Non-rigid Registration Of Proatate Image Based On Multi-modality

The prostate is an important gonad of man,which is related to the physical health of man and the happiness of life.Prostate cancer is the life killer of man,causing great pain to the patient.Prostate cancer is a malignant tumor that occurs in the prostate,it can cause pain,difficulty in urinating and other symptoms,it can also lead to death seriously.In Europe and the United States,the death rate of man caused by prostate cancer in male cancer disease ranked second,only lower than lung cancer.In China,the incidence rate is increasing due to the aging population in recent years.Radiation therapy is one of the main treatment of prostate cancer,the basic principle is to make use of radioactive substances to kill tumor cells.The technique used in clinical practice is image-guided radiotherapy,which monitors real-time tumor and prostate normal organ through medical imaging equipment during patient treatment.In this process,it is necessary to register the prostate image in order to confirm the patient’s placement and track tumor.Prostate MRI imaging resolution is excellent,prostate CT imaging realistic,anatomical relationship clear,in the radiotherapy process these two modal images need to register.As the prostate is soft tissue,the shape of prostate to deformation in different periods prone,it is not suitable for rigid transformation,therefore,the image of non-rigid registration is required.Non-rigid registration is still a challenge problem in medical imaging processing area compared to rigid registration of images.In this paper,the non-rigid registration algorithm for prostate multi-modal image is mainly studied as follows:1.A non-rigid registration algorithm for multi-modal prostate image based on hybrid model is proposed.This method first segments the target area from the whole image and registers the target area.In the process of registration,on the basis of the traditional Demons algorithm,the multi-resolution strategy is used to register from coarse to fine,and then the B-spline theory is used to simulate the local deformation of the image,and then the steepest descent method is used to optimize the parameters.Finally,the registered image is obtained.2.An improved energy function Demons algorithm for multi-modal prostate image is proposed.This method is based on the traditional Demons algorithm,using the spatial neighborhood information of the image and combined with the maximum a posteriori probability-Markov random field frame.The equivalence of Markov random field and Gibbs distribution is used to construct a new energy function,then the transformation matrix is solved,and the registered image is obtained by bilinear interpolation.3.A non-rigid registration method for fast prostate image based on transform matrix migration is proposed.In this method,we consider one of the CT images and MRI images as the basic images.Then we use the Demons registration algorithm to obtain the transformation matrix between them.When we register the other images,we directly map the resulting transformation matrix to the floating image to obtain the mapped image and then adjusted according to the reference image.The floating image is interpolated by the thin plate spline interpolation method,and finally the registered image is obtained.