Study Of Segmentation Technics For MRA Cerebrovascular

Angiography is an important means for the diagnosis of vascular disease; magneticresonance angiography has been widely used because it is a non-invasive method. TheMRI collects a set of cross-sectional images of the body, which are stacked together toconstitute a group of volume data. This article aims to segment vascular data from thethree-dimensional volume data acquired with magnetic resonance angiography.An automatic shimming algorithm is designed to adjust the MRI for imageacquisition, paving the way for the follow-up work. Then vessel segmentation using thefinite mixture model and normal distribution case EM algorithm iterative form areintroduced. MRA segmentation is performed using the finite mixture normal distributionmode with a MRA dataset as instance. The strategy of initialization and termination ofEM algorithm is studied through experiment to optimize the performance of thesegmentation method. We use connectivity filter to get rid of the remaining backgroundsignal.With multiple sets of data to the effectiveness of the algorithm designed in thispaper is verified.