DCE-MRI at high temporal resolution using undersampled radial FLASH: A phantom study
Jost Michael Kollmeier1, Volkert Roeloffs1, and Jens Frahm1

1Biomedizinische NMR Forschungs GmbH, Max-Planck-Institut für biophysikalische Chemie, Göttingen, Germany

Synopsis

We present a DCE-MRI experiment using a commercial perfusion flow phantom for quantitative analysis of image series with high spatial and high temporal resolution (107 ms). Both can be obtained by current real-time MRI methods, i.e. radial undersampled radial FLASH and image reconstruction by nonlinear inversion (NLINV). Contrast agent bolus tracking with high CNR and quantitative parameter maps are presented.

Purpose

Established DCE-MRI protocols present a tradeoff between spatial and temporal resolution especially for bolus tracking of first-pass perfusion procedures. Quantitative analyses require both high temporal resolution for the rapid arterial input functions (AIF) and high spatial resolution for monitoring corresponding tissue responses. These prerequisites are met with recent advances in real-time MRI1. The aim of this work is to develop a suitable real-time MRI sequence for quantitative DCR-MRI using undersampled radial FLASH and image reconstruction by nonlinear inversion (NLINV) and test its performance with use of a commercial perfusion phantom.

Methods

Flow Phantom and Injection Protocol: As a ground truth experiment a commercial multi-modality DCE Perfusion Flow Phantom (Shelley Medical Imaging Technologies, London, Ontario Canada) was used driven by a positive displacement pump to allow for flow of demineralised water. Inside the phantom inflowing liquids are divided into two outputs. The first is the outlet of a perforated distribution tube which leaks into a cylindrical compartment whose output provides the phantom’s response2. Gd-based contrast agent (Gadovist, Bayer HealthCare Pharmaceuticals, Berlin, Germany) is injected via a clinical power injector connected by 1/4'' PVC tubing with 6 m tubing length from the injection site to imaging plane, where a plastic bottle provides a water surrounding for the three tubes (ID 9.5 mm) connected to the phantom. Pump flow rate was set to 300 ml/min and the ratio of output to input flow to 1/2. 10 ml of contrast agent with a concentration of 10 mM of Gd at a rate of 1ml/s was injected. Acquisition and Reconstruction: A radial FLASH sequence (TE/TR 1.58/2.55 ms, resolution 1 x 1 x 8 mm3) acquired two perpendicular sections in an interleaved way, i.e. parallel to the tubing for saturation of the inflowing spins [Fig.1a)] and perpendicular for imaging [Fig.1b)]. To achieve high temporal resolution 21 radial views were acquired and reconstructed using NLINV1 (temporal resolution of 107 ms). However, the coil sensitivity profiles were kept fixed (similar to3) to allow for quantitative analysis of time series. The influence of the flip angle was investigated for 15°, 20°, and 25°. All experiments were performed at 3 T (MAGNETOM Prisma, Siemens Healthcare, Erlangen, Germany) using a 18 channel body array coil. Image Analysis: All images were analyzed using Matlab (Math Works, Massachusetts, USA). The diameter of the three ROIs were chosen to be 80% of the diameter of the inner tube. Quantitative perfusion parameters (Ktrans and kep) were calculated pixelwise for the normalized signal intensity using a least-squares fit and the standard Tofts model4.

Results

Bolus tracking with high CNR was possible for all three acquisition protocols. Figure 2 shows the normalized ROI signal intensities over time for the 25° flip angle measurement. Both, AIF (ROI 1) and Phantom Outlet (ROI 2) show the same enhancement indicating a successful suppression of the inflow effect, because the flow rates in ROI 1and ROI 2/ROI 3 differ by factor of 2 (300 ml/s vs. 150 ml/s). The measured phantom response (ROI 3) was fitted by the standard Tofts model (using ROI 1 as AIF). Values for maximum enhancement (Peak Intensity) and area under curve (AUC) for all data sets are listed in Table 1. All values increase with increasing flip angle. Figure 3 shows the quantitative parameter maps Ktrans and kep obtained by a pixelwise fit.

Discussion

The deviation in peak intensity and AUC across the ROIs decreases with higher flip angles. This has probably two reasons: Firstly, the suppression of the inflow effect is more effective when presaturation with a higher flip angle is performed, and secondly, linearity between contrast agent concentration and signal enhancement for T1-weighted images increases when going to higher flip angles. All phantom responses can be well described by the standard Tofts model, independent of the chosen flip angle. However, the quantitative analysis is affected by both, inflow effects and nonlinearity between concentration and signal enhancement, since all three data sets show an overestimation of the expected Ktrans (0.9 ml/min/g). Future research will focus on a separation of these effects to obtain a full ground truth experiment for DCE first pass perfusion with clinical parameter settings.

Conclusion

We presented a ground truth DCE-MRI experiment for quantitative analysis of image series with high spatial and high temporal resolution as obtained by current real-time MRI methods1. The quantitative analysis for flip angles over 20° revealed good agreement with expected values rendering this setup an excellent candidate for future optimization of protocol and reconstruction parameters.

Acknowledgements

No acknowledgement found.

References

1. Uecker, M., Zhang, S., & Frahm, J. (2010). Nonlinear inverse reconstruction for real-time MRI of the human heart using undersampled radial FLASH. Magnetic Resonance in Medicine, 63(6), 1456-1462.

2. Driscoll, B., Keller, H., & Coolens, C. (2011). Development of a dynamic flow imaging phantom for dynamic contrast-enhanced CT. Medical physics, 38(8), 4866-4880.

3. Wang, X., Roeloffs, V. B., Merboldt, K. D., Voit, D., Schätz, S., & Frahm, J. (2015). Single-shot multi-slice T1 mapping at high spatial resolution–Inversion-recovery FLASH with radial undersampling and iterative reconstruction. Open Medical Imaging Journal, 9, 1-8.

4. Tofts, P. S., Brix, G., Buckley, D. L., Evelhoch, J. L., Henderson, E., Knopp, M. V., ... & Weisskoff, R. M. (1999). Estimating kinetic parameters from dynamic contrast-enhanced T 1-weighted MRI of a diffusable tracer: standardized quantities and symbols. Journal of Magnetic Resonance Imaging, 10(3), 223-232.

Figures

Figure 1: Imaging setup: Water filled bottle with crossing tubes. a) Section of coronal slice for saturation of inflowing spins. b) Section of imaging slice (transversal) showing ROI 1 to 3.

Figure 2: Normalized ROI signal intensity over time for measurement with flip angle 25° and least-squares fit of the standard Tofts model for time series of phantom response.

Table 1: Pixelwise analysis of peak intensity, area under curve (AUC) (Mean \pm SD) and quantitative parameters Ktrans and kep (for ROI 3 only).

Figure 3: Quantitative parameter maps of Ktrans and kep for ROI 3 (phantom response) obtained by pixelwise fit of the standard Tofts model.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1478