Study On Brain Medical Image Registration Algorithm

Medical image registration is an important part of computer-aided diagnosis.It is the first step of image fusion.As the basis of image fusion,the accuracy of registration directly affects the fusion quality.Moreover,image registration can play a significant role in the process of surgery and radiation therapy.In addition,the registration of images between different modalities can provide the doctors with functional information and anatomical information of the lesion site at the same time,so as to improve the accuracy of diagnosis.With the rapid development of computer technology and medical imaging technology,the use of computer for image registration has also become a research hotspot of the two key disciplines.According to the process of image transformation,image registration can be divided into rigid registration and non-rigid registration.Rigid registration is mainly used for the registration of relatively hard tissues,such as the skull,or coarse registration of images.Compared to rigid registration,non-rigid registration is more complex,which is mainly used for soft tissue registration or fine registration.Since they both have their own usefulness in clinical practice,the research in this thesis includes both aspects.This thesis first introduces the background,significance and research status of medical image registration at home and abroad.Then,the basic concept of image registration and the four key points involved are further explained.Next,the mutual information and other key points will be introduced,on this basis,some of them will be improved.Some algorithms in non-rigid registration are introduced later,and an improved algorithm is proposed.In general,the main research work of this thesis is as follows:1)The current research status of image registration technology at home and abroad will be investigated,and the key technologies of brain medical image registration will be summarized.2)The basic knowledge of information theory will be introduced.On the basis of this,the concept of mutual information will be introduced.Aiming at some shortcomings of mutual information,a new brain medical image registration algorithm called IMI will be proposed by adding initial values and image gradient information.3)The non-rigid registration algorithm will be studied.The main research are the registration algorithms which are based on the optical flow field.The original Demons algorithm will be introduced in detail.In addition,part of the algorithms based on it will be introduced.After introducing the concept of image multi-resolution,this thesis will propose a brain medical image registration algorithm based on Demons algorithm to improve the accuracy of registration.The algorithm proposed in this thesis will be tested on clinical medical images and synthetic medical images.Moreover,several other algorithms will also be introduced for comparison.Through comparative experiments,the accuracy of the proposed algorithm in rigid and non-rigid registration will be verified.