Algorithm for automatic segmentation of arterial bed structures involved in transcatheter implantation of aortic valve prosthesis based on computed tomography data

Abstract

Justification. One of the significant causes of premature mortality and disability of the population is currently the pathology of the heart valves. The most common among this group of pathologies is aortic stenosis. The most popular method of its correc-tion today is transcatheter implantation of aortic valve prosthesis. For the teaching of medical residents and manual training, ana- tomical material and operations on animals are more often used, but this kind of practice has a high cost and is not always avail-able to a wide range of students. Alternatively, virtual surgical simulators are appearing on the market, including a workstation with manipulators and software for computer imaging, but to date, there is no ready-made solution on the market for segmentation of vascular structures involved in transcatheter implantation of aortic valve prosthesis. The Purpose. Development of an algorithm for automatic segmentation of the structures of the arterial vascular bed in the area of transcatheter implantation of the AV prosthesis based on the processing of computed tomography data. Material and methods. In the course of the work, 14 CT studies of randomized patients with no abnormalities and/or pathology of AV and aortic bulb were used. Their analysis and further transformations were carried out using Vidar DICOM Viewer software, SolidWorks 2016, VMTKLab and other existing types of software and libraries. Due to the fact that visualization of the semilunar cusps of the AV during CT is practically impossible, the construction of a virtual 3D-template of the AV was performed taking in- to account the parameters given in the literature. Results. As a result of the first part of the work, a ready-made virtual 3D-template of AV was obtained, capable of changing its ge- ometry under external influences. The performance of the parametric model has been proven. In the course of working with the model, no problems with the geometry during deformation were identified. As a result of the second part of the work, a ready- made STL-model of the vessels from the aortic ostium to the level of the transition of the external iliac artery to the femoral artery was formed. After that, we imported the vessels model into SolidWorks and applied the parameters of the mechanical properties of biological materials. At the last stage, two models were assembled into a single system, as a result of which a ready STL model of the zone of interest with segmented structures was formed. Conclusions. 1. When implementing the developed algorithm for automatic segmentation of the arterial bed structures, ready-made software packages were used, which made it possible to implement the developed algorithm using a fairly simple and orig- inal technique. 2. Considering that the operator is involved in most stages of using the algorithm, segmentation in the developed software is semi-automatic in nature, but the actions performed by the operator do not require deep immersion in the algorithms of the software and are superficial. 3. As a result of the study, an individualized 3D model of the anatomical structures in the area of surgical intervention was obtained for transcatheter implantation of aortic valve prosthesis, which allows the surgeon or stu- dent to work out in detail on a virtual 3D model all stages of the forthcoming surgery.