Multiphase Image Segmentation Based On Improved LBF Model

Brain disease has become one of the diseases with high incidence in recent years. In order to better analyze brain structure, brain magnetic resonance imaging(MRI) technology is widely used in medical field. Because the brain disease will lead to the changes of the structure of the tissue in a certain extent, it’s important for medical diagnosis to segment MR brain imaging accurately. In practice, brain imaging will be affected by the device itself and the magnetic field. And these factors lead to the inhomogeneity of the gray level of the tissue in MR brain map. In addition, the brain map is usually made of dark gray, white, light gray and black. So it can’t be realized by using the previous fitting of the global information of the segmentation model.So far, the segmentation method is numerous in the image processing field, and the PC multi-phase model is mainly aimed at the multi-phase images of intensity inhomogeneity. And then the PS multi-phase model is proposed, but the model is complex and it needs to be reinitialized level set function as a symbol distance function(SDF), which greatly reduces the segmentation efficiency. However, there is no need to reinitialize the level set function in the iterative process, and LBF model can be segmented effectively by the image with intensity inhomogeneity. But the model can only be applied to the two phase images, which can’t be applied to the segmentation of magnetic resonance images(MRI).This paper proposes a new multi-phase image segmentation algorithm based on HLBF model, the proposed model replaces the Gauss kernel function in the original LBF model with the new kernel function to improve the time efficiency. Meanwhile, the HLBF model is further integrated into the variational level set of multi-phase image segmentation strategy to achieve the segmentation of multi-phase image with intensity inhomogeneity. Before the formal segmentation, the image is pre-processed by the method of dimension reduction and histogram equalization for good segmentation results. And the proposed model is compared with the LBF model, the PC model and the PS model, and then the segmentation results of the model are analyzed in two aspects. Finally, the segmentation results of each model are analyzed from two aspects: the segmentation time efficiency and the segmentation accuracy.By statistical analysis of experimental results on multiple magnetic resonance images,the results show that the proposed HLBF multi-phase model can segment the white matter, gray matter and cerebrospinal fluid in the brain.The proposed model has advantages over the traditional segmentation method in terms of time efficiency and accuracy.