Bart Van De Bank

MR Imaging or Spectroscopy cannot be performed without a RF coil. Building such a coil can be challenging, especially for MR systems with magnetic field strengths of 7 Tesla or higher, which are delivered without a body coil. Currently, this field of hardware engineering is a paradise for the MR Coil technician, as there is need to develop and construct innovative experimental coil designs for MR experiments. For multi-nuclear MR spectroscopy there are few commercial coils available and an engineer is needed to design a dedicated coil for the appropriate resonance frequency and body organ. In chapter 4 we presented one of these special coil designs; a dedicated coil setup for an ultra-high field (UHF) strength of 7 Tesla that featured multi-transmit capabilities on 1H channels, volume excitation for PNP and an additional PNP receive-only array with a 7-fold local SNR improvement close to the receive elements. This coil contains three main components: an 8-channel multi-transmit 1H volume (head) coil, an insertable and actively tunable birdcage coil for PNP and a local 7-channel PNP receive-only array. The increased PNP sensitivity in combination with the limited field of view of the local receive coils made it possible to perform PNP MRSI at higher spatial resolutions (3.0cm3 voxel) in the occipital lobe (the visual processing center) of the human brain in clinically acceptable scan times (~15 min). This setup was used for the research presented in chapters 6 and 7.

For the study presented in chapter 5, we assessed variability and reproducibility of a 1H spectroscopy measurement, which has a limited chemical shift displacement error. We implemented an identical single voxel semi-LASER 1H-MRS pulse sequence on four different 7T systems. These systems were manufactured by two different vendors, Philips and Siemens. We assessed metabolic variations and reproducibility in 7 healthy volunteers by obtaining high quality, short echo time, 1H MR spectra from two clinically relevant brain regions, the posterior cingulate cortex (gray matter) and the corona radiata (white matter). Data processing and analysis of all MR systems was harmonized. As such, this work initiates standardization efforts for MR spectroscopy at UHF. This is a critical step towards a wider utility and higher impact of the methodology for neuroscience and clinical applications. In this study we have shown that harmonization of data acquisition and post-processing of single voxel 1H-MRS produces very similar results at four different 7 Tesla systems. The accuracy and reproducibility obtained with the semi-LASER sequence in this multi-center, multi-vendor setting at ultra-high magnetic field strength of in-vivo measured neurochemical profiles can be used as guidance for quantifying metabolite concentrations in future (clinical) studies.

‘To NOE or not to NOE’, is the question we addressed in chapter 6. Signal enhancement of PNP signals can at first sight be seen as a promising tool to either shorten scan times, in some cases to increase spatial resolution, or just to increase signal intensity. Unfortunately, it was not known if this signal enhancement was beneficial for all PNP nuclei at ultra-high field strengths and if there was no additional variability introduced to the acquired data. To answer this question, we looked at the variations in data between non NOE-enhanced (native) MRSI data and NOE-enhanced data. To a great extent, this question can be answered with ‘yes’. The signal enhancement generated by NOE improved the relative repeatability of brain PNP MRSI in healthy volunteers. Variations in NOE enhancements per metabolite could be explained almost completely by the repeatability of native and NOE-enhanced PNP MRSI. For these reasons, the use of NOE-enhanced PNP MRSI is encouraged.

Knowing that NOE-enhancement would do no harm, and given the fact that we have an excellent RF coil, we demonstrated the feasibility of functional PNP-MRSI (fMRSI) at 7T in chapter 7. The role of the high-energy buffer present in the human visual cortex has been explored before. However, we tried to explore this in more detail by using the optimized coil setup for localized spectroscopic imaging as presented in chapter 4, combined with local NOE-enhancement (chapter 6) we were able to increase signal intensity of the PNP nuclei. This enabled us to do repeated localized 3D MRSI with sufficient quality in a limited scan time, enabling localized spectroscopy during visual stimulation and during rest. With this very sensitive measurement, we showed that the high energy phosphates like PCr, Pi and ATP hardly change during a visual stimulation protocol, in contrast to previously reported studies. This shows that it is very challenging, or even impossible to drain the PCr-buffer and questions its role as a long-term energy buffer in the brain. Although we examined the high energy phosphate levels at ultra-high magnetic field strength combined with a dedicated coil and a very sensitive coil setup, we couldn’t replicate previously reported changes during visual stimulation. This gave us the impression that the high energy buffer is not acting as a main energy source on our applied time-scales of stimulation inside the human visual cortex.

Future Directions
This thesis describes studies on the accuracy and stability of measuring neurochemical profiles and high energy phosphate levels in the human brain at 7T. Measurement stability and accuracy are topics for ongoing debate among MR spectroscopists worldwide. Although MR measurements of the tissue concentrations of metabolites of a particular brain region are expected to produce similar numbers, within certain ranges, it is common that different post-processing steps including quantitation may produce different results. Therefore I would propose to investigate the stability and influence of different quantitation algorithms, including differences in prior knowledge, on the same spectroscopic datasets. This would confirm the metabolic insights of spectroscopy. Or - if this is not applicable - at best, point out the hurdles for quantification with spectroscopy.

Another possible further direction based on this thesis is to investigate feasibility of multi-nuclear excitation of the region of interest only. We presented a possibility to combine B1-shimming and NOE-enhanced PNP MRSI with a multichannel receiver. Using this setup it should be possible, as the next step, to combine multichannel transmit, or transmit SENSE, to selectively excite a geometric region of interest for instance through polarization transfer experiments. This would open up new possibilities for even higher resolution PNP-MRSI, or for more accurate localization of internal structures.

Combining the reproducibility measurements with the dedicated multi-channel, multi-nuclear coil setup, opened up the possibility to measure brain energy metabolism during longer visual exposures. Although we could not detect any alterations in high-energy phosphates during these stimulation paradigms, it could still be the case that the high energy buffer is needed on initiation of activation, just for one or a couple of seconds and is refreshed very quickly, faster than detectable. Another possible explanation might be that the high-energy reservoir inside the human visual cortex is barely addressed at all in the case of visual stimulation. Consensus about this issue has not been presented in literature yet. The tools and technical opportunities in this thesis enable further mechanistic studies in this direction.