Reducing artifacts occurring due to vessel overlaps in a four-dimensional angiography dataset

A computer-implemented method for providing a trained function, the computer-implemented method comprising:

• receiving input training data comprising:
• two-dimensional digital subtraction angiography projection images showing a blood circulatory system and having been acquired at respective time points in a specific time interval;
• and
• a three-dimensional vessel dataset of the blood circulatory system or a reduced dataset derived therefrom;
• receiving output training data comprising distribution data relating to a concentration of a contrast agent in the blood circulatory system at the respective time points for the three-dimensional vessel dataset or for the reduced dataset, wherein the output training data is related to the input training data;
• training a function based on the input training data and the output training data, wherein output data is generated by applying the function to the input training data and parameters of the function are adjusted based on a comparison of the output data with the output training data;
• and
• providing the trained function.

An image processing device comprising:

• a determination unit configured to determine a four-dimensional angiography dataset, acquired with administration of a contrast agent, of an acquisition region of interest of a blood circulatory system of a patient, having a reconstruction subunit for reconstructing a three-dimensional vessel dataset of the blood circulatory system from two-dimensional digital subtraction angiography projection images showing the blood circulatory system and having been acquired at respective time points in a specific time interval, and a backprojection subunit for determining the four-dimensional angiography dataset by backprojection of the two-dimensional digital subtraction angiography projection images into the three-dimensional vessel dataset;
• a computing unit configured to generate distribution data relating to a concentration of the contrast agent in the blood circulatory system at the respective time points for the three-dimensional vessel dataset or for a reduced dataset derived therefrom, wherein the distribution data is generated as output data of a trained function, and wherein the trained function is applied to the two-dimensional digital subtraction angiography projection images and the three-dimensional vessel dataset or the reduced dataset as input data;
• and
• a distribution unit configured to distribute a color intensity of a pixel of a respective two-dimensional digital subtraction angiography projection image of the two-dimensional digital subtraction angiography projection images, wherein the pixel shows a contrast agent filling state according to the distribution data at a respective time point of the two-dimensional digital subtraction angiography projection image across voxels of the three-dimensional vessel dataset that lie along a ray associated with the pixel in the backprojection.

A training system configured to provide a trained function, the training system comprising:

• at least one processor configured to:
• receive input training data comprising:
• (1) two-dimensional digital subtraction angiography projection images showing a blood circulatory system and having been acquired at respective time points in a specific time interval;
• and (2) a three-dimensional vessel dataset of the blood circulatory system or a reduced dataset derived therefrom;
• receive output training data comprising distribution data relating to a concentration of a contrast agent in the blood circulatory system at the respective time points for the three-dimensional vessel dataset or for the reduced dataset, wherein the output training data is related to the input training data;
• train a function based on the input training data and the output training data, wherein output data is generated by applying the function to the input training data and parameters of the function are adjusted based on a comparison of the output data with the output training data;
• and
• provide the trained function.