Research On Vascular Centerline Extraction From CTA Image And Interventional Guidewire Simulation

Vascular interventional surgery robot system is the most widely used clinical auxiliary diagnosis and treatment system in current minimally invasive vascular interventional surgery,which is characterized by less radiation hazard,high surgical accuracy,short operation time and etc.For this system,interventional guidewire simulation and vascular centerline extraction are particularly important,which largely determines whether the performance of Image Guided Surgery is excellent.However,due to artifacts,noise,clot and tumor occlusion,patient movement and etc,the image features of vascular centerline and guidewire are often blurred,resulting in poor extraction and simulation effect.Meanwhile,interventional guidewire simulation and vascular centerline extraction can be classified as the same feature extraction problem.Therefore,in order to solve the feature extraction problem,this dissertation carries out algorithmic research,aims to find out the shortcomings of current research,and facilitates more accurate traditional diagnosis and treatment.Regarding guidewire simulation,intraoperative guidewire pathway of the 2D coronary artery,3D coronary centerline,and 3D carotid artery centerline are used to carry out algorithmic research.At present,the guidewire simulation algorithm described in published literature may have some defects.For example,in the guidewire simulation phase,the surgical path(reference guidewire)has a track change because of complex clinical environment,particularly,vascular motion,blood flow velocity,and etc.All such interference factors increase the difficulty of real-time simulated surgical path(reference guidewire).To address these challenges,this dissertation redefines the mathematical model of process objects through geometry characteristics of the reference guidewire,and proposes some real-time and accurate guidewire simulation algorithms.Regarding vascular centerline extraction,this dissertation mainly participates in the cardiac CTA(Computed Tomography Angiography)coronary tree extraction competition organized by Rotterdam Coronary Competition Institute.At present,the cardiac CTA coronary tree automatic extraction algorithm described in published literature have some defects.For example,in the cardiac CTA image acquisition phase,the image may have low quality due to the complex structure of vessels and peripheral tissues,image noise,partial volume effect,vessel plaques,artifacts,etc.All such interference factors increase the difficulty of complete coronary tree extraction from the cardiac CTA image.To address these challenges,this dissertation redefines an abstract distribution of vascular centerline characteristics through geometry characteristics of vascular angiography and proposes a real-time and accurate cardiac CTA coronary tree automatic extraction algorithm using bidirectional minimal path propagation with backtracking.Besides,to highlight clinical practicability of the proposed algorithm,this dissertation designs an Image Guided Surgery in the vascular interventional surgery robot system and evaluates comprehensive performance of the proposed algorithm.The main contributions of this dissertation are summarized as follows:Aiming at the subject of intravascular intervention guidewire simulation,this dissertation conducts a Systematic Mapping Study.According to research papers published from2007 to 2019,some challenges faced by guidewire simulation are summarized in aspects such as accuracy,real-time,etc;the representative techniques or methods used in the guidewire simulation are summarized in aspects such as guidewire model,collision detection,etc;some promising research directions for future work are identified,such as static guidewire simulation technique,guidewire tip modeling method,and etc.Aiming at solving the problems of very poor real-time,low accuracy and very weak robustness of current guidewire simulation algorithm,this dissertation proposes the following two algorithms:(1)a guidewire simulation algorithm based on an AOE network(Activity On Edge Network),where an iterative refinement algorithm based on merging smaller edge strategy is put forward to obtain some maximum curvature points,the critical path algorithm of AOE network is used to calculate the shortest path,and a surgical path optimization algorithm based on two ends of the guidewire is put forward to preplan a surgical path.(2)a guidewire simulation algorithm based on a fusiform ternary tree model,where an iterative refinement algorithm based on splitting larger edge strategy is put forward to obtain some maximum curvature points,the optimal path algorithm of fusiform ternary tree is used to calculate a shortest path,and an improved surgical path optimization algorithm based on two ends of the guidewire is put forward to preplan a surgical path.Aiming at dealing with the deficiency of current cardiac CTA coronary tree automatic extraction algorithm,this dissertation proposes a cardiac CTA coronary tree automatic extraction algorithm using bidirectional minimal path propagation,where a bidirectional minimal path propagation with backtracking algorithm based on symmetric convexity energy is proposed to extract vessel centerline,the Dijkstra algorithm is used to calculate an optimal path per coronary artery,and an evolutionary algorithm of coronary centerline based on the Open-Snake model is proposed to optimize coronary centerline.