Functional magnetic resonance imaging fMRI is becoming a popular noninvasive tool for - PDF document

Functional magnetic resonance imaging (fMRI) ishas dominated this field since its discovery [1]. BOLD How well do we up to 95% of regional cerebellarblood flow increases might be activity. effect on local metabolism would be small comparedwith the apparent changes recorded in imagingexperiments. By comparison, the major determinantof cortical oxygen and glucose consumption is there-establishment of ionic concentrations via theNa+–K+ATPase after synaptic activity, defined here, defined hereup to 95% of regional cerebellar blood flow increasesmight be dependent on postsynaptic activity [11].is synaptic activity, to which one would expect fMRIsingle-cell activity. These data emphasize theapparent dominance of synaptic activity, but howRelevance of the relationship between action potentialsAneuronal action potential can be defined asTRENDSin Neurosciences Vol.25 No.1 January 2002http://tins.trends.com Box 1.The haemodynamic response and fMRI BOLD signals TRENDS in Neurosciences (3) HaemodynamicresponsefMRI BOLDresponse Neuronalactivity - Excitatory activityand inhibitory activityunknown- Blood flow oxygenation level- Blood volume- Haematocrit- TR, repetition- TE, echo time- Spin or gradient Neurovascular Detection byStimulusin background The blood oxygenation level-dependent (BOLD)signal can be thought of as having several keydeterminants (shown from left to right in Fig. I): therelationship between neuronal activity and triggeringresponse itself; and the way in which this response isscanner.The many experimental parameters in functionalMRI (fMRI) scanning that affect the amount of BOLDsignal observed by any particular scanner includemagnetic field strength, echo time and the type ofimaging technique involved. For example, a 1% BOLDsignal at an echo time of 30 ms is equivalent to 2% atan echo time of 60 ms, even if the haemodynamicsusceptible to various artefacts, including headtefacts, including headTogether, many factors will affect the amount bywhich the BOLD response reflects a givenhaemodynamic response, which makes the responsedifficult to quantify.Much work has been carried out to identify thenature of the haemodynamic response itself, inparticular by Friston [b], furthering the Balloon[c].The vascular basis of therelative imbalance between increases in localcerebral blood flow (CBF) and concurrent (albeitsmaller) increases in oxygen metabolism,whichcauses a transient drop in thedeoxyhaemoglobin:oxyhaemoglobin ratio, and isxyhaemoglobin ratio, and isphysiological factors that also contribute to changesin deoxyhaemoglobin concentration include bloodvolume, vascular geometry, haematocrit and basaloxygenation levels [f,g]. These important initialresponse to a stimulus or background modulation; (2) the complexrelationship between neuronal activity and triggering ahaemodynamic response (termed neurovascular coupling); (3) thehaemodynamic response itself; and (4) the way in which thisresponse is detected by an MRI scanner. EPSPs summed linearly, then at least 75 afferent, then at least 75 afferentdecrease the probability of cell firing (Box 2). Spikingactivity thereafter adapts quickly, whereas synapticLFPactivity might be maintained during stimulusactivity might be maintained during stimulusbetween cortical synaptic activity and cell spikingactivity is difficult to standardize and quantify, andareas. As fMRI BOLD inherently measures relativeunclear.Logically, however, one would expect an, one would expect anbetween axon and dendrite of the same neurone.Synaptic activity should therefore correlate with thecorrelated to spiking activity. Ultimately, thisfor the sake of a single flower-bed’[29]. So whatfMRI BOLD detects population activityects population activitycompromise shorter acquisition times with a lowerspatial resolution of cells. As a result of this scale, conventional BOLDimaging (and similarly, scalp electrophysiology), scalp electrophysiology)currently unclear whether fMRI can differentiatebetween these small activity changes in large cellularrate. Taking things to the next stage, a furtherand inhibitory synaptic connections [32] (Box 2). Therelationship of the haemodynamic response to theseTRENDSin Neurosciences Vol.25 No.1 January 2002http://tins.trends.com factors aside, the haemodynamic response can vary widely across corticalareas and between species. Different aspects of the haemodynamicresponse might change on different timescales, and might have differentneural determinants and different consequences for the BOLD signal.It is also widely recognized that the BOLD signal occurs not only at thedownstream from the neuronally active regions [f,h]. By implication, suchsignal changes would be spatially displaced from the activated neurallikely to be limited by the microvascular density, which will always be lowerays be lowerknown as the ‘brain versus vein’ debate [i]. Spin-echo fMRI techniquesminimize these venous contributions and thus might be useful in resolvingto noise [j,k]. At higher field strengths, the capillaries exert a larger effect onimage intensity [l,m]. In combination, these two might therefore becomeReferencesaTurner, R. and Ordidge, R.J. (2000) Technical challenges of functional magneticIEEE Eng. Med. Biol. Mag.bFriston, K.J. (2000) Nonlinear responses in fMRI: the Balloon model, VolteracBuxton, R.B. (1998) Dynamics of blood flow and oxygenation changes during braindTurner, R. (1992) Magnetic resonance imaging of brain function. eMenon, R.S. and Kim, S.G. (1999) Spatial and temporal limits in cognitive neuroimagingTrends Cognit. Sci.fOgawa, S. contrast magnetic resonance imaging. Acomparison of signal characteristics with agKennan, R.P. (1994) Intravascular susceptibility contrast mechanisms in tissues.hKim, S.G. iFrahm, J. jBoxerman, J.L. kLee, S.P. (1999) Diffusion-weighted spin-echo fMRI at 9.4 T: microvascular/tissuelMenon, R.S. mLogothetis, N.K. (1999) Functional imaging of the monkey brain. in overall population or background activity. It isneuronal synchrony. Global scaling techniques canTRENDSin Neurosciences Vol.25 No.1 January 2002http://tins.trends.com There is currently no evidence that the recycling and repackagingof neurotransmitters and the restoration of ionic concentrationsafter synaptic transmission differs between excitatory andinhibitory synapses. Because cortical glucose use reflectspresynaptic rather than postsynaptic activity [a], the release ofinhibitory or excitatory transmitters must both be energy-consuming processes; for example, inhibitory activity results invity results inInhibitory synaptic activity might modulate the functionalmagnetic resonance imaging (fMRI) blood oxygenation level-dependent (BOLD) response by changing metabolic demand, oractivity. Increases in inhibitory activity demand greaterexcitatory input in order to achieve supra-threshold activity, inother words, more synaptic activity is required for each actionneurotransmitters might also feasibly cancel out the reduction inactivity of the inhibited postsynaptic cell. However, it is unlikelythat a substantial volume of cortex could sustain a high level ofinhibitory activity, producing a simultaneously low firing rateand high metabolic rate [c]. The majority of the cortex (70–80%)consists of pyramidal cells, which are excitatory regular-spikinginhibitory [d,e] (approximately one inhibitory synapse for everyfive excitatory synapses [f]). It has been argued that because oftheir reduced number, strategically superior location andincreased efficiency [g], there could be lower metabolic demandduring inhibition compared with excitation. Accordingly, onegroup has proposed that inhibition, unlike excitation, does notelicit a measurable change in the BOLD signal [h]. However,conditions that appear to involve inhibitory interactions [i]. Bothinhibitory neurones are likely to be balanced so that it is unlikelyelyno cell would reach threshold. A recent model suggests thath threshold. A recent model suggests thatThe cerebellum provides an interesting opportunity to studythe neural origin of fMRI BOLD, because its principal cortical cells,the Purkinje cells, are inhibitory. However, in rats, no simplecorrelation has been found between blood flow and Purkinje cellurkinje cell()to postsynaptic GABA (inhibitory) activity; but are attenuatedwhen synaptic potentials are abolished by blocking glutamate-activity alone provides the basis for the vascular responsesactivation of inhibitory interneuronal firing. The cerebellum is amulti-layered structure containing many types of cell, and thereblood flow (S-J. Blakemore, PhD thesis, University CollegeLondon, 2000). However, these results seem to generalize tosomatosensory cortex [o]. Further investigation into the cellularbasis of haemodynamic change in the cerebellum might addressrates in different areas of cortex with different levels of excitatoryand inhibitory activity might therefore create distinctrelationships between neural activity and the BOLD response.vity and the BOLD response.(but not directly vascular) drugs, for example, GABA-mediatedinhibitory blockers, might reveal some of the true contributions ofinhibitory and excitatory activity to the BOLD response in cortex-ReferencesaJueptner, M. and Weiller, C. (1995) Review: does measurement of regionalbAckermann, R.F. duration recurrent inhibition of hippocampal pyramidal cells. cRaichle, M.E. (1987) Circulatory and metabolic correlates of brain function(Mountcastle, V.B. , eds), American Physiological SocietydConnors, B.W. and Gutnick, M.J. (1990) Intrinsic firing patterns of diverseTrends Neurosci.eDeFelipe, J. and Farinas, I. (1992) The pyramidal neuron of the cerebralfBeaulieu, C. and Colonnier, M. (1985) Alaminar analysis of the number ofgKoos, T. and Tepper, J.M. (1999) Inhibitory control of the neostriatalhWaldvogel, D. iHeeger, D.J. (1999) Motion opponency in visual cortex. jScannell, J.W. and Young, M.P. (1999) Neuronal population activity andkShadlen, M.N. and Newsome, W.T. (1998) The variable discharge of corticalneurons: implications for connectivity, computation, and information coding.lTagamets, M.A. and Horwitz, B. (2001) Interpreting PET and fMRImMathiesen, C. nAkgoren, N. (1996) Cerebral blood flow increases evoked by electricaloNielsen, A. and Lauritzen, M. (2001) Coupling and uncoupling of activity-dependent increases of neuronal activity and blood flow in ratpLeslie, R.A. and James, M.F. (2000) Pharmacological magnetic resonance Box 2.Does inhibitory activity contribute to the BOLD signal? regional firing activity [23,33–35]. Assuming thatfMRI signals are representative of global synapticfMRI signals are representative of global synapticbearing upon the non-absolute nature of all fMRIBOLD signals, and the varying relationship betweenaction potentials and synaptic energy demand. It isarguable that background modulations induced byrelative changes between active and ‘resting’state,state,predictable how fMRI might express stimulus-correlated activation changes on top of simultaneousbackground modulations.To conclude, fMRI BOLD signals are clearlydependent on the variability and inter-relationshipsaction potentials. Further investigation of thisTRENDSin Neurosciences Vol.25 No.1 January 2002http://tins.trends.com AcknowledgementsO.J.A is supported byMerck Sharp and Dohme.The authors thank theWolfson Brain Imagingchairman, J.D. Pickard,and also E.T. Bullmoretheir comments in theReferences1Ogawa, S. 2Pauling, L. and Coryell, C.D. (1936) The magnetic3Thulborn, K.R. 4Lou, H.C. 5Villringer, A. and Dirnagl, U. (1995) Coupling ofMetab. Rev.6Logothetis, N.K. investigation of the basis of the fMRI signal.7Arthurs, O.J. 8Ogawa, S. (2000) An approach to probe some97, 11026–110319Tsubokawa, T. 10Ngai, A.C. potentials during somatosensory stimulation in11Mathiesen, C. 12Brinker, G. 13Nielsen, A. and Lauritzen, M. (2001) Couplingneuronal activity and blood flow in rat14Allison, T. human and monkey cerebral cortex bystimulation of the median nerve. Areview of scalp114,15Eccles, J.C. (1951) Interpretation of actionElectroencephalogr. Clin. Neurophysiol.16Lopes da Silva, F. and Storm van Leeuwan, W.(Brazier, M.A.B. and Petsche,17Lopes da Silva, F. (1991) Neural mechanismsElectroencephalogr. Clin.18Heeger, D.J. 19Rees, G. (2000) Adirect quantitative20Ueki, M. 21Ances, B.M. activation to blood flow in the somatosensorylinear. 22Rees, G. cerebral blood flow measurements by varying the23Creutzfeldt, O.D. (1975) NeurophysiologicalBrain Work: The Coupling of Function,(Ingvar,24Yarowsky, P. 25Raichle, M.E. (1987) Circulatory and metabolic(Mountcastle, V.B. , eds), American26Takahashi, S. 27Kandel, E.R. and Schwartz, J.H. (1991) Directlygated transmission at central synapses. In(Vol. 3, 3rd edn)28Kadekaro, M. 29Malonek, D. and Grinvald, A. (1996) Interactionsbetween electrical activity and cortical30Kim, D.S. 31Scannell, J.W. and Young, M.P. (1999) Neuronal32Taylor, J.G. 33Creutzfeldt, O.D. (1995) 34Livingstone, M.S. and Hubel, D.S. (1981) Effects35Chawla, D. their mean activity levels. 11,1389–141136Kastner, S.