Research On Multimodality Medical Image Fusion Based On Multi-scale Decomposition

With the rapid development of computer science technology and sensor technology,the modality medical images used by doctors are increasing gradually.However,because of the different imaging principle of sensor,the data obtained from the single-model medical images is limited,which lead to the difficult for clinicians to provide the reliable and effective medical information.In order to solve this problem,a multimodality medical image fusion algorithm has been put forward by academicians.The image obtained from the same or different imaging devices can be appropriate matching and superposition.to increase the complementary and effective information,which is helpful to clinical diagnosis and treatment.At present,the medical image fusion has been widely used in the field of medical image processing and played an important role in the field.Hence,based on the theory of multi-scale decomposition,T mixture model and low rank matrix model,the thesis aim at researching the medical image fusion and obtained some innovative research results.The main innovations of this thesis are as follows:1.Aiming at the problem of poor data fitting and the loss of detailed information in Gauss distribution,a medical image fusion algorithm using Contourlet transform and T mixture model is proposed.Firstly,the RGB space of MRI and PET image are transformed to GIHS space separately,and then using Contourlet transform decompose the intensity of GIHS to high frequency and low frequency subbands.The absolute maximum method is applying to the high coefficients.For low frequency coefficients,the proposed T mixture model using EM algorithm is used to fuse them.Lastly,the fusion results are transformed by the inverse Contourlet transform.The effectiveness of the proposed algorithm is verified by experiments on the brain images of MRI and PET.2.Pointing at the problem of traditional medical image processing losing detailed information and the basing on the single pixel,an algorithm based on NSCT and Improve Robust Principal Component Analysis for medical image fusion has been introduced,which is also called NSCT_IRPCA.The proposed NSCT_IRPCA is based on Robust Principal Component Analysis(RPCA).NSCT_IRPCA algorithm is divided into three steps:(1)using Nonsubsampled Contourlet transform(NSCT)to decompose the source images to high frequency and low frequency subbands;(2)decomposing to the low-rank key information and sparse saliency information ,the improved RPCA is applied to the low frequency of saliency detection;(3)the absolute value of the elements in the represents the significance of the feature,so that the absolute value of the fusion rule are used to fuse the main information of the two kinds of source images.At the same time,the same fusion rules are also applied to fuse the sparse saliency information.MRI and PET brain images as the source image are applied to the experiment of proposed algorithm.The effectiveness of the NSCT_IRPCA is verified by the experiment.