary arteries which lead to blockages. This block may totally block the blood supply to the heart muscles leading to heart attack. Accurate segmentation of coronary artery is crucial for detection of blocks. Manual segmentation requires a skilled operator to trace the region boundaries in 2D slices. Several automatic segmentation methods have been proposed. One such method of segmentation is based on intensity or gradient information in which a boundary model is constructed from a group of training data and adapts the boundary model to extract boundaries from the …show more content…
An iterative method is used to extract the centreline and vessel boundaries [8-9]. In this the image derivatives were often used to detect the boundaries or centerlines of tubular structures. The vessel boundaries are extracted based on local intensity gradients which lead to inaccurate segmentation. Several methods are proposed for automatic centreline extraction of coronary vessel based on medialness filter [10], fuzzy corrected function [11] and model fitting [12]. The disadvantages of these methods are the difficulty in extracting the distal parts of the coronary arteries, motion artifacts and calcifications. Another method of automatic coronary vessel extraction is based on morphological operators[13]. This method resulted in premature termination of the algorithm due to intensity distribution and difference in morphology from the normal vessel. To avoid isotropic diffusion in Gaussian scale space Bauer et al. Proposed a method that combined Frangi vesselness function [14] and Gradient vector flow[15]. This method extracted coronary artery and calcifications with high gray