A Method Of Deformation Registration With Planned CT And Online CT With Preconditioning

In the radiotherapy plan,the researchers will deform the CT image of the tumor patient and the online CT image before radiotherapy to obtain information on the movement and deformation of the tumor and its surrounding organs,improve the radiotherapy plan,and thus reduce the location of the tumor.Increased unnecessary radiation range due to changes in shape reduces radiation damage in normal tissues.The CT images used in the above process are generally acquired by ordinary scanning or enhanced scanning.In order to observe the lesion more clearly,a contrast agent is intravenously injected in the enhanced scan,and blood vessels of the acquired CT image show extremely high HU values.On-line CT usually does not perform enhanced scans,and the corresponding blood vessels will display normal values.Even if the enhanced scans are performed,it is generally not the same as the location of the planned CT.Therefore,the HU value of the blood vessels in the two scan images may vary greatly.The existing registration methods are designed based on the consistency of the density.Therefore,the huge differences in the HU values of the above-mentioned blood vessels may cause the registration results of the planned CT and online CT images to be distorted and misfit.Aiming at this problem,this paper proposes a deformation registration method with preprocessing.For the registration algorithm proposed in this paper,detailed comparison and verification experiments were conducted.In the study,based on the information of the CT image organization and structure,the blood vessels were segmented using the threshold segmentation method,and the largest communicative region in all segments was used as the initial segmentation enhancement vessel,followed by the segmentation of the enhanced vascular region with a 5 mm extension.As an externally augmented blood vessel,blood vessels are filled with a fixed HU value.Finally,after the image is filled,the Demons algorithm is used for deformed registration.In the study,gray-scale Demons deformationregistration was performed on the planned CT and online CT images without preprocessing,Demons deformation registration was performed on the planned CT and on-line CT images with the same parameters and the two parameters were matched.The data of the quasi-results were compared,and mutual information(MI)and mean square error(MSE)were used to judge the registration results.Planned CT and online CT image deformation registration uses a direct gray-based algorithm,and deformation registration uses an algorithm based on automatic organ delineation.The experimental results show that the deformation registration method with pretreatment proposed in the article can better solve the CT mismatch problem caused by water-soluble iodine contrast agent.In the case of scanning enhanced CT,this study will help to improve the accuracy of planning CT and online CT registration,and it will also improve the quality of radiation therapy.