Research On Algorithms Of Medical Image Fusion Based On EMD

With the development of medical imaging, a large number of high-precision imaging equipment are emerged, but the imaging mechanism of different forms of imaging equipment is different, so the focus of describing information is different, has its own limitations and usually only better reflects on some aspects'information of human body, Therefore, in order to make an accurate diagnosis or treatment plan for disease, we usually require doing image fusion for a variety of medical image. Long-term practice proved that the fusion method based on multi-scale and multi-resolution can obtain better fusion result. However, the bidimensional empirical mode decomposition as a new multi-scale and multi-resolution image decomposition method has advantages which other multi-scale and multi-resolution methods don't have, therefore, it is applied to medical image fusion in this article. The main content and achievements in this article are as follows:1. Analysis and research the origin, principles, algorithm steps, properties and problems of one-dimensional empirical mode decomposition, and then extend it to two-dimensional, namely the bidimensional empirical mode decomposition, detailedly analysis the algorithm framework and the key technologies of bidimensional empirical mode decomposition.2. In view of the traditional bi-dimensional empirical mode decomposition exists the problem of taking a long time, decomposing in low efficiency and existing serious gray spots in decomposition result, we improve it and propose the improved bidimensional empirical mode decomposition. It sets the maximum neighborhood of every decomposition in the time domain, and does a certain limit for every decomposition, so effectively reduces the gray spots in the decomposition result, in addition in neighborhood it uses the order statistics filter based on spatial domain do envelope estimation for the discrete extreme points, and then obtain the upper and lower envelope, effectively reduce the inefficiency and other problem when applying various interpolation algorithms to form upper and lower envelope, and make the efficiency of the algorithm greatly improved. Finally, the improved algorithm makes the quality and efficiency of decomposition significantly improved. 3. Apply the improved bidimensional empirical mode decomposition to medical image fusion, firstly, use the improved bidimensional empirical mode decomposition decompose the medical image of preparing fusion, then use different fusion rules to fuse the bidimensional intrinsic mode function and Residue. Through compare the fusion result with other fusion algorithm, prove that the improve bidimensional empirical mode decomposition introduced to the medical image fusion is feasible.4. According to the CT image and MRI image's imaging characteristics and the function in the fusion result, proposes the medical image fusion algorithm based on the improved bidimensional empirical mode decomposition and region segmentation. It first uses the variational level set segmentation algorithms segment the CT image into reference region and lesion region, and maps the segmentation model to the MRI image, then also segment the MRI image into reference region and lesion region, second use the improved bidimensional empirical mode decomposition decompose the CT image and MRI image respectively, then according to the properties of different region and different decomposition layer of image propose a new fusion measurement standard to guide fusion. Finally, through experiments proved that the new algorithm can obtain better fusion result.5. Implement a medical image fusion system, mainly implement variety medical image fusion algorithm involved in the study, and evaluate the final fusion result.