Research And Design Of Medical Image Registration System

As one of the basic methods of image processing,image registration plays an important role in medical field.In the era of high morbidity and mortality of malignant tumors,precision radiotherapy has posed new challenges to image registration technology.Both the precise segmentation and localization of the GTV and its surrounding normal organs as well as the comprehensive diagnosis and treatment of patients with a variety of image information,are inseparable from high-precision image registration.In addition,a good image registration system can not only make the doctor work efficiently,but also improve the accuracy of doctors' judgment.Based on the current demand,it has become an urgent need in the medical field to develop a registration system with high precision,complete functions and friendly interface.Due to the low accuracy of multi-mode registration and non-rigid registration,and the uneven quality of registration system,the traditional image registration algorithm based on mutual information is improved in this paper.And the corresponding image registration system is designed and developed.The main research and development contents are as follows:(1)Aiming at the problem that mutual information registration algorithm is prone to local extreme value,two regular terms including bending energy penalty and Euclidean distance between landmark pairs were added on the basis of the original objective function.And each weight in the objective function was reasonably controlled to complete image registration of different dimensions,different parts and different modes.Through a large number of experiments,it is proved that this method has a good effect on the image registration with much soft tissue and big deformation.(2)Integrated medical image Insight Segmentation and Registration Toolkit ITK,medical image Visualization Toolkit VTK,graphical user interface Qt and medical image registration Toolkit Elastix under Windows system to build the system development platform.The platform can provide rich image registration and visual analysis methods for system development.(3)Under the system development platform,we development an medical image registration system which including image registration,visual assessment,deformation field processing etc.This system supports standard DICOM medicalimage file,it also provide mutual information,correlation coefficient of classic registration method and the registration method based on mutual information and the regularization.(4)During the functional testing of this system,aiming at the problem that poor effect on multi-mode registration due to the large difference between the images,for multi-mode registration,we use the intermediate image as a bridge to complete image registration step by step,the experimental results show that this strategy can simplify the complexity of the registration and meet the requirements of high precision registration.(5)The system test results show that this registration system has full functions,friendly interface,strong scalability and other characteristics.