Identification and boundary extraction of blood vessels viewed with MRI

Identification and boundary extraction of major blood vessels (e.g., aorta) in Magnetic Resonance Image (MRI) is difficult but important for further processing, such as classification of wall tissue types. In this paper, a progressive method is developed which involves the following major steps: [1] Locating the approximate position of the aorta; [2] Extracting the aorta; [3] Estimating the outer wall boundary of the aorta. A Laplacian of a Gaussian (LOG) operator is utilized as a spot detector to locate the approximate position of the aorta. A histogram pyramid of the subimage extracted from the input image is generated for automatically thresholding the subimage and segmenting the aorta. A shrink and expand operation is initially utilized to reduce the irregular protuberances of the aorta. The new boundary of the segmented aorta is then reparameterized and resampled in the 1-D vector form. The boundary estimation is completed by using Fourier descriptors of the 1-D curve.