Quantitative Proton MR Spectroscopic Imaging of Normal Human Cerebellum and Brain Stem

FULL PAPERS Magnetic Resonance in Medicine 46:699 705 (2001) Quantitative Proton MR Spectroscopic Imaging of Normal Human Cerebellum and Brain Stem Michael A. Jacobs, 1 Alena Horská, 1 Peter C.M. van Zijl, 1,2 and Peter B. Barker 1,2 * Quantitative, multislice proton MR spectroscopic imaging (MRSI) was used to investigate regional metabolite levels and ratios in the normal adult human posterior fossa. Six normal volunteers (36 3 years, five male, one female) were scanned on a 1.5 T scanner using multislice MRSI at long echo time (TE 280 msec). The entire cerebellum was covered using three oblique-axial slice locations, which also included the pons, midbrain, insular cortex, and parieto-occipital lobe. Concentrations of N-acetylaspartate (NAA), choline (Cho), and creatine (Cr) were estimated using the phantom replacement technique. Regional variations of the concentrations were assessed using ANOVA (P < 0.05). High-resolution MRSI data was obtained in all subjects and brain regions examined. Metabolite concentrations (mm) (mean SD) were as follows: cerebellar vermis: 2.3 0.4, 8.8 1.7 and 7.6 1.0 for Cho, Cr, and NAA respectively; cerebellar hemisphere: 2.2 0.6, 8.9 2.1, 7.5 0.8; pons 2.2 0.5, 4.3 1.1, 8.3 0.9; insular cortex, 1.8 0.5, 7.8 2, 8.0 1.1, parieto-occipital gray matter, 1.3 0.3, 5.7 1.1, 7.2 0.9, and occipital white matter, 1.4 0.3, 5.3 1.3, 7.5 0.8. Consistent with previous reports, significantly higher levels of Cr were found in the cerebellum compared to parietooccipital gray and occipital white matter, and pons (P < 0.0001). NAA was essentially uniformly distributed within the regions chosen for analysis, with the highest level in the pons (P < 0.04). Cho was significantly higher in the cerebellum and pons than parieto-occipital gray and occipital white matter (P < 0.002) and was also higher in the pons than in the insular cortex (P < 0.05). Quantitative multislice MRSI of the posterior fossa is feasible and significant regional differences in metabolite concentrations were found. Magn Reson Med 46:699 705, 2001. 2001 Wiley-Liss, Inc. Key words: cerebellum; pons; proton magnetic resonance spectroscopic imaging Proton magnetic resonance spectroscopy (MRS) and spectroscopic imaging (MRSI) of the human brain are becoming increasingly used for both research and clinical applications (1). To date, the majority of spectroscopic studies of both normal and pathological brain have focused mainly on supratentorial brain regions, although there have been some reports of MRS and MRSI in human cerebellum and pons (2 8). Generally, infratentorial metabolite levels have been less often studied because of potential technical difficulties (e.g., field inhomogeneity) and the relatively 1 The Russell H. Morgan Department of Radiology and Radiological Science, The Johns Hopkins University School of Medicine, Baltimore, Maryland. 2 F.M. Kirby Research Center for Functional Brain Imaging, Kennedy Krieger Institute, Baltimore, Maryland. Grant sponsor: NIH; Grant number: T32 CA09630. *Correspondence to: Peter B. Barker, Department of Radiology, MRI 143C, Johns Hopkins University School of Medicine, 600 N. Wolfe Street, Baltimore, MD 21287. E-mail: barker@mri.jhu.edu Received 22 December 2000; revised 16 April 2001; accepted 17 April 2001. 2001 Wiley-Liss, Inc. 699 lower frequency of disorders of the brain stem and cerebellum. While white and gray matter seem to have characteristic metabolic profiles at the level of the lateral ventricles and above (4,5,9 12), significant regional variations have been demonstrated in structures such as the thalamus, insular cortex (13), and temporal lobe (14). In this article, we describe the use of infratentorial multislice proton MRSI for the estimation of metabolite concentrations in normal adult cerebellum and brain stem. MATERIALS AND METHODS Subjects Six normal volunteers (age range 31 40 years, five male, one female, 36 3 years (mean SD)) with no known neurological disorders and normal brain MRI were scanned in a 1.5 T Philips Gyroscan ACS NT Powertrak 6000 (Netherlands) using a combined MRI and spectroscopic imaging protocol. The Philips transmit/receive quadrature birdcage head coil was used for all scans. The protocol was approved by the Johns Hopkins Institutional Joint Committee on Clinical Investigation and informed consent was obtained from all subjects. MR Protocol The MR protocol consists of a routine brain MRI examination (sagittal and axial T 1 -weighted images, oblique axial fast spin-echo, and FLAIR images) and MRSI. The routine MR images were used to identify anatomical structures and confirm the absence of any structural or signal abnormalities. MRSI was performed using a spin-echo sequence with 2D phase-encoding and outer-volume saturation pulses for lipid suppression (15). Three 15 mm thick slices were recorded from the base of the cerebellum to the basal ganglia. The slices were angulated so that the most inferior slice covered the base of the cerebellum and inferior pons, while the most superior slice included the tip of the cerebellar vermis and the midbrain (Figs. 1, 2). The MRSI parameters were TR/TE 1900/280 msec, 28 28 matrix, 24 cm FOV using 28 28 circular phase-encoding scheme (16) and the total data acquisition time was 20 min (one signal average per phase-encode step). The nominal voxel size was 1.5 0.9 0.9 cm (approx. 1.1 cm 3 ), while the estimated actual voxel size (calculated from the integration of the 2D point spread function) was approximately 2.8 cm 3 with a full width at half maximum of 1.7 cm. The echo signal was digitized with 256 data points and a spectral width of 1500 Hz. Water suppression was accomplished with a single CHESS pulse and a bandwidth of

700 Jacobs et al. FIG. 1. a: Slice locations for proton MRSI of the cerebellum and brain stem, overlaid on midsagittal T 1 -weighted localizer image. The locations of the sagittal outer-volume saturation pulses are indicated in gray. b: Maps of the B 0 field homogeneity generated from the double gradient-echo scans for the three slice locations used for MRSI. 150 Hz, applied 40 Hz downfield from the water resonance. Extracranial lipid signals were attenuated by using eight outer-volume saturation (OVS) pulses, arranged in a conical, octangular pattern to match the contours of the skull. Compared to Ref. 15, the OVS cone was inverted (i.e., with the sat bands closest together in the most inferior slice location) to best match the shape of the skull in the posterior fossa region. T 1 -weighted MR images were recorded at the same slice locations as the MRSI dataset for anatomical correlation. Prior to MRSI, shimming was performed to optimize field homogeneity and water suppression was optimized using automated routines provided by the manufacturer. After the MRSI acquisition, a final set of double gradient-echo images were recorded at the same slice locations in order to calculate maps of the B 0 magnetic field strength and to ensure that the subject had not moved during the MRSI scan. MRSI Data Analysis Spectroscopic imaging data were reconstructed using inhouse software on a Sun Enterprise 5500 computer system (Sun Microsystems, Mountain View, CA). Multislice 2D MRSI datasets were processed by 3D Fourier transformation, with cosine filters in the spatial (phase-encoding) domains after zero-filling to 32 32 matrix size and exponential line broadening of 3 Hz, zero-filling to 2048 data points, and a high-pass convolution filter to remove the residual water signal (50 Hz stop-band) in the time-domain. After magnitude calculation, a susceptibility correction (i.e., left- or right-shifting of each spectrum in the frequency domain on a voxel-by-voxel basis) was applied. Baseline correction was performed using a cubic spline routine. After setting the chemical shift of N-acetyl aspartate (NAA) to 2.02 ppm, spectroscopic images (Fig. 2) were created by numerical integration over the following frequency ranges: choline (Cho, 3.34 3.14 ppm), creatine (Cr, 3.14 2.94 ppm), NAA (2.22 1.82 ppm), and lactate (Lac, 1.55 1.15 ppm). For display, metabolic images are linearly interpolated to 256 256 points. In the second slice, regions of interest (ROIs) were selected for quantitative analysis from the cerebellar vermis, white and gray matter of both cerebellar hemispheres, pons (Fig. 3). In the third slice, ROIs were chosen in the insular cortex and parieto-occipital gray and occipital white matter (Fig. 4). For each ROI, peak areas were determined by fitting the major resonances of the spectrum (Cho, Cr, NAA) to a Gaussian function using a simplex curve-fitting routine (12). Metabolite concentrations were estimated using the phantom replacement methodology as described previously (12). Briefly, the MRSI procedure was performed on a 4-L sample containing 73 mm NAA. In vivo metabolite concentrations were estimated from the ratio of the in vivo and phantom signals, with corrections for differences in T 1

Proton MRSI of the Cerebellum 701 FIG. 2. T 1 -weighted and metabolic images (Cho, Cr, and NAA) from the three slice locations indicated in Fig. 1. Particularly apparent are the high levels of Cho and Cr in slice 2, corresponding to the cerebellar hemispheres and vermis. and T 2 relaxation times (using the equation S S 0 * exp(-te/t 2 )*(1 exp(-tr/t 1 ))), RF coil loading, S corr S uncorr *10 A/20 where A is the difference in RF power (measured in db) for a 90 pulse between the phantom and the brain and receiver gain. Metabolite concentrations (in mm) are calculated from the ratio of the corrected metabolite and phantom signals: [M] [P]*S 0 (M)/S 0 (P). Metabolite relaxation times were taken from the literature (17); the values used for T 1 for NAA, Cr, and Cho were 1.41, 1.50, and 1.32 sec, respectively, and for T 2 350, 220, and 350 ms, respectively. These values represent average results of multiple measurements from different groups (17). No corrections were applied for partial volume effects or RF coil inhomogeneities. Experiments performed on a uniform phantom indicated a maximum signal change of 6% over the volume of the posterior fossa using a spinecho pulse sequence and the spectroscopy head coil. Statistical Analysis Descriptive statistics consisted of means and standard deviations for estimated metabolite concentrations and peak area ratios for each ROI. Regional variations of the metabolite concentrations and ratios were assessed using ANOVA. Statistical significance was assigned for P 0.05. RESULTS Good quality data from all six subjects were obtained, with the most caudal slice exhibiting artifacts from lipid suppression or increased linewidths due to field inhomogeneity in one subject only. Representative field homogeneity maps for all three slices are shown in Fig. 1. Figure 2 shows metabolic images for each slice location and spectra from representative ROIs are shown in Figs. 3 and 4. Table 1 contains the results of the quantitative analysis, while Table 2 provides metabolite concentration values selected from the literature for comparison. The largest differences in regional metabolite concentrations were in Cr levels. Cr was highest in the cerebellar vermis and hemispheres, next highest in insular cortex, lower in parieto-occipital gray and white matter, and lowest in the pons. Cerebellar Cr concentrations were significantly higher than parieto-occipital gray and white matter and pons (P 0.0001). In addition, pontine Cr levels were significantly lower than the insular cortex (P 0.0003), parieto-occipital gray (P 0.02) and white matter (P 0.003). Regional variations in Cho also existed, with highest Cho in the cerebellar vermis and hemispheres, next highest in pons, then insular cortex, and lowest in the parietooccipital gray and white matter. Statistically significant differences were found between the parieto-occipital regions and the pons and cerebellum (P 0.002). In contrast, no significant regional variations in NAA could be determined, except for slightly higher NAA in the pons than parieto-occipital gray matter (P 0.04). No significant differences (P 0.05) were found in any metabolite concentration between cerebellar vermis and hemispheres. Likewise, no significant differences (P 0.05) were found between parieto-occipital gray and white matter. DISCUSSION Quantitative multislice MRSI of the posterior fossa has not been previously reported, although prior studies have

702 Jacobs et al. FIG. 3. Representative spectra from regions of interest selected in slice 2; anterior and posterior cerebellar hemispheres, vermis, and pons. High levels of Cho and Cr are present in all regions except the pons, which has low levels of Cr. NAA is slightly higher in the pons than other regions. Field homogeneity is sufficient and there is minimal lipid contamination in each region. No left right asymmetries are present. been performed using either quantitative single voxel MRS (4,5,18,19) or nonquantitative MRSI (2,15). The estimated metabolite concentrations reported in this study are, in general, in good agreement with prior literature values obtained using short TE single-voxel spectroscopy. For instance, Michaelis et al. (5) reported cerebellar levels of Cho, Cr, and NAA to be 2.5, 9.1, and 9.6 mm, while for the pons the concentrations were 2.9, 6.0, and 12.1 mm (Table 2), similar to the numbers reported here (Table 1) with the exception of higher NAA in the pons. The lower pontine NAA concentrations in our long TE MRSI study could be the result of several differences between the current study and previous single voxel studies at short TE; in addition to differences in voxel sizes, at short echo time, signals from a number of compounds (glutamate, glutamine, GABA, and macromolecules) may overlap with NAA and (if not accounted for) increase its apparent concentration. Concentrations similar to the current study have also been reported for the cerebellar vermis by Pouwels and Frahm (4) (2.2, 9.0, and 8.7 mm for Cho, Cr, and NAA, respectively) and Hennig et al. (18) (2.2, 8.1, and 8.7 mm). Creatine has also previously been shown to be higher in the cerebellum than in other brain regions using nonquantitative multislice MRSI (2,15). Although we failed to detect a significant difference in Cr concentrations between parieto-occipital gray and white matter, as demonstrated previously (11), we did observe a trend for Cr to be higher in the parieto-occipital gray matter (P 0.2, Table 1). We attribute the lack of statistical significance to spectral overlap between Cr and Cho and significant partial volume effects of gray and white matter in the voxels selected for analysis. The general concordance between our measurements and single-voxel metabolite concentrations in the literature suggests that the phantom replacement quantitation technique performed satisfactorily in the posterior fossa. However, there are certain limitations to the technique as applied here; first, since MRSI was performed at long echo time (280 msec), the signals are very sensitive to T 2 -decay. Because time constraints of the examination did not permit determination of regional metabolite T 2 values, we applied corrections for T 2 signal decay based on literature values (17), which were mostly determined in supratentorial brain regions. The reasonable concordance between

Proton MRSI of the Cerebellum 703 FIG. 4. Representative spectra from ROIs selected in slice 3; insular cortex, parieto-occipital gray matter, and occipital white matter. Spectra from the insular cortex exhibit higher levels of Cho than the parieto-occipital gray and occipital white matter regions. our concentration estimates and those in the literature may suggest that metabolite relaxation times in the posterior fossa are not much different from those of other regions of the brain. Second, we applied no correction factors for inhomogeneity of the B 1 field. However, for all subjects in the current study it was possible to place the entire head well within the sensitive volume of the RF coil, and so B 1 inhomogeneity was likely to be only a minor correction factor, given the high homogeneity of the commercial imaging coil used. B 1 inhomogeneity may be more significant for patients (e.g., obese, and/or with short necks) who are difficult to position in the center of the RF coil. In principle, the phantom replacement method may also be extended to correct for B 1 inhomogeneity, by comparing similar voxel locations (relative to the position of the RF Table 1 Estimated Metabolite Concentrations (mm) and Peak Area Ratios for Selected Regions-of-Interest (Mean SD) [Cho] [Cr] [NAA] Cerebellar vermis 2.3 0.4 8.8 1.7 7.6 1.0 Cerebellar hemispheres 2.2 0.6 8.9 2.1 7.5 0.8 Pons 2.2 0.5 4.3 1.1 8.3 0.9 Insular cortex 1.8 0.5 7.8 2.0 7.9 1.1 Occipital gray matter 1.3 0.3 5.7 1.1 7.2 0.9 Occipital white matter 1.4 0.3 5.3 1.3 7.5 0.8 coil) between the phantom and in vivo. However, because of the high B 1 homogeneity of the RF coil over the volume of the posterior fossa, this was not deemed to be necessary in the current study. Beside quantitation, there are a number of other technical issues regarding multislice MRSI of the posterior fossa. First, compared to supratentorial MRSI (15), if OVS pulses are used the angulation of the OVS cone is inverted (Fig. 1) to provide optimal fat suppression and it may not be possible to simultaneously provide suppression of the fat in the clivus without suppressing some of the brain signal from the inferior frontal lobe (Fig. 1). Second, B 0 magnetic field homogeneity in the posterior fossa has been predicted and measured to be less worse than in other parts of the brain, particularly the occipital and parietal lobes (20). Magnetic susceptibility-associated field inhomogeneity is caused by the air tissue interfaces both internal and external to the cranium and the cerebellum and brain stem experience a shift to a lower field because of their spatial relationship to the nasal cavity and pharynx, with a particular shift in the region of the posterior cerebellum because of the curvature of the nape (20). Consistent with these observations, we obtained lower-resolution spectra than those attainable with supratentorial MRSI, but spectra (e.g., Fig. 3) were of sufficiently high resolution to be interpretable and only occasionally was field inhomogeneity a problem in the most caudal slice. The lower spectral

704 Jacobs et al. Table 2 Literature Values of Metabolite Concentrations for Selected Regions-of-Interest (mm, Mean SD) Reference Location [Cho] [Cr] [NAA] (5) Cerebellar hemispheres 2.5 0.3 9.1 1.2 9.6 1.6 (4) Cerebellar hemispheres 2.2 0.3 8.7 1.3 7.4 1.0 (18) Cerebellar hemispheres 2.2 1.2 8.1 3.3 8.7 2.1 (29) Cerebellar hemispheres 2.1 0.3 8.3 0.9 8.0 0.9 (29) Cerebellar vermis 1.8 0.2 6.8 0.7 7.3 1.3 (5) Pons 2.9 0.7 6.0 0.3 12.5 0.8 (4) Insular cortex 1.3 0.2 7.0 0.6 8.7 0.8 (4) Occipital gray matter 0.9 0.1 6.9 0.7 9.2 0.9 (18) Occipital gray matter 1.3 0.4 5.8 2.4 10.4 2.7 (5) Occipital gray matter 1.4 0.3 8.2 1.4 11.7 2.2 (29) Occipital gray matter 1.2 0.1 6.5 0.1 8.9 0.8 (4) Occipital white matter 1.6 0.2 5.5 0.8 7.8 0.9 (5) Occipital white matter 1.8 0.3 6.1 0.8 8.8 1.0 (18) Occipital white matter 1.3 0.5 5.3 2.5 8.2 2.2 (29) Occipital white matter 1.6 0.2 4.9 0.6 6.1 0.7 resolution may explain the fairly large standard deviations in the reported metabolite concentrations. We only had linear shim coils available to us for shimming and it is possible that spectral resolution in multislice MRSI of the posterior fossa could be improved in the future by the use of higher-order shim corrections. The largest metabolic difference between the cerebellum and other brain regions is the high levels of Cr both in the vermis and the cerebellar hemispheres. The origin of the increased Cr signal is unclear, but the most likely explanation appears to be the different cellular composition of cerebellar cortex compared to that of the neocortex. Whereas the neocortex has a characteristic six-layer structure, the cerebellar cortex has a uniform three-layer structure, consisting of a superficial molecular layer containing mainly axons and dendrites of the cerebellar neurons, a Purkinje cell layer, and a granular layer consisting of a multitude of densely packed small granule cells. In one preparation, isolated cerebellar granular neurons (from 7-day-old rats) had lower levels of Cr than cultured astrocytes or oligodendrocytes (21), so the high levels of Cr in the cerebellum may originate either from the Purkinje cells or the molecular cortical layer. Since the peak commonly referred to as creatine in the proton spectrum represents the sum of both creatine and phosphocreatine, compounds which are involved with energy metabolism in the brain, the high levels of Cr may reflect increased energy demand of Purkinje cells. In this regard, it is interesting to note that the cerebellum has high levels of the enzyme creatine kinase (CK) (22,23), which catalyzes the reversible exchange of a phosphoryl group between Cr and ATP. The CK-PCr system serves as a temporal buffer for ATP homeostasis when rates of synthesis are transiently exceeded by rates of consumption. In the cytosol, there are two distinct isoforms of CK a brain type (CKB) and a muscle type (CKM). It has been reported that the cerebellum (in particular, Purkinje neurons (24)) is the only part of the brain which contains the CKM isoenzyme and that high levels of CKB and mitochondrial CK were also found in the granular layer; Bergmann glial cells of the molecular layer contained CKB only (25). In addition, recent studies using 31 P spectroscopic techniques have indicated a much higher PCr/ATP ratio in the cerebellum along with a trend towards a moderately increased (5 10%) level of PCr compared to the cerebrum (26,27). Since PCr may act as an energy buffer, higher PCr (and Cr) levels in the cerebellum suggests that these metabolites are present for short-term, on-demand energy supply in response to cerebellar activation. However, further studies are needed to clarify the biological implications of the increased concentrations of these metabolites in the cerebellum. The other main significant finding of the current study was the observation of high levels of NAA and Cho and low levels of Cr in the pons. The pons is characterized by a high density of fiber bundles (i.e., white matter) and the metabolite profile observed in the pons may be regarded as consistent with the same pattern generally observed in white matter (compared to gray) in supratentorial brain regions (4,8,11). The high NAA signal presumably reflects the high axonal/neuronal density of the pons (28). In conclusion, quantitative multislice proton MRSI of the human posterior fossa is feasible. Significant regional variations in creatine and choline levels exist between the cerebellum, pons, and supratentorial brain regions. When considering pathological processes involving the posterior fossa using proton MRSI, it is important that these normal regional variations are recognized. ACKNOWLEDGMENTS This work was done during the tenure of an Established Investigatorship from the American Heart Association (PBB), and supported in part by NIH T32 CA09630 (MAJ). Scanner time for this study was donated by the F.M. Kirby Center for Functional Brain Imaging of the Kennedy Krieger Institute. REFERENCES 1. Ross B, Michaelis T. Clinical applications of magnetic resonance spectroscopy. Magn Reson Q 1994;10:191 247. 2. Tedeschi G, Bertolino A, Massaquoi SG, Campbell G, Patronas NJ, Bonavita S, Barnett AS, Alger JR, Hallett M. Proton magnetic resonance spectroscopic imaging in patients with cerebellar degeneration. Ann Neurol 1996;39:71 78.

Proton MRSI of the Cerebellum 705 3. Yang B, Wang Y, Pioro E, Ng TC. Metabolite concentrations of pons, medula, motor cortex in normal human brain using 2D CSI. In: Proc 8th Annual Meeting ISMRM, Denver, 2000. p 1937. 4. Pouwels PJW, Frahm J. Regional metabolite concentrations in human brain as determined by quantitative localized proton MRS. Magn Reson Med 1998;39:53 60. 5. Michaelis T, Merboldt KD, Bruhn H, Hanicke W, Frahm J. Absolute concentrations of metabolites in the adult human brain in vivo: quantification of localized proton MR spectra. Radiology 1993;187:219 227 [erratum Radiology 1993;188(1):288]. 6. Frahm J, Bruhn H, Gyngell ML, Merboldt KD, Hanicke W, Sauter R. Localized proton NMR spectroscopy in different regions of the human brain in vivo. Relaxation times and concentrations of cerebral metabolites. Magn Reson Med 1989;11:47 63. 7. Hattori N, Inoue N, Yoshikubo S, Umeda M, Fukunaga M, Tanaka C, Naruse S, Sawada T, Engelhardt RT. Quantitative analysis of human and macaque brain metabolites using 3 Telsa system. In: Proc 8th Annual Meeting ISMRM, Denver, 2000. p 1067. 8. Hetherington HP, Pan JW, Mason GF, Adams D, Vaughn MJ, Twieg DB, Pohost GM. Quantitative 1H spectroscopic imaging of human brain at 4.1 T using image segmentation. Magn Reson Med 1996;36:21 29. 9. Narayana PA, Johnston D, Flamig DP. In vivo proton magnetic resonance spectroscopy studies of the human brain. Magn Reson Imag 1991;9:303 308. 10. Kreis R, Ernst T, Ross B. Absolute quantitation of water and metabolites in the human brain. II Metabolite concentrations. J Magn Reson B 1993;102:9 19. 11. Hetherington HP, Mason GF, Pan JW, Ponder SL, Vaughan JT, Twieg DB, Pohost GM. Evaluation of cerebral gray and white matter metabolite differences by spectroscopic imaging at 4.1T. Magn Reson Med 1994;32:565 571. 12. Soher BJ, van Zijl PC, Duyn JH, Barker PB. Quantitative proton MR spectroscopic imaging of the human brain. Magn Reson Med 1996;35: 356 363. 13. Barker PB, Szopinski K, Horska A. Metabolic heterogeneity at the level of the anterior and posterior commissures. Magn Reson Med 2000;43: 348 354. 14. Vermathen P, Laxer KD, Matson GB, Weiner MW. Hippocampal structures: anteroposterior N-acetylaspartate differences in patients with epilepsy and control subjects as shown with proton MR spectroscopic imaging. Radiology 2000;214:403 410. 15. Duyn JH, Gillen J, Sobering G, van Zijl PC, Moonen CT. Multisection proton MR spectroscopic imaging of the brain. Radiology 1993;188: 277 282. 16. Maudsley AA, Matson GB, Hugg JW, Weiner MW. Reduced phase encoding in spectroscopic imaging. Magn Reson Med 1994;31:645 651. 17. Henriksen O. In vivo quantitation of metabolite concentrations in the brain by means of proton MRS. NMR Biomed 1995;8:139 148. 18. Hennig J, Pfister H, Ernst T, Ott D. Direct absolute quantification of metabolites in the human brain with in vivo localized proton spectroscopy. NMR Biomed 1992;5:193 199. 19. Frahm J, Bruhn H, Gyngell ML, Merboldt KD, Hanicke W, Sauter R. Localized high-resolution proton NMR spectroscopy using stimulated echoes: initial applications to human brain in vivo. Magn Reson Med 1989;9:79 93. 20. Li S, Dardzinski BJ, Collins CM, Yang QX, Smith MB. Three-dimensional mapping of the static magnetic field inside the human head. Magn Reson Med 1996;36:705 714. 21. Urenjak J, Williams SR, Gadian DG, Noble M. Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types. J Neurosci 1993;13:981 989. 22. Hemmer W, Wallimann T. Functional aspects of creatine kinase in brain. Dev Neurosci 1993;15:249 260. 23. Ilyin SE, Sonti G, Molloy G, Plata-Salaman CR. Creatine kinase-b mrna levels in brain regions from male and female rats. Brain Res Mol Brain Res 1996;41:50 56. 24. Hemmer W, Zanolla E, Furter-Graves EM, Eppenberger HM, Wallimann T. Creatine kinase isoenzymes in chicken cerebellum: specific localization of brain-type creatine kinase in Bergmann glial cells and muscletype creatine kinase in Purkinje neurons. Eur J Neurosci 1994;6:538 549. 25. Kaldis P, Hemmer W, Zanolla E, Holtzman D, Wallimann T. Hot spots of creatine kinase localization in brain: cerebellum, hippocampus and choroid plexus. Dev Neurosci 1996;18:542 554. 26. Hetherington HP, Spencer DD, Vaughan JT, Pan JW. Quantitative (31)P spectroscopic imaging of human brain at 4 Tesla: assessment of gray and white matter differences of phosphocreatine and ATP. Magn Reson Med 2001;45:46 52. 27. Sappey-Marinier D, Vighetto A, Peyron R, Broussolle E, Bonmartin A. Phosphorus and proton magnetic resonance spectroscopy in episodic ataxia type 2. Ann Neurol 1999;46:256 259. 28. Kanazawa I, Kwak S, Sasaki H, Mizusawa H, Muramoto O, Yoshizawa K, Nukina N, Kitamura K, Kurisaki H, Sugita K. Studies on neurotransmitter markers and neuronal cell density in the cerebellar system in olivopontocerebellar atrophy and cortical cerebellar atrophy. J Neurol Sci 1985;71:193 208. 29. Pouwels PJW, Brockmann K, Kruse B, Wilken B, Wick M, Hanefeld F, Frahm J. Regional age dependence of human brain metabolites from infancy to adulthood as detected by quantitative localized proton MRS. Pediatr Res 1999;46:474 485.