We propose a method for extracting quantitative hemodynamic information, such as the time varying volumetric flow rate from image sequences in the form of 2D Digital Subtraction Angiography (DSA) acquisitions. An intermediary 3D+t image representation is created by backprojecting the 2D intensities to a working volume. Dense 3D flow field estimation adapted to the tubular vascular geometries, is then used to recover displacement fields representing fluid motion. Whereas previous related attempts performed calculation within the 2D image domain only, our proposed method quantifies blood flow directly within the 3D vascular representation, which allows constraints motivated by physical principles, to be applied for regularization. From the extracted dense 3D flow fields, quantities of clinical interest such as the volumetric flow rate are derived. Our experimental setup for validating our proposed algorithm involves synthetic and phantom datasets. Whereas the phantom data results only allows for qualitative result inspection due to missing ground truth information, for the synthetic cases, flow rate measurements are quantitatively validated.     «

We propose a method for extracting quantitative hemodynamic information, such as the time varying volumetric flow rate from image sequences in the form of 2D Digital Subtraction Angiography (DSA) acquisitions. An intermediary 3D+t image representation is created by backprojecting the 2D intensities to a working volume. Dense 3D flow field estimation adapted to the tubular vascular geometries, is then used to recover displacement fields representing fluid motion. Whereas previous related attempts...     »