Research On Estimation Methods For Brain Cerebral Blood Volume Based On MR Perfusion Imaging

Brain magnetic resonance perfusion imaging is one of the imaging methods of brainfunction, which can reflect the distribution of microvascular perfusion within theorganization. It is a new technology which can evaluate tissue in vivo and provide relevanthemodynamic parameters. Usually we scan levels selected continuously during intravenousbolus injection of contrast agent, measure tracer signal after vascular changes, and obtaindifferent hemodynamic indexes. Through the analysis of perfusion parameters, theperfusion states within tissue and organ can be evaluated, and the microcirculationsituations of brain tissue, such as the vascular lesion changes, are acquired.In this thesis, a detailed overview of the basic principle and experiment method ofmagnetic resonance perfusion imaging technique is made firstly. Secondly, the Fouriertransform based minimum mean square error (FT-based MMSE) method is used to simulatecerebral blood volume (CBV). By using Wiener filter, the estimated values of the CBV arevery close to the ideal values. Experiments show that the signal to noise ratio (SNR) andtracer delay have little effects on the estimations of CBV, but metric echo time has obviousinfluence on experimental values. Therefore we amended echo time in different regions.Then, the singular value decomposition (SVD) method is adopted to estimate CBV. Thesimulation results demonstrate that echo time has little effect on estimated values, but theSNR and tracer delay are relative sensitive. So we improve the method by settingappropriate thresholds in different regions to well evaluate CBV values. Finally, thecolor-coded cerebral blood volume maps are achieved from the real brain perfusion images.The color-coded images can reflect the brain structure clearly, and estimate the distributionof cerebral blood volume. It is helpful for doctors to determine the lesion type and location.