Logo of the Physikalisch-Technische Bundesanstalt

Flow quantifications techniques at high and ultra-high fields

Being a very versatile imaging technique, magnetic resonance imaging allows quantifying blood flow in addition to soft tissue properties such as T1- and T2-relaxation, diffusion, perfusion and many more. This allows to gain extensive knowledge on normal and pathologically altered hemodynamics in various cardio- and cerebrovascular diseases. Therefore, we are investigating a technique called 4D flow MRI at high and ultra-high fields to measure non-invasively and temporally resolved, complex, three-dimensional velocity vector fields at various field strengths. This development is performed in close collaboration with Charite hospital (see Fig.1) [1].

Fig. 1: Time-resolved animation of the blood velocity as a function of the cardiac cycle through the aortic arc of a patient with bicuspid aortic valve (courtesy of workgroup Prof. J. Schulz-Menger, Charite, Berlin).

Blood velocity quantification in the body at Ultra-High Fields

Quantifying the blood velocity vector at UHF is challenging mainly due to the heterogeneous B1+ field distribution within the human body at UHF [2]. The lab has focused on developing methods to counteract the B1+ field and flip angle heterogeneities for blood velocity measurements. For example, in a joint work with the German Cancer Research Center, Heidelberg, Germany, dynamic pTx methods have been applied (see section " RF Pulse design and parallel transmission") to flow quantification techniques and the impact of the RF pulses on the flow quantification has been investigated. Using this development, the flip angle throughout the target region could be homogenized while correct quantification of the velocity vector was demonstrated [3]. Based on this work, the impact of the non-instantaneous encoding process on selected hemodynamic parameters has been studied recently [4].

4D flow imaging in the carotid arteries at Ultra-High Fields

The carotid bifurcation is a location where plaques may accumulate with age. In case the plaques detache from the vessel wall they can travel with the blood into the brain where they may occlude vessels and cause stroke. Indications exists that the blood flow has impact on the formation and growth of such plaques, therefore, the group is interested in the development of high-resolution MRI-based 4D flow methods to quantify the blood velocity vector within the carotid arteries and the bifurcation (see Fig. 2). In recent joint work with the Center for Magnetic Resonance Research (CMRR) in Minneapolis, USA, simultaneous multi-slice (SMS) methods have been applied to 4D flow MRI at UHF to measure the velocity vector field in both carotid arteries [5]. In further work, SMS techniques have been applied to measure the velocity of the myocardium during cardiac contraction [6,7].

Fig. 2: Left: maximum intensity projection image showing the carotid bifurcation derived from a gradient-echo image obtained at 7 Tesla. Right: segmented carotid artery with color-coded velocity overlay.

4D Flow Imaging and 2D-Selective Excitation

Still, long acquisition times prevent the routine clinical application of 4D flow imaging. One promising acceleration approach restricts the excitation in two dimensions by employing 2D-spatially-selective excitation (2DRF). Spatial selectivity is achieved by applying gradients along both selective directions during RF-application using e.g. a spiral excitation-k-space trajectory as shown in Fig. 3. For 1D-selective excitation (1DRF), a gradient along only one dimension is required. Since only the excited region needs to be encoded, the FOV can also be reduced in two dimensions. Consequently, acquisition times could be reduced to approximately ¼ of the original scan time by using 2DRF 4D flow instead of conventional 1DRF 4D flow (Fig. 4) [8]. Furthermore, the reduced size of the excitation pattern limits potential motion artefacts to the excited region. Resulting velocity quantifications agree with conventional methods using 1D excitations (Fig. 5). To counteract the increase in RF pulse duration and thus the decrease in temporal resolution, additional methods can be employed to restore the temporal resolution of a conventional scan with 1D excitation [9].

Fig 3: (a) Spiral k-space trajectory (blue line) and k-space weighting (colored mesh) needed to excite a cylindrical excitation pattern (b). The weighting of k-space is achieved by applying a corresponding RF pulse. (c) Anatomical image with 1D slab selective RF pulse exciting the full field-of-view for comparison.

Fig. 4: Magnitude and 3-directional velocity in the Circle of Willis acquired at 7T with a 2D-selective excitation RF pulses (2DRF) using fat saturation and for comparison with SINC excitation. Acquisition times were from top to bottom 5.19±0.39 min and 18.32±0.97 min.

Fig. 5: a) Comparison of time-resolved flow through the basilar artery and internal carotids between i) 1D-selective excitation with full FOV and ii) 2D-selective excitation with half a FOV. b) GRE Magnitudes (gray-scale) and velocities in z-direction (color-coded) of the slices in which flow is analysed in (a).

Flow Magnetic Resonance Fingerprinting (Flow-MRF)

For certain pathologies, such as for the characterization of atheroma plaque progression, not only the blood hemodynamics are of interest but also the composition of the surrounding tissues, in particular the vessel walls and the plaque. A novel method termed Flow-Magnetic Resonance Fingerprinting (Flow-MRF) [10] which has recently been proposed in a joint development with the German Cancer Research Center (DFKZ), allows this joint quantification of blood flow and relaxation times of surrounding tissue simultaneously in a single acquisition.

The method relies on an MRI acquisition with pseudorandomized parameters such as the flip angle and - in the case of flow MRF - the velocity encoding moment (Fig 6). Such pattern generates for each tissue or flowing region a unique signal evolution (or ‘fingerprint’) that depends on the tissue's properties. Then, theoretical signal evolutions generated from these acquisition parameters are computed for a wide variety of tissues and collected in a database, called ‘dictionary’. To extract the tissue parameters, a pattern recognition algorithm is used to find the dictionary entry that best represents the acquired signal in each voxel of the image. (Fig. 6).

Fig 6: Flip angle pattern (top left) and velocity encoding pattern (top right) used in the flow MRF sequence (sequence scheme: bottom) (10).

Phantoms for velocity validation measurements

The development of novel MR-based velocity quantification techniques requires validations setups to characterize the accuracy and precision of individual techniques. For such purposes so-called flow-phantoms are being developed and used at PTB that study are characterized by well-known and reproducible measurements. As an example, Figure 7 illustrates a rotational phantom, which is used to validate novel MR acquisition techniques. Here, the ground truth angular velocity and thus the velocity vector field within the phantom is given by an external laser-based measurement.

Furthermore, 4D flow MRI performed in MRI phantoms can be used to validate computational fluid dynamics (CFD) simulations.

Fig. 7: left: image of a rotational phantom, used validation of MR-based velocity measurements, manufactured in a joint project by the German Cancer Research Center. The phantom is powered by compressed air and the angular velocity is measured and controlled by an external, MR-compatible, laser-based setup. Center: 2D rotational velocity vector field measured in transversal orientation. Right: velocity profile through the center of the phantom.

References:

  1. Wiesemann S, Schmitter S, Demir A, Prothmann M, Schwenke, C, Chawla A, von Knobelsdorff-Brenkenhoff, Greiser A, Jin N, Bollache E, Markl M, Schulz-Menger J. Impact of sequence type and field strength (1.5, 3, and 7T) on 4D flow MRI hemodynamic aortic parameters in healthy volunteers. Magnetic resonance in medicine 2021;85:721–733 doi: 10.1002/mrm.28450.
  2. Schmitter S, Schnell S, Ugurbil K, Markl M, Moortele P-F van de. Towards high-resolution 4D flow MRI in the human aorta using kt-GRAPPA and B1+ shimming at 7T. J Magn Reson Imaging 2016;44:486–499 doi: 10.1002/jmri.25164.
  3. Schmidt S, Flassbeck S, Bachert P, Ladd ME, Schmitter S. Velocity encoding and velocity compensation for multi-spoke RF excitation. Magn Reson Imaging 2019;66:69–85 doi: 10.1016/j.mri.2019.11.007.
  4. Schmidt S, Flassbeck S, Schmelter S, Schmeyer E, Ladd ME, Schmitter S. The impact of 4D flow displacement artifacts on wall shear stress estimation. Magnet Reson Med 2021;85:3154–3168 doi: 10.1002/mrm.28641.
  5. Schmitter S, Adriany G, Waks M, Moeller S, Aristova M, Vali, A, Auerbach EJ, Van de Moortele PF, Ugurbil K, Schnell S. Bilateral Multiband 4D Flow MRI of the Carotid Arteries at 7T. Magnet Reson Med 2020;84:1947–1960 doi: 10.1002/mrm.28256.
  6. Ferrazzi G, Bassenge JP, Wink C, Ruh A, Markl M, Moeller S, Metzger GJ, Ittermann B, Schmitter S. Autocalibrated multiband CAIPIRINHA with through-time encoding: Proof of principle and application to cardiac tissue phase mapping. Magnetic Resonance in Medicine 2019;81:1016–1030 doi: 10.1002/mrm.27460.
  7. Ferrazzi G, Bassenge JP, Mayer J, Ruh A, Roujol S, Ittermann B, Schaeffter T, Cordero-Grande L, Schmitter S. Autocalibrated cardiac tissue phase mapping with multiband imaging and k-t acceleration. Magnetic Resonance in Medicine 2020;84:2429–2441 doi: 10.1002/mrm.28288.
  8. Wink C, Ferrazzi G, Bassenge JP, Flassbeck S, Schmidt S, Schaeffter T, Schmitter S. 4D flow imaging with 2D-selective excitation. Magnetic Resonance in Medicine 2019;82:886–900 doi: 10.1002/mrm.27769.
  9. Wink C, Bassenge JP, Ferrazzi G, Schaeffter T, Schmitter S. 4D flow imaging with UNFOLD in a reduced FOV. Magnetic Resonance in Medicine 2019;7:55–338 doi: 10.1002/mrm.28120.
  10. Flassbeck S, Schmidt S, Bachert P, Ladd ME, Schmitter S. Flow MR fingerprinting. Magnetic Resonance in Medicine 2019;81:2536–2550 doi: 10.1002/mrm.27588.