Magnetic Resonance Imaging at ultra high fields requires novel techniques, both concerning hardware and imaging sequences. At field strengths of 9.4 T (human) and 14.1 T (rodents), I investigate the capabilities of ultra high field MRI. Specifically, our aim is to

• quantify the MR parameters at ultra high field strength,
• find solutions to some of the major challenges involved in ultra high field MRI, like B1-homogeneity or safety,
• develop techniques that take full advantage of the potential of the high field strength and assess the gain compared to conventional MR instruments
• help in appying those techniques to medical or biological applications
• help in combining ultra-high field MRI with other techniques to gain additional information in neuroscientific studies.

In addition, I assist other groups by developing or optimizing specialized techniques for their applications.

For a comprehensive overview over our projects, please see our project group web page.

Evaluation of Signal Gain and MR tissue parameters at ultra-high field

While the number of ultra-high field MR scanners is still growing rapidly, mainly due to the desired SNR gain with increasing fields, surprisingly little is known about the actual SNR that can be obtained with higher fields. In this project we compared the SNR and tissue parameters at 3 T, 7 T and 9.4 T. The same subjects were scanned with the same gradient echo sequence at all three fields, and the SNR of the resulting maps were corrected for the differences in relaxation times and transmit field. Since it is not easily possible to obtain reliable values for the sensitivity of the receive coils, arrays with similar number of elements and geometry were used. The SNR was found to increase supralinearily with the field strength, following a relation of SNR ≈ B_0^1.65.

SNR (top) and relative SNR differences (bottom) at 3 T, 7 T and 9.4 T. Data are from the same subject at all fields. SNR increases supralinearily, growing by 3.3 from 3 T to 7 T and another factor of 1.67 from 7 T to 9.4 T.

Ultra-high Resolution Imaging with Acquisition Weighting

High spatial resolutions are one of the main goals of ultra-high field imaging. In this study we demonstrated the possibility to obtain images with voxel sizes down to 14 nl with sufficient SNR in an acceptable scan time by combining the ultra-high field with a highly sensitive 31 channel array coil and a special, SNR optimized imaging technique. By performing a k-space weighting during acquisition, it can be shown that this method not only avoids ringing artifacts, but also increases SNR by up to 36% without losses in spatial resolution or scan time (publication).

High resolution images (0.2 × 0.2 × 0.5) mm3 acquired with a conventional (top) and the acquisition weighted sequence (bottom). The weighted images show greater fine detail and a higher SNR.

High sensitivity perfusion imaging with Arterial Spin Labeling

Arterial Spin Labeling is a promising technique for quantitative perfusion imaging without the need for contrast agents. Since ASL generally suffers from a low signal-to-noise ratio, we concentrate on implementing and developing techniques with high sensitivity. To reach this goal, we follow several different approaches:

• In contrast to the most commonly used techniques of pulsed arterial spin labeling, continuous arterial spin labeling has a considerably increased sensitivity, but also a higher complexity. We have implemented a number of different variants of continuous ASL, which are now used in neuroscientific and biological applications (publication).

  

  Perfusion images acquired with different CASL and a PASL technique.

• In contrast to the cortex in the brain, which has a very high perfusion, the white matter or, even more, the skeletal muscle are only poorly perfused, making highly quantitative measurement difficult. For making it possible to obtain quantitative values of the perfusion from those tissues, a highly sensitive technique based on single-voxel spectroscopy was developed. This FAIRPRESS techniques has been successfully used to quantify the perfusion in the white matter as well as in muscels in the legs of rats and humans.
• In addition to the gain in intrinisc signal-to-noise ration at ultra-high field, the longer longitudinal relaxation time will help to further improve the accuracy of perfusion images. Using the most sensitive CASL sequences is, however, hampered by the high SAR with is necessary to label the inflowing blood. Using a dual-coil CASL technique, that applies the labelling field with separate, local coils at the neck of the volunteer, it will be possible to reduce SAR far enough to be able to take full advantage of the improved signal at higher field.

Multiband fMRI in animal studies

High temporal resolution is a crucial factor in many fMRI studies. Multiband imaging is a novel technique to increase the speed of fMRI acquisition by using parallel imaging methods to obtain the signals from two or more slices simultaneously. While this technique is already used frequently in human fMRI, it has so far not been applied in animal studies. We have implemented multiband EPI on an animal scanner and evaluated its performance with different fMRI protocols.

fMRI data from a rat forepaw stimulation experiment at 7 T, acquired with a conventional sequence (left) and a dual-band EPI with double temporal resolution.

Measurement of MR imaging parameters at ultra high field

When increasing the field strength, MR relaxation times, magnetization transfer effects, and susceptibility-induced field variations will change. We have measured and quantified those parameters at field strengths up to 16.4 T and analysed the effects of those changes on SNR and contrast of the resulting images (publication).

The relaxation times T₁, T₂ and T₂^* were measured with high accuracy and spatial resolution using inversion recovery, single spin echo and gradient echo sequences, respectively, with varying delays. The MTR was measured by off-resonance irradiation at different frequencies and power levels and the resulting values were used to determine the parameters of a two-pool model [2] to allow for comparison to other field strengths. Local frequency variations were deduced from phase changes in gradient echo images with varying echo time.

With values between 1834 ms (white matter ) and 2376 ms (hippocampus), T₁ was significantly increased compared to lower fields, while T₂ and T₂^* are relatively low (corpus callosum: 20 ms / 13.6 ms, cortex: 24 ms / 21 ms). The MTR varies between 51% and 61% and thus is considerably stronger than at lower field. Image phase shows distinct differences between different anatomical structures and can be a valuable contrast mechanism at high field. In all parameter maps, all major anatomical structures were clearly visible.

Comparisons to publications at lower field show an increase in contrast-to-noise ratio with field strength for all contrast mechanisms. Especially phase and T₂^* imaging have great potential for use in neuroscientific and preclinical applications.

Flip angle mapping

Highly accurate, fast, and simple mapping of the excitation field is a crucial requirement for ultra high field MR imaging. In an extensive study, we are comparing the accuracy and precision of the most popular flip angle mapping techiques, both theoretically and in experiments under different settings. The results will be used to further improve the performance of these sequences to enable highly accurate shaping of the transmit field using B1-shimming or Transmit SENSE (publication).

For information about the other projects in our group, please see  our project group web page.