Research And Application Of Medical Image Fusion Algorithms Based On Multi-scale Transform

With the fast development of science and technology, medical image fusion technology has become an important part of information technology, and has been widely used in the field of clinical medical imaging diagnosis. The single-modal medical images can not provide enough digital information. In order to overcome the limitations, multi-modal image fusion technology is proposed. The medical image fusion technology can obtain the image which can presents human organs and lesions more enrich, more clear, more comprehensive. Thus it can provide a reliable basis for doctors to diagnose disease and to establish reasonable treatment methods. Firstly, we introduce the basic theory of image fusion, then we focus on medical image fusion algorithm and develop image fusion software, the main research contents are summarized as follows:1) The existing algorithms of medical image fusion do not consider the difference between different source images. In view of this problem, a multi-modal fusion algorithm based on mutual information is proposed. The target image is decomposed by lifting wavelet transform into high frequency sub bands and low frequency sub bands. Because of the correlation between the adjacent low frequency sub bands, and a lot of basic information of the image is in the low frequency sub band, so the fusion rule of low frequency sub bands is regional average energy weighted. For high frequency sub band with low mutual information, the method that gradient energy is combined with local standard deviation is used as fusion rules. For the high frequency sub bands with low mutual information, the method that the bigger edge strength is selected is used as fusion rule. Experiment results of multi group target image fusion shows the method proposed in this paper in much better because of its rich information, clear edge and with good visual features.2) Lots of simulation experiments of medical image fusion based on image mutual information are conducted. In the gray image fusion, fusing images consist of CT image, MRI image, MR-T1 image and MR-T2 image. In the color image fusion, fusing images consist of SPECT image and MRI image. The experiment results show that the fused image of this proposed fusion algorithm is more detailed and retains much more edge details with good visual effect. At the same time, the objective evaluation indicators also show that the proposedalgorithm can significantly improve the performance of fusion image in terms of quantify of information and retain the important information of the source image effectively.3) Pointing at the characteristics of medical image and the new image fusion algorithm, a software of multi-modal medical image fusion is designed. The image fusion software is developed by the Visual C++ programming language. Windows system as a platform. The image is processed by using the library function in OpenCV(Open Source Computer Vision Library).Firstly, the edge texture of the source image and other important information of the image is improved by using image pre-processing method. Secondly, manual and automatic registration method is used in image registration. This method can improve the precision of image registration. Then, different fusion schemes are designed in image fusion: fusion of gray level images and fusion of gray image and color image. The fusion algorithm based on mutual information feature is used in the software of image system. Finally, design image quality evaluation index scheme in image quality evaluation. Based on the OpenCV library function, the system has good real-time performance and stability. Image processing system based on windows platform can better transplant to other platforms. Further improve the generality of the image processing software.