Semi-automatic segmentation of vessels by mathematical morphology: application in MRI

We propose a semi-automated method to isolate vessels or other structures from magnetic resonance imaging acquisitions. This method is divided into two parts. The first part is a nonlinear filter, using morphological operators, based on the invariance properties of the image. The second consists of finding the extrema contours. The contour which is proposed here is called the virtual contour, because it is located between the extrema contours. We have tested our method on 20 patients. The results showed that the method is robust and efficient when it comes to segmenting vessels from MRI.