Research On Space Registration Of Image-guided Vascular Intervention Operation System Based On Electromagnetic Localization

Endovascular minimally invasive surgery(MIS) has been widely applied insurgeries due to its merits of less bleeding, less trauma and quick recovery. The imagenavigation system is extremely important to intervention operation as it can providedoctors with a lot of important information which helps to make surgical proceduressafer, faster and more accurate. However, Digital subtraction angiography is usuallyused during the surgery for positioning, which may lead to fussy intra-operativeimaging and X-ray radiation. To solve these problems, electromagnetic trackingtechnology is applied to image navigation system to help locate and track the position ofthe end of the robotic catheter. The spatial registration is a key issue that needs to besolved in image navigation system of vascular interventional surgery usingelectromagnetic sensor for positioning and tracking. Without using artificial markers, anpath-based algorithm has been researched to implement the association between thepatient coordinate system and image coordinate system by searching for thecorrespondence between the path data of the tip of the robotic catheter and the centerline data extracted from the three-dimensional model reconstructed from pre-operationCT slices.The principle of space registration using electromagnetic tracking has beendescribed and the mathematical model has been built for registration. Considering thenecessity of initial transformation matrix, the initial registration algorithm and theprecise registration algorithm based on the former have been designed based on thevascular silicone model.Considering the large deformation between the path line and the correspondingcenter line in the initial registration process, an improved ICP algorithm has beendesigned by preprocessing the path data and changing the classical ICP strategy ofsearching for the correspondence between two data sets.The dynamic precise registration algorithm and the optimization method used tominimize the objective function of this phase which contains multiple parameters havebeen studied. The mechanism of selection of vascular branches and the method ofcomputation of corresponding point pairs have been built to implement the preciseregistration in real time, meanwhile, an stable selection mechanism to improve therobustness of choosing corresponding points pairs when the catheter was located aroundthe bifurcation has been designed. The method of recognition of the corner points hasbeen designed based on the large registration error brought by initial process forconversion parameters’ compensation. To eliminate the influence on the distancebetween the path of robotic catheter and the corresponding center line brought by theradius of different vascular branches, an objective function with the parameter of radius of vascular introduced to has been built. Considering the special characteristic of thetarget function whose parameters were partially coupled, the method of separating andconverting the parameters has been researched and the alternative optimizationalgorithm to find the minimum of the function using golden search algorithm andimproved Powell algorithm has been designed.The implementation method of the registration function based on the imageguidance and the principle of making virtual path data of the robotic catheter have beenintroduced. Then based on the virtual path data, the method of calculating theregistration error is researched. The accuracy and feasibility of the proposed algorithmhave been validated by applying the algorithm to six groups of virtual path data andcalculating the registration error. The final average registration error is2.6millimeter.