Image Processing And Analysis For Aortic Dissection Surgery

Aortic dissection(AD)occurs when an injury to the intima,results in blood continuously flowing into the aortic wall to form an extra lumen separated from the original lumen.This phenomenon seriously affects the blood circulation system and threatens the life of the AD patient.More and more doctors choose thoracic aortic endovascular repair(TEVAR)as the preferred treatment for patients with aortic dissection.The remodeling of the true lumen is achieved by the occlusion of the tear and the support of the stent graft.With the application of endovascular treatment in AD patients,the process of preoperative planning,postoperative prediction and individualized treatment all require clinical imaging data.However,doctors spend a lot of time on image acquisition and analysis,but the subjective experience will affect the accuracy of diagnosis and treatment.Therefore,automatic and accurate computer-aided diagnosis and treatment planning has attracted widespread attention from the clinical and academic community.The purpose of this thesis mainly focuses on the actual clinical problems of AD patients.The methods of preoperative automatic angle planning,intraoperative interventional imaging assistance and postoperative follow-up and prognosis were optimized and innovated.The main work and innovation of this thesis are as follows:(1)An adaptive Angle optimization algorithm based on patient’s individualized image is proposed to automatically select the optimal viewing angle.In the process of endovascular intervention,DSA is used as the image to guide stent placement.The selection of C-arm angle depends on manual selection by experts,which is a key factor in imaging quality.However,the selection process is influenced by the doctor’s subjective experience and it also brings additional radiation exposure to the patient and doctor.Without introducing additional radiation dose,we propose an adaptive method that utilizes the CTA image acquired before TEVAR to automatically calculate the optimal C-arm angle.The projection foreshortening rate(PFR)is used to obtain a full display of the aortic arch,and the projection overlapping rate(POR)is used to avoid the overlap of branches on the aortic arch.There is no significant difference between the angle determined by experienced experts and automatic algorithm.The optimal viewing angle algorithm has the potential to optimize the planning time and reduce the radiation exposure during TEVAR.(2)A projection network with spatio-temporal information(PNet-ST)is proposed for the aortic segmentation of DSA.During the process of TEVAR,the aorta is partially displayed in the DSA frames at different times due to its morphological characteristics,which cannot provide doctors with the entire aortic structure to guide stent placement.Combined with the characteristics of DSA images.The spatial encoder module is introduced to preserve all the frames for feature learning to recognize the anatomical structure of each frame fully.The network fuses the contour features preserved by the spatial encoder and transmits the comprehensive sequence information through max intensity projection.Dense-biased connection merges multi-receptive field feature maps to obtain more advanced feature information.The small receptive field is used to identify the blurred boundaries,and the large receptive field is used to consider the contour information of the aorta.Our experiment results show that the proposed PNet-ST outperforms the state-of-the-art methods in aortic segmentation.The segmentation results of our PNet-ST can offer assistance in endovascular surgery and help the doctors observe the entire aorta to place the stent more accurately.(3)A retrospective analysis based on three-dimensional morphological parameters is proposed to correlate morphological parameters with the postoperative outcomes of AD patients.More attention should be given to the complicated AD patient to evaluate the risk of false lumen expansion and to determine the necessity of a second intervention.Morphological parameters of the aorta are important factors in aortic remodeling.The three-dimensional morphological parameters were analyzed based on the imaging-reconstructed models.AD patients were divided into the Stable and Enlarged group based on the change of FL volume after TEVAR,and the current study tried to correlate the different 3D morphological parameters over time with the FL development after TEVAR.The results of the morphological analysis of this study indicate that several morphological parameters before TEVAR are associated with poor remodeling after TEVAR.This parameter can provide scientific decision-making for doctors in preoperative planning.On the other hand,some of the 3D morphological parameters before TEVAR could act as predictors to identify those patients with a high risk of poor aortic remodeling after TEVAR.