high performance computing on graphics processing units: hgpu.org

Anatomical and functional information of cardiac vasculature is a key component in the field of interventional cardiology. With the technology of C-arm CT it is possible to reconstruct static intraprocedural 3-D images from angiographic projection data. Current approaches attempt to add the temporal dimension (4-D). In the assumption of periodic heart motion, ECG-gating techniques can be used. However, arrhythmic heart signals and slight breathing motion are degrading image quality frequently. To overcome those problems, we present a reconstruction method based on a 4-D time-continuous B-spline motion field. The temporal component of the motion field is parameterized by the acquisition time and does not assume a periodic heart motion. The analytic dynamic FDK-reconstruction formula is used directly for the motion estimation and image reconstruction. In a physical phantom experiment two vessels of size 3.1 mm and 2.3 mm were reconstructed using the proposed method and an algorithm with periodicity assumption. For a periodic motion both methods obtained an error of 0.1 mm. For a non-periodic motion the proposed method was superior, obtaining an error of 0.3 mm / 0.2 mm in comparison to 1.2 mm / 1.0 mm for the algorithm with periodicty assumption.For a clinical test case of a left coronary artery it could be further shown that the method is capable to produce diameter measurements with an absolute error of 0.1 mm compared to state-of-the-art measurement tools from orthogonal coronary angiography.Further, it is shown for three different clinical cases (left/right coronary artery, coronary sinus) that the proposed method is able to handle a large variability of vascular structures and motion patterns. The complete algorithm is hardware-accelerated using the GPU requiring a computation time of less than 3 minutes for typical clinical scenarios.