Font Size: a A A

CT And MRI Medical Image Fusion Method Research

Posted on:2012-07-10Degree:MasterType:Thesis
Country:ChinaCandidate:R S LiFull Text:PDF
GTID:2218330338454036Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
In medical imaging technology rapid development today, used for clinical diagnosis and treatment of new imaging equipment constantly to be developed, the Treasury of the different modes of medical image for medical researchers provided Intuitive,rich and qualitative analysis of human physiological information, but due to the different imaging principle, the medical images that reflect the perspective of the human body is different messages, have focused on information content, single-modal medical image diagnostic information can not be comprehensive, so they have their own scope of application and limitations. Medical image fusion technology is the same part of the different medical image of image information integration, get together on the same site more comprehensive, more intuitive and more reliable new medical image, thus for doctors in clinical treatment make more accurate diagnosis of more appropriate, and puts forward effective treatment for the scheme.This paper takes CT and MRI medical image as an example for medical image fusion algorithm and study in theories. First, let CT and MRI medical images registration, the strict alignment of two images in registration is the basis of accurate and re-integration of the image. The quality of CT image and MRI image registration will be directly impact on the effect of fusion, the precise image registration and suitable mage fusion method both in close alliance is a guarantee to provide for doctors'medical diagnosis. In the registration process based on Fourier transform phase correlation theory to obtain the initial position's geometric transformation parameters of two images, this greatly reducing the search area to improve the registration rate, through calculating the Normalized mutual information reducing the traditional method with image registration error rate, Using GPVE methods to improve produce local extremism interpolation phenomenon, Test shows that the registration method both in matching effect, iterations, registration precision or registration are superior to the traditional s image registration method based on mutual trust.After FTNMI registration operation ,this paper puts forward the fusion method of combining NSCT and BPCNN .this method has already registration's images adopts NSCT technology to multiple directions, multi-scale decomposition, acquire each band-pass sub-band coefficient and low frequency sub-band coefficient, then use BPCNN technique to determine the image fusion rules, the selection of fused image fusion coefficient, eventually corresponding NSCT inverter by change gain the effect excellent CT and MRI fusion image. Experimental results show that the presented in the detail of image fusion method with more information of messages, image with increasing image contrast than traditional aspects based on wavelet transform is better Contourlet transform or fusion effect.
Keywords/Search Tags:Image registration, Image fusion, Normalized mutual information Non-subsampled Contourlet transform, Bidirectional search PCNN-BPCNN
PDF Full Text Request
Related items