Resting Brain Metabolic Activity in a 4 TeslaMagnetic Field

Nora D. Volkow,1,3* Gene-Jack Wang,1 Joanna S. Fowler,2,4 William D. Rooney,2

Christoph A. Felder,1 Jing-Huei Lee,2 Dinko Franceschi,1 Laurence Maynard,1

David J. Schlyer,2 Jullie W. Pan,1 Samuel J. Gatley,1 and Charles S. Springer, Jr.2,4

MRI is a major tool for mapping brain function; thus it is impor-tant to assess potential effects on brain neuronal activity attrib-utable to the requisite static magnetic field. This study usedpositron emission tomography (PET) and 18F-deoxyglucose(18FDG) to measure brain glucose metabolism (a measure ofbrain function) in 12 subjects while their heads were in a 4 T MRIfield during the 18FDG uptake period. The results were com-pared with those obtained when the subjects were in the earth’sfield (PET scanner), and when they were in a simulated MRIenvironment in the PET instrument that imitated the restrictedvisual field of the MRI experiment. Whole-brain metabolism, aswell as metabolism in occipital cortex and posterior cingulategyrus, was lower in the real (4 T) and simulated (0 T) MRIenvironments compared with the PET. This suggests that themetabolic differences are due mainly to the visual field differ-ences characteristic of the MRI and PET instruments. We con-clude that a static magnetic field of 4 T does not in itself affectthis fairly sensitive measure of brain activity. Magn ResonMed 44:701–705, 2000. © 2000 Wiley-Liss, Inc.

Key words: functional imaging; PET; visual stimulation; cerebel-lum; FDG

Brain activity is correlated with metabolism and bloodflow, and changes in these quantities are used by imagingtechnologies to assess regional brain function (1). UsingPET with 18F-fluorodeoxyglucose (18FDG) or with [15O]-labeled water, one can measure regional brain glucosemetabolism and cerebral blood flow, respectively, in arelatively non-invasive manner (2). These have formed thebasis for the use of PET as a common tool for assessing theeffects of perceptual, cognitive, and emotional stimuli onthe human brain. Functional MRI (fMRI), which non-inva-sively measures changes in the MRI signal due to alter-ations in cerebral blood flow and blood oxygen level, isalso an important tool for assessing regional brain function(3). The higher spatial and temporal resolution of fMRI areso advantageous for mapping human brain activity (4) thatmost such studies currently employ fMRI. However, aconcern with fMRI is the potential influence on brainfunction of the strong magnetic field used in this modality.

Rapidly fluctuating magnetic fields can influence brainactivity (5,6). However, the effects on brain function of thestatic magnetic field encountered inside a human MRIinstrument are unclear. Since the strengths of such fieldsused for human research continue to increase (7), it isimportant to assess their effects.

The purpose of this study was to investigate the effectsof a 4 T magnetic field on resting brain activity, which wasmeasured with PET and 18FDG. Subjects were imaged aftereach of three different conditions obtained during the first35 min following 18FDG injection (uptake period): a real(4 T) MRI environment, a simulated (0 T) MRI environ-ment, and the PET environment. Because metabolic activ-ity as assessed with 18FDG mostly reflects that during theinitial 35-min uptake period of the radiotracer into thebrain (8), this allowed us to assess the effects of the 4 Tmagnetic field on resting brain metabolism, and to com-pare it to that obtained with the earth’s magnetic field(10–4 T).

MATERIALS AND METHODS

Subjects

We studied 12 healthy, right-handed subjects (11 malesand one female, 30 6 8 years of age), who were screenedfor the absence of medical, psychiatric, or neurologicaldiseases. Special attention was given to ensure that sub-jects did not abuse addictive substances, and urine toxi-cology studies were performed prior to the imaging ses-sions to ensure lack of psychoactive drug use. The protocolwas approved by the Institutional Review Board atBrookhaven National Laboratory. After a complete de-scription of the study to the subjects, written informedconsent was obtained.

PET Scan and FDG Uptake Conditions

Subjects were scanned using a Siemens HR1 tomograph(FWHM, 63 slices; voxel sizes x: 1.72 mm; y: 1.72 mm; andz: 2.43 mm) with procedures previously described (9).Briefly, emission scans were obtained for a total of 10 min,beginning 50 min after injection of 4–5 mCi of 18FDG.Arterial blood was obtained throughout the procedure tomeasure plasma concentration of 18F, glucose, pO2, andpCO2.

Each subject underwent PET FDG scans that measuredresting brain metabolism under three different experimen-tal conditions obtaining during the 35-min 18FDG uptakeperiod. The temporal order of the first two conditions wasrandomly assigned to be either that in the 4 T MRI magnetor that in the PET scanner (earth’s field). For the third scan,

1Medical Department, Brookhaven National Laboratory, Upton, New York.2Chemistry Department, Brookhaven National Laboratory, Upton, New York.3Department of Psychiatry, State University of New York, Stony Brook, NewYork.4Department of Chemistry, State University of New York, Stony Brook, NewYork.Grant sponsor: U.S. Department of Energy (Office of Biological and Environ-mental Research); Grant number: DE-AC02-98CH10886; Grant sponsor:NIDA.*Correspondence to: Nora D. Volkow, M.D., Medical Department, Bldg. 490,Upton, NY 11973. E-mail: volkow@bnl.govReceived 17 February 2000; revised 25 May 2000; accepted 28 June 2000.

Magnetic Resonance in Medicine 44:701–705 (2000)

© 2000 Wiley-Liss, Inc. 701

the 18FDG uptake period occurred while the subject was ina simulated MRI environment in the PET camera. Thehead was positioned inside a volume (birdcage) MRI RF-coil covered with opaque blue paper, and situated withinan ;25-cm-diameter cylindrical tube sufficiently long(;90 cm) to mimic the restricted visual field of the MRImagnet bore. For the MRI scanner condition, foam blockswere used within the head RF-coil to reduce head motion,and foam earplugs (noise reduction rating 32 dB; AearoCompany) were inserted in the ear canal to reduce noise. Agradient echo sequence was used to simulate the fMRIacquisition. The recycle time was 75 msec, the gradientamplitudes were all less than 10 mT/m, and the slew ratewas 15 T/m/sec. To improve access to the arterial line,subjects were positioned so their brains were located;75 cm from the magnet isocenter. Interpolating a fieldmap, we find that the magnetic field strength at this pointis no less than 3.9 T. For the two PET scanner conditions,an individually-fitted foam headholder was used to reducemotion. The PET and the MRI rooms were illuminated atsimilar levels, maintained at similar temperatures, andambient noise was kept to a minimum. Prior to 18FDGinjection, subjects had two catheters placed: a venous onefor radiotracer injection and an arterial one for blood sam-pling. Also before 18FDG injection, subjects were asked toverbally provide self-reports for sleepiness or anxiety us-ing a relative scale of 0 (none) to 10 (maximum sensation).Subjects were allowed to acclimatize to the environmentfor 30 min prior to injection of 18FDG. This simulated thetuning, matching, shimming, and scouting period of anMRI examination. Beginning ;20 min before the injection,and continuing throughout the uptake period, currents inthe gradient coils of the MR instrument were pulsed inorder to replicate the acoustical fMRI environment.Though no RF was actuated, the gradient pulses of a gra-dient-recalled echo sequence were applied. This causedthe equivalent of an amplitude-modulation of the 4 T field,never amounting to more than 10–2 T in amplitude and15 Hz in frequency, at the brain. The noise level at thesubject’s head produced by a similar gradient echo se-quence run on our MRI instrument was 93 dBA. Subjectswere asked to refrain from moving or speaking during the35-min 18FDG uptake period. They were also requested tokeep their eyes open to ensure that they did not sleepduring the 18FDG uptake period. At the end of this testing,the subjects were taken out of the PET or the MRI scannerand into another room. Ten minutes later, the subjectswere brought into the PET scanner, where they were po-sitioned in the gantry as described (10). The individual-ized headholder allowed accurate repositioning for eachPET scan. The subjects’ head was positioned in the gantrywith the aid of two orthogonal laser beams, one placed atthe corner of canthus and the other parallel to the sagittalplane. Metabolic images were obtained as previously de-scribed (10). A ten-min emission scan was started 50 minafter 18FDG injection, and transmission scans were per-formed simultaneously.

Data Analysis

Regions were selected using a template of 423 non-overlapping regions of interest (ROIs) based on Talairach andTournoux’s atlas (11). The ROIs were grouped first into

32 subregions, and were then further compressed into10 composite cortical, subcortical, and cerebellar regions.Measurement of global brain metabolism was obtained byaveraging the values from all of the ROIs. In addition, we alsocomputed “relative” measures for the 10 composite regionsby normalizing the activity to that of the global metabolic rate(ROI/global brain).

Differences in regional metabolism between the threeuptake environments were tested with repeated ANOVAfor the 10 composite brain regions. In those regions show-ing significant differences, post hoc t tests were then per-formed to assess which environments caused these differ-ences. In consideration of the “multiple comparison prob-lem” incurred by analyzing values for the 10 compositebrain regions, the significance level for the ANOVA wasset at P # 0.01. To determine the influence of sleepiness oranxiety on the regional differences seen across environ-ments, we calculated the correlations (Pearson productmoment) between the regions that differed and the scoresfor these self-reports of sleepiness or anxiety. Also, toassess the effects on regional brain metabolism of the tem-poral order of the first and second FDG studies (for sixsubjects the PET condition was first, and for six subjectsthe MRI was first), we used a repeated measure ANOVA.

Metabolic images were also analyzed using the Statisti-cal Parametric Mapping SPM98 software package (12). Sta-tistical parametric maps were displayed in coronal, trans-verse, and sagittal views showing only those pixels thatreached a statistical significance of P , 0.001. The pixelwith the highest Z-score within each isolated significantregion was chosen to report its coordinates in Talairachand Tournoux’s atlas reference space (11).

RESULTS

Global brain glucose metabolism differed across environ-ments (F 5 3.3, df 2,35, P , 0.05). Post hoc t tests showedthat it was significantly lower for the real (4 T) MRI (22.9 62 mmol/100g/min) than for the PET (25.2 64 mmol/100g/min) environment (P , 0.05). Though notsignificant (P , 0.13), global brain metabolism tended toalso be lower for the sham (0 T) MRI (23.6 64 mmol/100g/min) than for the PET environment (Fig. 1).Table 1 shows the percent changes in metabolism betweenthe three conditions for the whole brain (global) and forthe various brain regions.

Regional measures showed significant environmental ef-fects in occipital cortex (F 5 5.9, P , 0.01), posteriorcingulate gyrus (F 5 5.4, P , 0.01), and cerebellum (F 55.5, P , 0.01). Post hoc t tests showed that these differ-ences were significant between the 4 T MRI and the PETenvironments for the occipital cortex (P , 0.01), the pos-terior cingulate gyrus (P , 0.02), and the cerebellum (P ,0.05). However, there were no significant differences be-tween the real (4 T) and sham (0 T) MRI environments, andthe regions that differed between the PET and sham MRIenvironments were the same as those that differed be-tween the PET and real MRI environments (Fig. 2). Sepa-rate analysis of subregions within the occipital cortexshowed that the differences were mainly in areas 17 (F 510.4, P , 0.001) and 18 (F 5 11.2, P , 0.0004).

The repeated measure ANOVA to assess the effect onregional brain metabolism of the temporal order (between

702 Volkow et al.

the first and second 18FDG studies) showed a trend towardregional differences in the cerebellum. Cerebellar metabo-lism was lower (P , 0.03) for the second than for the first18FDG study, regardless of whether it was the PET (first:26.8 6 3 mmol/100 g/min; second: 24.0 6 3 mmol/100g/min) or the MRI environment (first: 23.8 6 3 mmol/100g/min; second: 22.6 6 3 mmol/100 g/min).

The analyses of the “relative” measures showed findingssimilar to those obtained for the analyses of the “absolute”metabolic measures, except for a significant environmentaleffect in the thalamus (F 5 4.9, P , 0.02) (Fig. 3). There,post hoc t tests showed that relative metabolism was sig-nificantly higher for the PET (1.21 6 0.12) and the real MRI(1.23 6 0.12) environments than for the sham MRI envi-ronment (1.14 6 0.10) (P , 0.05). Also, the post hoc t testsrevealed that relative cerebellar metabolism for the shamMRI was significantly lower than for the PET (P , 0.01)and the MRI environments (P , 0.01).

SPM analyses revealed significant differences betweenthe three environments that corroborated the ROI analysisfindings. Differences between the real MRI and PET werelocalized in the occipital cortex (higher occipital metabo-

lism for PET); those between the real and sham MRI werein the cerebellum (higher cerebellar metabolism for realMRI); and those between the PET and the sham MRI werelocalized in the occipital cortex and cerebellum (higheroccipital and cerebellar metabolism for the PET) (Fig. 4).

Self-reports for sleepiness or anxiety were not correlatedwith the absolute metabolic measures in occipital cortex,posterior cingulate gyrus, or cerebellum for any of theenvironments. For the “relative” metabolic measures dur-ing the sham MRI environment, self-reports of sleepinesswere negatively correlated with cerebellar metabolism (r 50.79, df 10, P , 0.003), and there was a trend for a negativecorrelation with thalamic metabolism (r 5 0.56, df 10, P ,0.06). The correlations for the other environments werenot significant. Self-reports for anxiety did not differacross environments, but showed a trend for higher sleep-iness scores for the sham MRI than for the PET or the realMRI conditions (P , 0.09).

DISCUSSION

This study shows higher global brain glucose metabolicactivity in the occipital cortex and posterior cingulate

FIG. 2. Absolute glucose metabolic rates for the 10 composite brainregions in the various environments (PET, MRI, and sham MRI).*Metabolism differed across environments in occipital cortex, pos-terior cingulate gyrus, and cerebellum (P , 0.01).

Table 1Percent Changes in Metabolism Between MRI and PET [(MRI 2PET) / PET 3 100], Between Sham MRI and PET [(Sham MRI 2PET) / PET 3 100] and Between Sham MRI and MRI [(ShamMRI 2 MRI) / MRI 3 100]

MRI versusPET

Sham MRIversus PET

Sham MRIversus MRI

Global 28 6 11 25 6 13 4 6 12Frontal 27 6 13 24 6 14 5 6 11Parietal 26 6 11 23 6 14 4 6 14Temporal 28 6 11 25 6 12 4 6 13Occipital* 211 6 10 28 6 13 4 6 13CG anterior 27 6 14 24 6 16 4 6 12CG posterior 210 6 13 28 6 13 3 6 12Insula 28 6 13 22 6 14 7 6 15Thalamus 26 6 15 210 6 16 24 6 16Striatum 26 6 13 25 6 14 2 6 11Cerebellum 27 6 10 210 6 12 22 6 14

*P , 0.01; Repeated ANOVA.

FIG. 1. PET images of brain glucose metabolism (18FDG) at thelevels of the occipital cortex and cerebellum for one of the subjectstested in the PET, the MRI, and the sham MRI environments. Noticethe higher occipital cortex metabolic rate in the PET than in the real(4 T) or sham (0 T) MRI environments. Also notice the lower cere-bellar metabolism in the sham MRI environment.

Brain Metabolism in a 4 T Magnetic Field 703

gyrus in the PET environment than in the MRI environ-ment. However, it failed to find any significant differencesin metabolism in these brain regions between the real (4 T)and sham (0 T) MRI environments. PET studies haveshown that metabolic activity in the occipital cortex isdirectly influenced by the intensity and complexity ofvisual stimuli presented to the subject during the 18FDGuptake period (13). The posterior cingulate gyrus is neu-roanatomically connected with primary visual regions (ar-eas 17 and 18) (14), participates in monitoring of eyemovements and eye position (15), and plays an essentialrole in the use of topographical information (16). The shamMRI environment was designed to mimic the visual fieldavailable to a subject in the bore of an MRI magnet duringa typical fMRI scan. The lack of a significant difference invisual area metabolism between the real and sham MRIindicates reasonable success in this attempt. Thus, thedifferences observed between the PET and MRI environ-ments can most likely be attributed to the complexity ofthe visual field a subject can observe in the PET instru-ment. In the MRI, the subject’s visual field is mainly lim-ited to the bore walls. In the PET, the thin gantry axialplane allows the subject—with eye movement—a diversevisual field. Consequently, greater eye movement withinthe PET instrument could account for the activation in theposterior cingulate gyrus, and the rich visual field couldaccount for the activation in the occipital cortex. Theoverall higher metabolic activity seen in the PET than inthe MRI condition is likely to also reflect the more limitedvisual sensory inputs in the MRI environment, which thenleads to an overall decrease in whole-brain metabolism(17).

The significant differences observed in the cerebellumbetween the environments are harder to explain. Therewas a lower cerebellar metabolism in the sham MRI thanin the real MRI or the PET environments. This also per-tains to the differences in “relative” metabolism in thethalamus, where there was also lower activity in the shamMRI environment. Therefore, these differences can not beattributed to the distinctions of the PET and the MRIenvironments. A systematic procedural aspect is that the

sham MRI environment was always the last studied of thethree conditions. It is possible that because the subjectswere by then more acclimatized to the procedure, theymay have been less alert and apprehensive, and morerelaxed about the experiment. In our study the subjects didnot report lower anxiety for the sham MRI condition thanfor the other conditions. However, the self-reports of sleep-iness tended to be higher during the sham MRI condition,which suggest that they may have been more relaxed andless alert during this condition. Moreover, self-reports ofsleepiness were inversely correlated with “relative” me-tabolism in cerebellum and thalamus. That the lower cer-ebellar metabolism in the sham MRI condition was due inpart to the order effect is supported by the trend towardlower cerebellar metabolism in the second than in the firstscan, regardless of which environment was studied first.We had previously shown a similar association betweenmetabolic activity in the cerebellum and thalamus andsleepiness induced by the anxiolitic drug lorazepam (18).Though the literature on the role of the cerebellum insleepiness and sedation is limited, there is evidence fromanimal studies that the cerebellum plays a role in thesleep-wakefulness cycle (19,20) and that the cerebellarPurkinje neurons are an important locus for the sedativeeffects of alcohol (21). The significant differences betweenthe first (PET or real MRI) and second (sham MRI) condi-tions highlights the sensitivity of regional brain metabolicmeasures to the environmental conditions, including nov-elty and familiarity, under which the experiments are per-formed. This is also consistent with a report that the sug-gestive power of the informed consent process for newvolunteers is at least as significant in this regard as thepresence of a 4 T magnetic field (22). An obvious improve-ment of the present study would be to randomize all threeconditions.

It should be noted that although the sham MRI environ-ment lacked the visual stimulation of the PET environ-ment, it also lacked the auditory stimulation of the realMRI environment. However, we did not detect signifi-cantly increased activation in auditory cortices during the

FIG. 3. Relative metabolic values (ROI/whole brain) for the10 composite brain regions in the various environments (PET, MRI,and sham MRI). *Metabolism differed across environments in oc-cipital cortex, posterior cingulate gyrus, thalamus, and cerebellum(P , 0.01).

FIG. 4. SPM images of the metabolic differences in the differentenvironments (P , 0.001).

704 Volkow et al.

real MRI condition. This was surprising, since severalfMRI studies have shown activation of primary auditorycortices attributable to gradient acoustic noise (23–26).However, the fMRI signal intensity appears to decreasewith repeated auditory stimulation (27). This could reflectthe fact that most auditory neurons respond to a suddenchange in intensity and are much less sensitive to aninvariant long-duration stimulus (28). There can also behabituation to continual noise stimulation (29). Thus, it ispossible that at the time of FDG administration, which was20 min after acclimatization to the MRI environment, thesubjects were already habituated to the gradient acousticnoise. Note that the PET 18FDG method lacks the temporalresolution to discriminate the differences in the time-de-pendent auditory response and the responses observedreflect those averaged over a 30-min period. An alterna-tive, though less likely, explanation is that when a subjectis wearing earplugs, the level of actual stimulation (again,averaged over 30 min) is not strong enough to elevateoverall auditory metabolism beyond that elicited in theregular PET environment.

This study focused on the effects of the static magneticfield on brain function, as assessed with brain glucose me-tabolism. Because an actual fMRI examination also involves(much smaller) varying magnetic fields and the use of RFirradiation, further studies can be designed to assess whethereither of these aspects affects brain metabolism. This is rele-vant because several studies have shown that fluctuatingmagnetic fields can influence brain activity (5,6). However,studies evaluating static magnetic fields are few, and thoughthere is some evidence from animal studies that a field mayinterfere with performance of certain behavioral tasks (30),such an effect has not been corroborated in humans. More-over, a recent study failed to detect any effect from an 8 Tmagnetic field on cognitive function in healthy human vol-unteers (7). The results from our study also fail to documentany effect from a 4 T magnetic field on brain activity asassessed with brain glucose metabolism. We therefore con-clude that static magnetic fields of themselves, at thestrengths currently used for MR imaging, do not significantlyaffect brain function.

ACKNOWLEDGMENTS

The authors thank David Alexoff, Robert Carciello, Rich-ard Ferrieri, Payton King, Alex Levy, Robert MacGregor,Noelwah Netusil, Naomi Pappas, Colleen Shea, DonaldWarner, and Christopher Wong for advice and assistance.

REFERENCES

1. Magistretti PJ, Pellerin L. Cellular mechanisms of brain energy metab-olism and their relevance to functional brain imaging. Philos Trans RSoc Lond B Biol Sci 1999;354:1155–1163.

2. Weng X, Ding Y-S, Volkow ND. Imaging the functioning human brain.PNAS 1999;96:11073–11074.

3. Ugurbil K, Hu X, Chen W, Zhu XH, Kim SG, Georgopoulos A. Func-tional mapping in the human brain using high magnetic fields. PhilosTrans R Soc Lond B Biol Sci 1999;354:1195–1213.

4. Volkow ND, Rosen B, Farde L. Imaging the living human brain: mag-netic resonance imaging and positron emission tomography. Proc NatlAcad Sci USA 1997;94:2787–2788.

5. Walsh V, Rushworth M. A primer of magnetic stimulation as a tool forneuropsychology. Neuropsychologia 1999;37:125–135.

6. Crasson M, Legros JJ, Scarpa P, Legros W. 50 Hz magnetic field expo-sure influence on human performance and psychophysiologicalparameters: two double-blind experimental studies. Bioelectromagnet-ics 1999;20:474–486.

7. Kangarlu A, Burgess RE, Zhu H, Nakayama T, Hamlin RL, AbduljalilAM, Robitaille PM. Cognitive, cardiac, and physiological safety studiesin ultra high field magnetic resonance imaging. Magn Reson Imaging1999;17:1407–1416.

8. Sokoloff L, Reivich M, Kennedy C, Des Rosiers MH, Patlak CS, Petti-grew KD, Sakurada O, Shinohara M. The [14C]deoxyglucose method forthe measurement of local cerebral glucose utilization: theory, proce-dure, and normal values in the conscious and anesthetized albino rat.Neurochem 1977;28:897–916.

9. Volkow ND, Wang G-J, Hitzemann R, Fowler JS, Wolf AP, Pappas N,Biegon A, Dewey SL. Decreased cerebral response to inhibitory neuro-transmission in alcoholics. Am J Psychiatry 1993;150:417–422.

10. Wang G-J, Volkow ND, Wolf AP, Brodie JD, Hitzemann RJ. Intersubjectsvariability of brain glucose metabolic measurements in young normalmales. J Nucl Med 1994;35:1457–1466.

11. Talairach J, Tournoux P. A co-planar stereotaxic atlas of a human brain.New York: Thieme, 1988. 108 p.

12. SPM95 Software. London: MRC Cyclotron Unit, Hammersmith Hospi-tal, 1995.

13. Phelps ME, Mazziotta JC, Kuhl DE, Nuwer M, Packwood J, Metter J,Engel J Jr. Tomographic mapping of human cerebral metabolism visualstimulation and deprivation. Neurology 1981;31:517–529.

14. Ahmed AK, Dong K, Sugioka K, Yamadori T. Afferent projections to thecingulate cortex in albino rats: a study with a retrograde labelingmethod using fluoro-gold. Kobe J Med Sci 1995;41:247–255.

15. Olson CR, Musil SY, Goldberg ME. Single neurons in posterior cingu-late cortex of behaving macaque: eye movement signals. J Neurophysiol1996;76:3285–3300.

16. Sutherland RJ, Whishaw IQ, Kolb B. Contributions of cingulate cortex totwo forms of spatial learning and memory. J Neurosci 1988;8:1863–1872.

17. Mazziotta JC, Phelps ME, Kuhl DE. Human cerebral metabolism stud-ied during limitation in sensory inputs. Neurology 1981;31:517–529.

18. Volkow ND, Wang GJ, Hitzemann R, Fowler JS, Pappas N, LowrimoreP, Burr G, Pascani K, Overall J, Wolf AP. Depression of thalamicmetabolism by lorazepam is associated with sleepiness. Neuropsycho-pharmacology 1995;12:123–132.

19. Cunchillos JD, De Andres I. Participation of the cerebellum in theregulation of the sleep-wakefulness cycle. Results in cerebellectomizedcats. Electroencephalogr Clin Neurophysiol 1982;53:549–558.

20. Paz C, Reygadas E, Fernandez-Guardiola A. Sleep alterations followingtotal cerebellectomy in cats. Sleep 1982;5:218–226.

21. Spuhler K, Hoffer B, Weiner N, Palmer M. Evidence for genetic correlationof hypnotic effects and cerebellar Purkinje neuron depression in responseto ethanol in mice. Pharmacol Biochem Behav 1982;17:569–578.

22. Erhard P, Chen W, Lee J-H, Ugurbil K. A study of effects reported bysubjects at high magnetic fields. Proc Soc Magn Reson 1995;3:1219.

23. Bandettini PA, Jesmanowicz A, Van Kylen J, Birn RM, Hyde JS. Func-tional MRI of brain activation induced by scanner acoustic noise. MagnReson Med 1998;39:410–416.

24. Hall DA, Summerfield AQ, Goncalves MS, Foster JR, Palmer AR, Bow-tell RW. Time-course of the auditory BOLD response to scanner noise.Magn Reson Med 2000;43:601–606.

25. Ulmer JL, Biswal BB, Yetkin FZ, Mark LP, Mathews VP, Prost RW,Estkowski LD, McAuliffe TL, Haughton VM, Daniels DL. Cortical acti-vation response to acoustic echo planar scanner noise. J Comput AssistTomogr 1998;22:111–119.

26. Talavage TM, Edmister WB, Ledden PJ, Weisskoff RM. Quantitativeassessment of auditory cortex responses induced by imager acousticnoise. Hum Brain Map 1999;7:79–88.

27. Jancke L, Buchanan T, Lutz K, Specht K, Mirzazade S, Shah NJ. Thetime course of the BOLD response in the human auditory cortex toacoustic stimuli of different duration. Cogn Brain Res 1999;8:117–124.

28. Benson DA, Hienz RD, Goldstein MHJ. Single-unit activity in the au-ditory cortex of monkeys actively localizing sound sources: spatialtuning and behavioral dependency. Brain Res 1981;219:249–267.

29. Rodriguez GP, Gerhardt KJ. Adaptation properties of the acoustic reflexin response to continuous-, intermittent- and industrial-noise stimula-tion. Audiology 1988;27:344–355.

30. Levine RL, Bluni TD. Magnetic field effects on spatial discriminationlearning in mice. Physiol Behav 1994;55:465–467.

Brain Metabolism in a 4 T Magnetic Field 705