Cardiovascular diseases are the major cause of death in the developed world. About half of these are due to ischemia heart diseases.
The high death rate caused by coronary artery diseases increases the need for preliminary detection. Perfusion magnetic resonance imaging has turned out to be very promising for this purpose. A contrast agent is injected intravenously to visualize the perfusion. Due to the extremely time-consuming manual analysis of these relatively large datasets, efforts for automatic approaches have been introduced. Most of these proposed methods focus on parts of the analysis process. The present thesis identifies four steps for an automatic analysis approach:
localization of the heart, suppression of motion artifacts, segmentation of the myocardium, and perfusion analysis. This thesis presents a method covering all these subtasks in an automatic manner with no need for any user interaction. First the acquired MR images are analyzed to roughly detect the heart. A registration step compensates motion artifacts based on the breathing of the patient. The segmentation step provides the contour of the myocardium at every time step. Based on these segmentations the perfusion is quantified.
The algorithm was tested on 11 datasets. Inspection of the results indicates that this method is very promising for an efficient perfusion analysis.