Automated segmentation of the aortic artery: evaluation of images obtained by triple rule-out protocol

Purpose: This study served as the initial step for developing a computer assisted diagnosis (CAD) system that detects images with signs of aortic diseases from patients´ Triple Rule-Out (TRO) protocol data. The process of the CAD system for seeking and segmenting the pathway of an aortic artery was tested. Methods: Three cardiac scans of patients who underwent the TRO protocol were selected for this test: one set with a normal condition, one set with an aortic dissection (AD) in the ascending aorta and one set with an AD in the descending aorta. Results: The segmentation process managed to segment all slices for the ascending and descending aorta and to create reconstructed images around the aortic arch. In patient A´s scan there were 32 of 51 slices in the ascending aorta that contained signs of AD. The process segmented all 32 slices. In patient B´s scan, there were 145 slices in the descending aorta that contained signs of AD and 6 slices in the aortic arch that contained signs of AD. The process segmented all 145 slices in the descending aorta and all 6 slices in the aortic arch. Conclusion: This process managed to create segmental aorta data including the slices with aortic dissection with no errors. The process also successfully reconstructed the cross-section images of the aortic arch which matched the original data in regarding AD. The segmentation process can be used for the next step in building a CAD system for detecting signs of aortic diseases.