A Study Of The Level Set Method Combined With Empirical Mode Decomposition In Magnetic Resonance Image Segmentation

To provide a reliable basis for auxiliary diagnosis to physicians,the significance of medical image segmentation is segment human tissue structure in the image.In a variety of medical imaging technology,magnetic resonance imaging are widely used in clinical medicine for its high spatial resolution and soft tissue resolution,what is more it has no radiation,can do multi-dimension,multiple parameter,multiple sequences imaging at the same time,contains more information than other imaging techniques,etc..However,every MRI will contain large amounts of data and intensity inhomogeneity,weak borders,noise in image,are difficulties in clinical medicine MR image processing.In recent years,MR image segmentation and recognition has gradually become the research focus of the industry,the level set method is be concerned once again for it is suitable for image segmentation.Level set method converts curve evolution to partial differential equations solving,has many advantages such as avoid tracking curve evolution,computation stability,suitable for any dimension etc..However,in the present various traditional level set method and its improved method is still difficult to meet the needs of clinical diagnosis,So a bi-dimensional ensemble empirical mode decomposition(BEEMD)method is introduced to improve accuracy of MR image segmentation by distance regularized level set(DRLSE)method.The main research works of this paper are as follows:(1)The main steps of the improved DRLSE method are:Firstly,the MR image is decomposed into numbers of two dimensional intrinsic mode functions(BIMF)by BEEMD method;different weighting coefficients are endued to BIMF for image reconstruction to enhance the segmentation target.Secondly,parts of BIMF components are added into edge indicator function of DRLSE to recover the blurring boundary caused by Gauss smooth operation.Then DRLSE is used to segment the reconstructed MR image.(2)Considering the slow decomposition in BEEMD method.in this paper,According to noise assistant complex empirical mode decomposition,which not only can remove mode mixing problem,but also can improve the decomposition speed.We proposed complex white noise assistant bi-dimensional empirical mode decomposition,by adding white noise as imaginary part in the two-dimensional complex matrix.Doing image rotation base on Euler equations,then decomposing image by bi-dimensional empirical mode decomposition,we can get a set of complex BIMFs in different directions form the two-dimensional complex matrix,thus,the average value of every real parts are the original image BIMFs.