Multiple View Geometry in Computer Vision
Multiple View Geometry in Computer Vision
Dynamic Cone Beam Reconstruction Using a New Level Set Formulation
MICCAI '09 Proceedings of the 12th International Conference on Medical Image Computing and Computer-Assisted Intervention: Part II
MICCAI '09 Proceedings of the 12th International Conference on Medical Image Computing and Computer-Assisted Intervention: Part I
Detecting patient motion in projection space for cone-beam computed tomography
MICCAI'11 Proceedings of the 14th international conference on Medical image computing and computer-assisted intervention - Volume Part I
| Hi-index | 0.00 |
The 3-D reconstruction of cardiac vasculature using C-arm CT is an active and challenging field of research. In interventional environments patients often do have arrhythmic heart signals or cannot hold breath during the complete data acquisition. This important group of patients cannot be reconstructed with current approaches that do strongly depend on a high degree of cardiac motion periodicity for working properly. In a last year's MICCAI contribution a first algorithm was presented that is able to estimate non-periodic 4-D motion patterns. However, to some degree that algorithm still depends on periodicity, as it requires a prior image which is obtained using a simple ECG-gated reconstruction. In this work we aim to provide a solution to this problem by developing a motion compensated ECG-gating algorithm. It is built upon a 4-D timecontinuous affine motion model which is capable of compactly describing highly non-periodic motion patterns. A stochastic optimization scheme is derived which minimizes the error between the measured projection data and the forward projection of the motion compensated reconstruction. For evaluation, the algorithm is applied to 5 datasets of the left coronary arteries of patients that have ignored the breath hold command and/or had arrhythmic heart signals during the data acquisition. By applying the developed algorithm the average visibility of the vessel segments could be increased by 27%. The results show that the proposed algorithm provides excellent reconstruction quality in cases where classical approaches fail. The algorithm is highly parallelizable and a clinically feasible runtime of under 4 minutes is achieved using modern graphics card hardware.