Research On Mipalgorithm Of Brain Vascular Reconstruction

Blood vessels are the channels where the blood flow through.In the human body,blood vessels everywhere and they play a huge role.As a branch of the arteries of the brain,they are responsible for the transport of substances and oxygen with other vessels.Three dimensional magnetic resonance angiography datas can provide high resolution and rich vascular information,which are necessary for the diagnosis of cerebrovascular diseases.In order to display the vascular morphology,doctors always use maximum intensity projection(MIP)technology to project the MRA datas in each orientation.The technology is of great significance in the diagnosis of cerebral aneurysms,cerebral arteriosclerosis,cerebral infarction and vascular malformation.The paper regards blood vessels in the brain as the research object and it uses the MIP as the reconstruction method.When introducing MIP,we talk about the advantages and disadvantages of MIP.In order to solve ruptures of small vessels and reduce the time for reconstruction,corresponding improvement measures put forward.Specific research contents are divided into the following four parts:1)The paper introduces the classic MIP reconstruction algorithm in details and analyzes the shortcomings and advantages of MIP reconstruction algorithm.Firstly,the paper introduces two pretreatment methods,they are image denoising and target enhancement.On this basis,the paper introduces in details the whole process of MIP reconstruction based on the classical ray casting algorithm.Finally,through the experimental test,it is proved that MIP algorithm is superior to shaded surface display(SSD)and volume rendering(VR)algorithm in brain vessel's reconstruction.At the same time,the paper points out two problems when reconstruct the vessels: rupture of small vessels and slow reconstruction.2)The paper resolves ruptures of small vessels in the reconstruction image.Aiming at small vessels' ruptures,a new method based on local maximum mean(LMM)is designed.Firstly,we use LMM to deal with the three-dimensional MRA datas for the enhancement of voxels in MRA volume datas.Then we use MIP reconstruction to get the LMM-MIP image.Finally,the small vessels' ruptures resolved.At the same time,in order to solve the problem of large vessels thickening which lead into by LMM-MIP,a weight function is adopted to combine the LMM-MIP image with the MIP image toensure the repair of the damaged blood vessels and restore the large vessels.3)The paper resolves slow MIP reconstruction.Aiming at the problem of slow MIP reconstruction,we use shear warp algorithm to replace ray casting algorithm.Then we improve shear warp algorithm and combine with the characteristics of CUDA's interpolation texture memory to reduce the reconstruction time.Finally,we test the MIP and LMM-MIP which achieves by ray casting,shear warp algorithm and improved shear warp algorithm combined with CUDA respectively.4)The paper uses the brain magnetic resonance angiography datas provided by manufactures to complete the experiments.Then,the performance of the algorithm is analyzed by vessel voxel projection probablity and the CNR of reconstruction image.Finally,the result proved that LMM algorithm is fitting for small vessels' ruptures,and the modified shear-warp algorithm which works on CUDA also improves the MIP reconstruction speed.