As a noninvasive method for measuring the concentration and distribution of chemicals in the living brain, magnetic resonance spectroscopy (MRS) is an important tool for studying brain function and disorders. However, acquiring robust measurement of MRS signals requires a sophisticated study design and the successful development, implementation and maintenance of various MRS techniques. As an active area for research, MRS technology attracts major efforts from top research centers around the world. Clinical magnetic resonance imaging scanners that are optimized for performing structural and functional imaging studies also present daunting obstacles for MRS technical development. Most MRS studies at NIH have been developed and implemented by the MRS Core under protocols 05-M-0144 (NCT00109174) and 11-M-0045 (NCT01266577). Over the last year we have continued to make progress in the following areas: 1) Correction of baseline in short echo time proton MRS. Data acquisition at short echo time is needed to capture the entire metabolic profile of brain visible to proton MRS and to acquire, at the same time, the broad macromolecule baseline. The macromolecule baseline contributes to Cremer Rao Lower Bound of the metabolite signals. To quantify this effect accurately in clinical short echo time MRS, we developed and optimized a method, based on mean squared error of the baseline, for determining the smoothness of baseline. This advance has led to a more objective and reliable determination of metabolite signals (Zhang et al., Magn Reson Med, 72(4):913-22. 2014). 2) Automatic correction of magnetic field inhomogeneity. It is essential to optimize the homogeneity of the magnetic field for all MRS experiments, because field inhomogeneity can easily destroy the critical separation of different chemicals. More importantly, an inhomogeneous field makes it difficult to suppress the tissue water signal effectively, making the reliable detection of more dilute chemicals impossible. This is particularly the case for anatomical regions of interest to psychiatric research. We have implemented and optimized an automatic shimming method that consistently out-performs the automatic shimming methods provided by MRI scanner manufacturers. This method has greatly improved the quality of clinical MRS data acquired at NIH. 3) Multi-slice chemical shift imaging of compounds containing N-acetylaspartate (NAA), creatine, and choline. The MRS core maintains a chemical shift imaging technique for mapping distribution of the neuronal marker N-acetylaspartate at 3 Tesla. This method simultaneously generates images of N-acetylaspartate (a prominent neuronal marker), creatine, and choline-containing compounds. 4) Glutathione detection. Glutathione is a marker for oxidative stress. Many psychiatric and neurological disorders (such as schizophrenia, Alzheimer's disease, and stroke) are associated with abnormal glutathione levels. In collaboration with Steven Warach (NINDS) a glutathione editing method was developed on the Philip 3 Tesla scanner at the Suburban Hospital for studying stroke patients using adaptive line-fitting. We have also developed a single-shot method for measuring glutathione at 7 Tesla (Lally et al., J Magn Reson Imaging 2016, 43(1):88-98). 5) Carbon-13 MRS. Using carbon-13 labeled glucose or the glial-specific substrate acetate, it is possible to measure brain energetics and glutamate and glutamine cycling flux. Previously we invented a method for carbon-13 MRS by combining low power stochastic decoupling and intravenous infusion of glucose with a carbon-13 label at the C2 position. This strategy makes it possible to perform viable carbon-13 MRS on single channel clinical MRI scanners. Using this strategy, we have acquired high quality carbon-13 MRS data from both the occipital and frontal lobes of healthy participants and showed that it is feasible to detect, simultaneously, two labeling pathways in the human brain. More recently, we succeeded in implementing and optimizing this strategy on the 7 Tesla scanner with enhanced sensitivity and spectral resolution (Li et al, Magn Reson Med. 2016, 75(3):954-61), and we were the first to detect GABA turnover in the human brain. Using our 7 Tesla carbon-13 MRS method we have demonstrated, for the first time, that specific enzyme activities can be measured noninvasively in the human brain following administration of carbon-13 labeled glucose (Li et al, Sci Rep, 8:2328:1-8 (2018)). 6) Proton glutamate editing. Previously we implemented a single-voxel glutamate editing method with correction of eddy current effects for measuring glutamate concentration at 3 Tesla. A method for simultaneously extracting both glutamate and glutamine from multi-echo MRS data has also been developed and implemented (Zhang et al, Magn Reson Med. 2016, 76(3):725-32). We have verified that, unlike the popular short-TE MRS, our multi-echo MRS has high resistance to errors originated from noise and linewidth variations across individuals because of its relatively flat baseline. Recently, we developed and optimized a new glutamate editing method that needs a single echo time to isolate the glutamate H4 signal at 7 Tesla. 7) GABA and beta-hydroxybutyrate editing. Previously we implemented and refined a method for measuring GABA. As with N-acetylaspartate imaging, participants' movements can lead to difficulty in the accurate determination of GABA. We used a navigator strategy, based on residual water, to track and correct for participants' movement. We also developed special data processing software to correct for phase changes because of participants' motion. Improvements to these corrections have now been quantified and successfully applied in studies of human participants. Currently, we are quantifyingWe have quantified glutamate/glutamine concentration using GABA-edited spectral data at 3 Tesla. For measurement of GABA at 7 Tesla we have shown that combining spectral editing with shorter echo time (56 ms) and taking advantage of strong coupling effect GABA, together with glutamate, glutamine and glutathione can be measured at 7 Tesla with high precision (An et al, Magn Reson Med 2018, 80:1776-1786 doi: 10.1002/mrm.27172.). More recently, we have developed an editing technique for detecting beta-hydroxybutyrate and applied it to studying ketogenic diet treatment. 8) NAAG editing. We developed MRS methods for measuring the dipeptide neurotransmitter N-acetylaspartylglutamate (NAAG), which plays an important role in glutamate signaling. Our methods use regularized line-shape deconvolution, based on the L-curve method and Wiener filtering, to measure NAAG reliably (An et al, Magn Reson Med 2014, 72(4):903-912). 9) 3 and 7 Tesla Phosphorus MRS imaging. We have developed a chemical shift imaging method for measuring phosphorus-containing chemicals in brain at 7 Tesla. The method has been successfully tested in both occipital lobe and. Development and testing for frontal lobe studies are currently in progress. We are also developing data fitting strategies to extract NAD+ and NADH from phosphorus and proton MRS spectra. Brain phosphorus MRS at 3 Tesla has also been developed and applied to clinical studies. 10) Highly accelerated full density simulation of multispin systems for spatially localized MRS. Prior to our work the state of the art quantum mechanical simulation of multispin systems for generating basis function for spectral fitting may require prohibitively long computation time. We recently invented a vastly accelerated one dimensional simulation method that that reduces computation time to minutes. As a result, highly accurate spatial digitization of MRS voxel can now be routinely used. Our method greatly improves of clinical proton MRS data and have already been adopted by many major MRS groups around the world.
Showing the most recent 10 out of 36 publications
