Development of the semi-automatic segmentation software for 3D modeling of cerebral vessels

Biomechanical modeling is one of the stages of preoperative planning of reconstructive surgical interventions and allows one to choose and justify the choice of a particular variant of the operation. It is required to create a patient-specific 3D solid geometric model of the object under study to perform biomechanical modeling. This task can be solved by processing computed tomography (CT) or magnetic resonance imaging (MRI) data. However, manual procedure for CT or MRI data processing is quite time-consuming. Therefore, there is a problem to automate the processing phase of these data in order to speed up the process of building models and improve their accuracy and quality. This problem remains unresolved to date. Semi-automatic processing of CT or MRI data uses various methods of image segmentation. The most popular and effective methods are active contour model and frontal growth method. Using these methods, one can segment the objects of interest from DICOM files and transfer them for further operation. In this paper, we present the results of the software development of three image segmentation methods: automatic method, base-color curve method, and recursive method of frontal growth. The work of these methods is illustrated on the basis of MRI data of cerebral vessels with a contrast agent. The results of comparative analysis of the developed methods of image segmentation are given in this article. The developed software allows to download DICOM images in semi-automatic mode and to obtain on their basis 3D models of vessels and also to transfer them for further processing to computer-aided design systems.