Markus Huellebrand, Presenter: Nothing to Disclose

Anja Hennemuth MS, Abstract Co-Author: Nothing to Disclose

Jens Frahm PhD, Abstract Co-Author: Research collaboration, Siemens AG

Lennart Tautz, Abstract Co-Author: Nothing to Disclose

2D phase-contrast (PC) MRI is an established technique for the analysis of vascular hemodynamics. A recently developed real-time MRI technique allows for respective acquisitions under free breathing and without the need for ECG synchronization. However, quantitative evaluations become more complicated than for conventional methods because of potential changes in contrast, the management of multiple cycles without manual interference, and the influence of respiratory displacements. In order to overcome such problems, we developed a new method for the automatic analysis of blood flow parameters from real-time PC MRI.

Real-time 2D PC MRI of the ascending aorta was performed in 5 healthy subjects (mean age 25 years) were at 3-T (TrioTim, Siemens, Erlangen, Germany). Acquisitions were based on a highly undersampled radial FLASH sequence with and without a bipolar flow-encoding gradient (VENC=200 cm/s, flip angle 10°) and image reconstruction by regularized nonlinear inversion. The spatial resolution was 1.33x1.33x6.0mm3 and the temporal resolution corresponded to 40 ms. The images were analyzed with use of the research software prototype CAIPI. After an initial segmentation of the aortic vessel wall, the vessel contour is automatically propagated to all frames using a registration based on a quadrature filter. The results of the automatic analysis were compared to the manual results of three experts.  

The segmentation results of the three observers and the automatic segmentation (duration 39 ± 4 s) were compared pair-wise. The average dice coefficient between observers and the algorithm was 0.86 ±0.04, the inter-observer comparison was 0.92±0.03. The average symmetric absolute surface distance error was 1.09±0.4mm for the algorithm and 0.71±0.22mm for the observers. The mean absolute error of the stroke volume was 4.67±2.28ml for the algorithm and 6.14±3.5ml for the observers.

The comparison of the manual and automatic quantification shows good agreement. Because no manual correction is needed, the proposed method is suited for the automatic analysis of the temporal evolution of flow velocities, peak velocities, stroke volumes and flow rates over multiple cardiac cycles.

Automatic quantification of real-time 2D PC MRI enables analysis of patients with aperiodic heartbeats (e.g. arrhythmias) and monitoring of hemodynamic responses to stress or physiologic manoevers.

Huellebrand, M, Hennemuth, A, Frahm, J, Tautz, L, Automatic Quantification of Blood Flow from Real-time Phase-contrast MRI.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13016377.html