Research On Multi-Modal Medical Image Registration Based On Demons Algorithm

With the rapid development of modern computer science and medical imaging technology,medical images have become a prerequisite for clinical diagnosis and treatment by doctors.However,each medical imaging equipment has their own advantages and disadvantages,and different tissues and organs of human body have different characteristics,an imaging equipment that can not directly examine the human body and diagnose illness,together medical images from different imaging equipment complement and interwork with each other.Therefore,in clinical practice,it usually need to make comprehensive use of patient's multi-modal medical images to improve the effectiveness of diagnosis and therapy,medical image registration technology provides an important technical support for it.The Demons algorithm based on gray medical image registration,using all grayscale information of images and avoid the error caused by too much manual intervention,which is a full automatic and efficient registration algorithm,so it is favored by researchers in the field of medical image registration.This dissertation studies and improves the Active Demons medical image registration algorithm from the requirement of clinical diagnosis and treatment.The main research works are as follows:(1)This paper introduces the theory of precise histogram normalization and improves the Active Demons medical image registration algorithm shortages for Demons algorithm is only used for mono-modal images registration,and the direction of deformation cannot be determined when the gray gradient is very small.Firstly,the method of precise histogram normalization is used to transform the mode of floating image,and eliminate the gray level difference between multi-modal images.Secondly,according to the principle of symmetry to average the grayscale gradient information of the images,and use Bilateral filter instead of Gaussian filter in the regularized transform.Finally,in order to improve the accuracy of the registration results and the efficiency of the algorithm,the image multi-resolution strategy was introduced into the registration algorithm.Experimental results demonstrate that the algorithm not only applicable to multi-modal images registration,but also can improve registration accuracy and efficiency.(2)In order to further improvement of registration accuracy,This article introduces the theory of local entropy and improves the Active Demons medical image registration algorithm shortages for the Active Demons algorithm only register the same modal images and lead to misregistration when gray gradient of the image is very small.Converting reference image and floating image to mono-modal images,namely the local entropy images.And improved Active Demons algorithm,which the algorithm is added with the square of the gray gradient of the floating image and the reference image,is used for registration,meanwhile,usingBilateral filter instead of Gaussian filter in the regularized transform to keep the detail edge information of the image better.Experimental results demonstrate that the algorithm enhances registration accuracy and the image quality.