Segmentation of ascending and descending aorta from magnetic resonance flow images

摘要

In this work, we propose an algorithm for segmenting the ascending and descending aorta from magnetic resonance phase contrast images, also referred to as MR flow imaging. The proposed algorithm is based on the active contour model combined with some refinements. In addition, false segmentation results due to severe image artifacts are automatically detected and corrected. The developed algorithm features three practical advantages: (1) fast; (2) requires minimal user interaction; and (3) robust to the changes in the algorithm parameters (e.g. same parameter set is used for all datasets). The algorithm is tested and validated on a number of datasets with different image qualities. The preliminary results show very satisfactory segmentation of the ascending and descending aorta even when the flow images are severely distorted.