Determining a three-dimensional model dataset of a blood vessel system with at least one vessel segment

摘要

A three-dimensional model dataset of a blood vessel system of a patient including at least one the vessel system is determined from a number of projection images, which have been recorded from different recording angles, of the blood vessel system The projection images are divided up into image areas each containing at least one pixel. A feature vector is determined for each of the image areas. Classification information, which describes how the respective image area belongs or does not belong to a vessel segment of the blood vessel system defined in accordance with anatomical specification data, is defined for each of the image areas by applying a classification function to the feature vector assigned to the image area. The classification function has been trained by training data records annotated with classification information obtained from at least one person other than the patient. The blood vessel system in the projection images is segmented by grouping image areas with the same classification information. The three-dimensional model dataset is calculated as a function of the segmented projection images and the classification information.