MR imaging in the non-human primate: studies of function and of ...

MR imaging in the non-human primate: studies of functionand of dynamic connectivityNikos K Logothetis

Since its early development in the late 1940s, nuclear magnetic

resonance has become a powerful tool for applications ranging

from chemical analysis or the study of the structure of solids to

biomedical investigations. In the early 1990s the potential of this

technique for functional brain mapping was demonstrated,

causing unprecedented excitement in both basic and clinical

neuroscience. It was shown that by using the appropriate pulse

sequences the so-called functional magnetic resonance imaging

(fMRI) technique can be made sensitive to local magnetic

susceptibility alterations produced by changes in the

concentration of deoxyhemoglobin in venous blood vessels. This

blood-oxygenation-level-dependent (BOLD) contrast

mechanism was successfully implemented in awake human

subjects, in small animals, and recently in the non-human

primate — the experimental animal of choice for the study of

cognitive behavior. Simultaneous imaging and electrode

recordings promise new insights into the mechanisms by which

large-scale networks in the brain contribute to the local neural

activity recorded at a given cortical site. Moreover, the use of

MRI-visible tracers and of electrical microstimulation applied

during imaging proves to be ideal for the study of connectivity

in the living animal.

AddressesMax Planck Institute for Biological Cybernetics, Department of

Physiology of Cognitive Processes, Spemannstr 38, 72076 Tubingen,

Germany

e-mail: [email protected]

Current Opinion in Neurobiology 2003, 13:1–13

This review comes from a themed issue on

New technologies

Edited by Karel Svoboda and Kamil Ugurbil

0959-4388/$ – see front matter

� 2003 Elsevier Ltd. All rights reserved.

DOI 10.1016/j.conb.2003.09.017

AbbreviationsBOLD blood-oxygen-level-dependent

CBF cerebral blood flow

CNS central nervous system

dHb deoxyhemoglobin

EEG electroencephalography

fMRI functional magnetic resonance imaging

FOV field of view

LFPs local field potentials

LGN lateral geniculate nucleus

mEFP mean extracellular field potential

MRI magnetic resonance imaging

MUA multiple-unit spiking activity

PET positron emission tomography

SNR signal to noise ratio

TE echo time

TR repetition time

V1 primary visual cortex

WGA-HRP wheat germ agglutinin conjugated to horseradish

peroxidase

IntroductionUnderstanding brain function requires comprehension of

the physiological workings of its cellular components, as

well as a detailed knowledge of their structural and

functional connectivity. Moreover, the functional plasti-

city of the brain, reflected in its capacity for anatomical

reorganization, means that a mere ‘snapshot’ of its archi-

tecture is not enough. Desirable, instead, are repeated

anatomical and physiological observations of its connec-

tivity patterns at different organizational levels. In vivoneuroimaging, especially the spatiotemporally resolved

magnetic resonance imaging (MRI), together with its

functional variant, functional magnetic resonance imag-

ing (fMRI), is therefore of obvious importance, and it is

currently one of our best tools for linking cognition and

action with their neural substrates in humans.

At the moment, brain fMRI relies mainly on the blood-

oxygen-level-dependent (BOLD) contrast mechanism,

which was first described in rat studies using high mag-

netic fields [1]. BOLD contrast is derived from the field

inhomogeneities induced by deoxyhemoglobin (dHb),

which is confined in the intracellular space of the red

blood cells that in turn are restricted to the blood vessels.

Magnetic susceptibility differences between the dHb-

containing compartments and the surrounding space gen-

erate magnetic field gradients across and near the com-

partment boundaries. Pulse sequences designed to be

sensitive to such susceptibility differences alter the signal

whenever the concentration of dHb changes.

Upon neural activation, an increase in dHb concentration

would be expected to enhance the field inhomogeneities,

reducing the BOLD signal. Yet, a few seconds after the

onset of stimulation the BOLD signal actually increases.

This enhancement reflects an increase in cerebral blood

flow (CBF) that overcompensates for the increased ox-

ygen consumption, and ultimately delivers an oversupply

of oxygenated blood [2,3]. BOLD imaging has been

successfully applied in humans [4–6] and was recently

applied in the non-human primate [7�–9�,10–12].

An example of BOLD imaging in the monkey is shown in

Figure 1a, a high resolution functional scan obtained from

1

CONEUR 94

www.current-opinion.com Current Opinion in Neurobiology 2003, 13:1–13

an anesthetized monkey in a moderately high field

(4.7 Tesla) scanner [11]. A gradient-recalled multi-shot,

multi-slice echo planar imaging method, optimized for

both the field and the subject-species, was used to acquire

volumes of 13 slices every 6 seconds while a polar-

transformed rotating checkerboard was presented to the

monkey through a fiber-optic system. The image shows

maps of the activation’s statistical significance (Z-scores

from a T-test) superimposed on T1-weighted anatomical

scans; whereby T1 refers to the time constant describing

the tissue’s magnetization re-growth. Significant clusters

of activation can be seen in the lateral geniculate nucleus

(LGN) and the striate cortex. The LGN of the monkey

is only about 6 mm across in the rostro-caudal direction

and approximately 5 mm in the dorso-ventral and medio-

lateral directions. The precise anatomical localization

of its activation is therefore important evidence for the

spatial specificity of fMRI in high magnetic fields.

Gradient echo sequences, like those used for the acquisi-

tion of the images in Figure 1, are sensitive to both small

and large vessels [13,14]. The significant contribution of

the large vessels can lead to erroneous mapping of the

activation site, as the flowing blood will generate BOLD

contrast downstream of the actual tissue with increased

metabolic activity. Thus, the extent of activation will

appear to be larger than it really is. The contribution of

large vessels depends on both the field strength and the

parameters of the pulse sequences [15–18], and they can

be de-emphasized in stronger magnetic fields, because

the strength of the BOLD that originates in the paren-

chyma (extravascular) increases more rapidly for small

Figure 1

dLGNV1

V1

STS

Lu

STSSTS

(a)

(d)

(b) (c)(c)(c)

STS

LuLuLu

Current Opinion in Neurobiology

Activation of LGN and visual cortex. (a) Volume imaging. The transverse section is parallel to the Frankfurt zero-plane. Measurements were made

on a vertical 4.7 T scanner having a 40 cm diameter bore (Biospec 47/40v, Bruker Medical Incorporation, Ettlingen, Germany). The system was

equipped with a 50 mT/m (180 ms rise time) actively shielded gradient coil of 26 cm inner diameter. The images were acquired using a field of view

(FOV) of 128 mm � 128 mm. T1-weighted, high resolution (256 � 256 matrix, 0.5 mm thickness) anatomical scans were obtained by using the 3D

modified driven equilibrium Fourier transform (3D-MDEFT) pulse sequence [82], with an echo time (TE) of 4 msec, repetition time (TR) of 14.9 msec, flip

angle (FA) of 20 deg, inversion time (t) of 800 msec, and 8 segments. Multi-slice fMRI (13 slices, 8 segments, 128 � 128 matrix, 1 mm thickness,

TE ¼ 20 msec, TR ¼ 740 msec, FA ¼ 50 deg, number of excitations ¼ 1) was carried out by the use of gradient-echo echoplanar imaging (GE-EPI).(b-d) Imaging with implanted radiofrequency coils. Intraosteally implanted RF coils were used to achieve sufficient signal- and contrast-to-noise ratios

for very high-resolution imaging that can be also combined with simultaneous electrophysiological measurements within the magnet. The parameters

of the EPI scan in b and c were FOV ¼ 32 � 32 mm2 on a matrix of 256 � 256 (0:125 � 0:125 mm2 resolution), slice thickness ¼ 0:720 mm,

TE/TR ¼ 20/750 msec, FA ¼ 30 deg, number of segments 8 or 16. The parameters of FLASH sequence in d were FOV 3:84 � 3:84, matrix 512 � 256,

nominal resolution 75 � 150 � 300 mm, TE 20 ms, TR 2040 ms, BW 25 kHz, 30 degree flip angle. Abbreviations dLGN, dorsal lateral geniculate

nucleus; Lu, lunate sulcus; STS, superior temporal sulcus; V1, striate cortex.

2 New technologies

Current Opinion in Neurobiology 2003, 13:1–13 www.current-opinion.com

vessels than it does for large ones (for review see [19]).

Thus, with a sufficiently high signal-to-noise ratio (SNR),

signals originating from the capillary bed are discernable

in strong magnetic fields and better reflect the site of

actual neural activation [20,21]. This property can be

exploited to obtain precise maps of areas defined on

the basis of functional architecture [22].

Specifically, cortex consists of specialized, separate

areas, which can be identified by their cyto- and

myelo-architecture, physiological properties and con-

nectivity (e.g. Felleman and Van Essen [23]). In the

visual system, the boundaries of some areas can be

determined by exploiting their retinotopic organization,

and several laboratories have already developed methods

for measuring visual field maps in the human brain (for a

review see Wandell [24]). Retinotopic maps were also

recently obtained in monkeys [25�], and were found to

register well with those derived from the same species

using anatomical and physiological techniques. In con-

trast to these techniques, however, non-invasive map-

ping of the retinotopic areas of individual monkeys is

likely to prove a valuable tool for several longitudinal

investigations, including studies of learning, plasticity

and reorganization.

When imaging of the entire brain volume is not the primary

interest, then small surface radiofrequency (RF) coils can

be used in either human or animal studies to obtain the

highest possible SNR in spatially resolved images. The

coils are often used as transceivers, namely both for excit-

ing the tissues and for receiving the RF signals transmitted

by them. Image quality and resolution can be further

improved by geometrically matching the coils to a specific

tissue region and, in animal experiments, by implanting

the coils in the body (for references see [26�,27]). The

small RF surface coil used in Figure 1b–d was implanted

within the bone (intraosteally). The achieved resolutions

in this case were 125 � 125 � 720 mm3, 125 � 125 � 720

mm3 and 75 � 150 � 300 mm3.

The contrast sensitivity of the image in Figure 1b is

sufficient to reveal the characteristic striation of the

primary visual cortex. The dark line shown by the white

arrow is the well-known, approximately 200 m thick Gen-

nari line formed by the axons of pyramidal and spiny

stellate cells contained in middle cortical layer (lamina

IVB). Figure 1c shows fMRI correlation coefficient maps

(in color) superimposed on the actual echoplanar (EPI)

(T�2-weighted) images. T�

2 refers to the transverse relaxa-

tion time constant in the presence of field inhomogene-

ities. The two most commonly varied MRI parameters are

the relaxation time constants T1 and T2. As mentioned

above, T1 is the tissue’s magnetization re-growth. T2, on

the other hand, refers to the so-called transverse relaxa-

tion in an ideal homogenous magnetic field. In actuality

the transverse relaxation is more rapid because of local

field inhomogeneities, including those within the imaged

tissue itself. It is then characterized by the decay constant

T�2 rather than T2. In the T�

2-weighted image of Figure 1c

both robust activation and good anatomical detail can be

discerned. Voxels of this size reflect the activity of as few

as 600–1200 cortical neurons, providing us with the

opportunity to study how neural networks are organized,

and how small cell assemblies contribute to the activation

patterns revealed in fMRI. Furthermore, they enable

direct comparisons between the imaging signals and those

signals obtained in intracortical recordings.

Intracortical recordings and fMRIThe successful application of BOLD fMRI in human or

animal brain research ultimately depends on a compre-

hensive understanding of the relationship between the

hemodynamic signal and the underlying neuronal activ-

ity. Functional MRI has already been combined with

optical imaging recordings of intrinsic signals [28] and

electroencephalography (EEG; e.g. [29]). However, the

first method also measures hemodynamic responses [30]

and thus can offer very little direct evidence of the

underlying neural activity, whereas EEG studies typically

suffer from poor spatial resolution and imprecise localiza-

tion of the electromagnetic field patterns associated with

neural current flow. Recently, combined intracortical

recordings and BOLD fMRI were successfully applied

in anesthetized and conscious monkeys. The BOLD

response was found to directly reflect an increase in

neural activity, best correlating with those electrical sig-

nals that are thought to represent synaptic inputs and

local intracortical processing [31��]. This conclusion was

derived from the detailed study of different components

of the digitized comprehensive mean extracellular field

potential (mEFP), which encompasses time-varying spa-

tial distributions of action potentials superimposed on

relatively slow varying field potentials.

Neural signals and their cellular origin

The mEFP represents the weighted sum of all sinks (e.g.

negativities caused by Naþ or Ca2þ moving from the

extracellular to the intracellular space) and sources (posi-

tivities) along multiple cells. If a microelectrode with a

small tip is placed close (within about 50 mm–100 mm) to

the soma or the axon of a neuron, then the measured

mEFP directly reports the spike traffic of that neuron, and

frequently that of its immediate neighbors as well [32].

The firing rate of such well isolated neurons has been the

usual measure for comparing neural activity to sensory

processing or behavior ever since the earliest develop-

ment of microelectrodes. A great deal has been learned

since then, and the single-electrode single-unit recording

technique still remains the method of choice in many

behavioral experiments with conscious animals. Like

any other method, however, it also has drawbacks, provid-

ing information mainly on single receptive fields but no

MR imaging in the non-human primate Logothetis 3


access to subthreshold integrative processes or the associa-

tional operations taking place at a given site. In addition,

it suffers from an element of bias towards certain cell types

(for a review and references see [33,34]). Spikes generated

by large neurons remain above the noise level over a greater

distance from the cell than spikes from small neurons, so

microelectrodes sample their somas or axons preferentially,

a prediction supported by experimental work.

When the activity of a single neuron is not the primary

concern of an investigation, microelectrodes of lower

impedance and appropriate tip geometry can be con-

structed that are less inundated by spikes and capture

the totality of the potentials in a given region. The mEFP

recorded under these conditions is related to both local

integrative processes (dendritic events) and spikes gen-

erated by several hundred neurons.

A large number of experiments have presented data

that indicate that the two different processes can be

segregated by subjecting the mEFP to frequency-band

separation. A high-pass filter cutoff of approximately

300–400 Hertz is used in most recordings to obtain

multiple-unit spiking activity (MUA), and a low-pass

filter cutoff of about 200–300 Hz to obtain the so-called

local field potentials (LFPs).

Fast and slow components of the mean extracellular

field potential

The MUAs are a weighted sum of the extracellular action

potentials of all neurons within a sphere of approximately

140–300 mm radius, with the electrode at its center.

Spikes produced by the synchronous firings of many cells

can, in principle, be enhanced by summation and thus

detected over larger distances [33,34].

The LFPs are slow fluctuations reflecting cooperative

activity in neural populations. Until recently, these sig-

nals were thought to represent exclusively synaptic

events. Evidence for this came from combined EEG

and intracortical recordings that showed that the slow

wave activity in the EEG is largely independent of

neuronal spiking [33,34]. They also showed that, unlike

the multiunit activity, the magnitude of the slow field

fluctuations is not correlated with cell size, but instead

reflects the extent and geometry of dendrites in each

recording site. The pyramidal cells, with their apical den-

drites running parallel to each other and perpendicular

to the pial surface, form an ideal open field arrangement

and contribute maximally to both the macroscopically

measured EEG and the LFPs. The LFPs are the weighted

average of synchronized dendro–somatic components of

the synaptic signals originating from within 0.5–3 mm of

the electrode tip (e.g. [35]).

Today we know that LFPs represent both synaptic

events and other types of slow activity, including voltage-

dependent membrane oscillations (e.g. [36]) and spike

afterpotentials. The soma–dendritic spikes in the neurons

of the central nervous system (CNS) may be followed by

afterpotentials, a brief delayed depolarization, the after-

depolarization, and a longer lasting after-hyperpolariza-

tion, which are thought to play an important role in the

control of excitation-to-frequency transduction (e.g.

[37,38]). Afterpotentials, which were shown to be gener-

ated by calcium-activated potassium currents (e.g.

[39,40]), have a duration in the order of 10’s of milli-

seconds and most likely contribute to the generation of

the LFP signals (e.g. [41]).

BOLD reflects the input and local processing of a

studied brain area

To understand the contribution of such different cellular

events to the generation of the BOLD signal, we exam-

ined the correlation of LFPs, MUA and single neuron

activity with the hemodynamic response in a large num-

ber of combined imaging-physiology experiments in the

striate and extrastriate areas of the monkey visual system

[31��]. Figure 2 shows the neural and BOLD signals

recorded simultaneously from an alert, fixating monkey.

At first sight, they all seemed to be correlated with the

BOLD response, although increases in the LFP range

were consistently greater in both spectral power and

reliability (SNR). Furthermore, correlation analysis

showed that LFPs are better predictors of the BOLD

response than multiunit spiking.

The relation of the two types of signals to BOLD was best

appreciated, however, in cases of a complete dissociation

between slow waves and spikes. Sites exhibiting a strong

neural response adaptation were characterized by MUA

that returned to the baseline a few seconds after stimulus

onset, and local field potentials that remained elevated for

the entire duration of the visual stimulus. In such cases,

LFPs were the only neural signal to be associated with the

BOLD response. This striking result suggests that spikes

are a ‘fortuitous’ predictor of the BOLD signal, simply

because the firing of neurons usually happens to correlate

with the LFPs. A similar dissociation between spikes and

CBF has also been demonstrated in microstimulation

studies in the cerebellar cortex (for a review of this work

see Lauritzen and Gold [42]).

The differential contribution of LFPs and MUA to the

BOLD response can also be demonstrated by neurophar-

macological manipulations that permit the selective mod-

ulation of interneuronal and pyramidal activity. Figure 3

shows the effects of intracortical serotonin injections into

the primary visual cortex of the monkey during simulta-

neous acquisition of BOLD and neural responses (N

Logothetis, unpublished data). Using a triple pipette

(electrode, saline and the neuromodulator) 20 micro-liters

of 0.01 Mol 5HT (5-hydroxytryptamine hydrochloride;

serotonin) were injected over a period of 10 minutes. A

4 New technologies


couple of minutes after the injection, a profound suppres-

sion of the MUA was observed (Figure 3a). The LFP

signal showed a slight increase and returned to baseline

within a few minutes. However, no significant change was

discernible in the BOLD response. Spectrograms

obtained before and after the 5HT injection during visual

stimulation showed that the stimulus-induced spikes

were entirely eliminated, whereas LFP activity was mod-

erately increased (Figure 3b). BOLD response to the

stimuli was unaltered, indicating once again the possibi-

lity of a total dissociation between spiking activity and

hemodynamic responses. On the basis of all of these

dissociations, we conclude that the LFP signal is the

key variable for the BOLD response.

Taken together, these results suggest that the BOLD

signal mainly reflects the incoming specific or association

inputs into an area, and the processing of this input

information by the local cortical circuitry (including exci-

tatory and inhibitory interneurons). Usually, the incoming

subcortical or cortical input to an area will generate the

kind of output activity that is typically measured in

intracortical single-unit recordings, and the recorded

spike rate will be correlated to the BOLD signal. If it

does not, perhaps because the activity of projection

neurons is shunted by concurrent modulatory input,

hemodynamic responses will still be generated, but the

spiking activity measured with microelectrodes will — in

such cases — be a poor predictor of the hemodynamic

response. Several experiments demonstrate the plausi-

bility of this explanation.

In a recent study, Tolias and co-workers [43�] used an

adaptation technique to study the brain areas that process

motion information [44]. They repeatedly imaged a mon-

key’s brain while the animal viewed continuous motion in

a single unchanging direction. Under these conditions,

the BOLD response adapts. When the direction of

Figure 2

0

Mag

nitu

de in

AD

C u

nits

Spike raster

5000

–50000 20 40 60 80 100 120

–1

–2

–3

–4

4

3

2

1

6

5

0SD

uni

tsS

pike

cou

nt

20 40 60 80 100 1200

Time in seconds

Time in seconds

(a)

(b)

(c)

(d)


Imaging and physiology in the alert, behaving monkey. (a,b) Electrode position and BOLD activation of striate cortex. (c) Neural and hemodynamic

responses from the region of interest (green circle in b). The black trace shows the comprehensive neural signal, that is, the signal of 50 mHz to

3 kHz band-width, which includes LFP, MUA and single spikes. Superimposed is the BOLD activation (red) for the region of interest around the

electrode tip. (d) Spike raster and peristimulus time histogram (PSTH) showing the activity of a small assembly of neurons (2–3 cells). Each raster dot

signifies the occurrence of an action potential. Each bin of the PSTH is equal to the repetition time of an image (250 msec).



motion reverses abruptly, the measured activity imme-

diately shows a partial recovery toward the initial activity

levels, or rebound. The extent of this rebound was

considered to be an index of the average directional

selectivity of neurons in any given activated area. The

results confirmed previous electrophysiological studies

that had identified a distributed network of visual areas

(V1, V2, V3, V5/MT) in the monkey that process informa-

tion about the direction of visual motion.

Surprisingly, however, strong activation was also observed

in area V4, which is only weakly involved in motion

processing (e.g. [45]). Such a discrepancy can be

explained on the basis of the arguments developed ear-

lier. Areas V4 and MT are interconnected [46]. Although

they process separate stimulus properties, each area may

influence the sensitivity of the others by providing some

kind of ‘modulatory’ input, which in itself is insufficient

to drive the pyramidal cells recorded in a typical electro-

physiology experiment. In such cases, BOLD fMRI will

reveal significant activation of an area whose output may

be only indirectly related to the stimulus or task, provid-

ing results that do not match those of neurophysiology

experiments. Similarly, attentional effects on the neurons

of striate cortex have been very difficult to measure in

monkey electrophysiology experiments [47,48]. Yet for

similar tasks strong attentional effects are readily measur-

able with fMRI in the human primary visual cortex (V1)

(e.g. [49]).

In summary, combined physiology and fMRI experiments

suggest that the BOLD signal primarily reflects the input

of neuronal information to the relevant area of the brain

and its processing there, rather than the output signals

transmitted to other regions of the brain by the principal

neurons; the cells that are most easily accessible in single

cell recordings in the behaving animal. These findings are

in agreement with several observations from experiments

on the brain’s energy metabolism and the mechanisms

that couple neural activation with metabolic demands.

Figure 3

0 10 20 30 40 0 10 20 30 40Time in seconds

BOLDresponse

Time in seconds

SD

uni

ts

MUA

LFP

Before injection (control) 10 min after injection

0 200 400 600 800 1000

–4

–2

0

2

4

–8

–4

0

4

8

BOLD response

Neural response

LFP

MUA

(a)

LFPMUA

0

10

20

30

20

0

5

10

15

0.04

0

0.01

0.02

0.03

SD

uni

tsS

pike

den

sity

(b)

(c)

Log

freq

uenc

y in

Hz

102

103

102

103


Effects of serotonin on the neural and hemodynamic responses. (a) Changes in the multiunit activity, local field potentials and hemodynamic

responses during the administration of serotonin. Using a triple pipette (electrode, saline and the neuromodulator) 20 micro-liters of 0.01 Mol 5HT wasinjected over a period of 10 min, as signified by the red ramp on the x-axis of the plot. A couple of minutes after the injection a profound suppression of

the MUA was observed. The LFP signal showed a slight increase, returning to baseline in a few minutes. The lower traces show the hemodynamic

response for the site of injection (pink) and a control site on the other hemisphere. No significant change was discernible in the BOLD response.

(b) Spectrograms before and after the 5HT injection. Note the MUA activity in the control spectrogram. The activity in the same frequency band is

eliminated after the injection. (c) LFP, MUA and hemodynamic responses before and after the 5HT injection. The two lower plots show the activity

of a small number of isolated neurons.

6 New technologies


Selective neurovascular and neurometabolic coupling

Basics of brain energy metabolism

The brain’s demand for substrate requires adequate

delivery of oxygen and glucose through elaborate

mechanisms regulating CBF. Early experimental evi-

dence that suggested a regional coupling between these

mechanisms and neural activity [50] was later verified by

means of the deoxyglucose autoradiographic technique

that enabled spatially resolved measurements of glucose

metabolism [51].

In humans, the first quantitative measurements of regio-

nal brain blood flow and oxygen consumption were per-

formed using radiotracer techniques, which were

followed by the introduction of positron emission tomo-

graphy (PET; for a historical review see Raichle [52]).

PET showed that maps of activated brain regions can be

produced by detecting the indirect effects of neural

activity on variables, such as cerebral blood flow [53],

cerebral blood volume (CBV) [2], and blood oxygenation

[2,3]. At the same time, optical imaging of intrinsic signals

demonstrated the spectacular precision of neurovascular

coupling (the strong correlation between neural and vas-

cular changes) by constructing detailed maps of cortical

microarchitecture in both the anesthetized and the alert

animal [30].

Although the existence of a regional coupling between

neural activity, metabolism, and hemodynamic changes is

now established, the nature of the links among these

processes remain largely unknown. One important ques-

tion is whether changes in CBF are driven directly by

energy demand or by neurotransmitter related signaling

(see review by Attwell and Iadecola [54��]). If energy

demand is indeed the trigger for CBF changes, then

which are the cellular processes and sites that dominate

the energy consumption? The structural and functional

organization of the neuro-vascular system provides some

insights into these questions.

Structural neurovascular coupling

The density of the vascular network in a given region

largely correlates with its average activity. Most impor-

tantly, however, the spatial correlations reported by a

large number of investigators have been mainly between

vascular density and the number of synapses rather than

the number of neurons. For instance, on the basis of its

density the human cortical vascular network can be sub-

divided into four layers parallel to the surface [55]. These

layers systematically overlap with certain portions of the

cytoarchitectonically defined Brodmann’s laminae. Nota-

bly, the first Duvernoy layer, which consists of vessels

oriented approximately parallel to neural fibers, is entirely

within the lower part of the molecular layer (Layer I),

which in the rodent has the lowest concentration of cell

bodies and highest density of synapses [56]. Similarly, in

the primate, this layer has the lowest concentration of

neurons, the highest concentration of astrocytes, and a

high density of synapses [57]. Its vascularization is there-

fore suggestive of the contribution of these cellular types

and components in the hemodynamic response.

The spatial correlation of blood vessels and neural com-

ponents reported for the human brain were later con-

firmed in animal studies [58]. The vascularization was

found to be lowest in Lamina I and highest in IVc, with an

average IVc/I ratio (across animals) of approximately 3:1.

Interestingly, the IVc/I ratio of synaptic density in the

striate cortex of macaque is 2.43:1, that of astrocytes is

1.2:1, and that of neurons is 78.8:1. Assuming some

similarity in the distribution of neurons, synapses, and

astrocytes between squirrel monkeys and macaques, the

recent data would also support the notion that vascular

density is correlated with the density of perisynaptic

elements rather than that of neuronal somata.

Functional neurovascular coupling

The cerebral metabolic rate (CMR), is commonly

expressed in terms of oxygen consumption (CMRO2)

because glucose metabolism is about 90% aerobic and

therefore parallels oxygen consumption (e.g. [59]). The

CMRO2 varies with neuronal shape, size and firing

properties, whereby large projection neurons, which

maintain energy-consuming processes such as ion pump-

ing over a large membrane surface, may have larger

energy requirements.

High-resolution autoradiography studies showed that

the uptake of deoxy-glucose is higher in the neuropil,

with its axonal, dendritic, and astrocytic processes, than

in the cell bodies. These results were confirmed by

others, which suggests that synaptic activity dominates

energy consumption (for example, Sokoloff [60]). They

are further corroborated by microstimulation experi-

ments, in which the increase in glucose utilization was

assessed during orthodromic and antidromic stimulation,

the former activating both pre- and post-synaptic term-

inals and the latter activating only postsynaptic term-

inals. Increases were only observed during orthodromic

stimulation ([61–63]; for a review see Jueptner and

Weiller [64]), which suggests that coupling occurs pri-

marily between energy metabolism and activity in the

presynaptic terminals.

The notion that perisynaptic activity (synaptic events and

restoration of gradients disturbed by postsynaptic poten-

tials) consumes a greater proportion of the available

energy receives support from the metabolic requirements

of other cellular components of the brain. Neurons are not

the only elements contributing to the energy metabolism

of the brain; glia and vascular endothelial cells do so as

well. In fact, research suggests a tightly regulated glucose

metabolism in all brain cell types (for detailed references

see Volterra and co-workers [65]).



An interesting case is the glial cell known as the astrocyte,

which forms a massive bridge system between neurons

and the brain’s vasculature. The structural and functional

characteristics of astrocytes make them ideal viaducts

between the neuropil and the intraparenchymal capil-

laries. It has been suggested that for each synaptically

released glutamate molecule taken up by an astrocyte

(with two to three Naþ ions), one glucose molecule enters

the same astrocyte, two ATP molecules are produced

through glycolysis, and two lactate molecules are released

and consumed by neurons to yield 18 ATPs through

oxidative phosphorylation. In other words, synaptic activ-

ity may be tightly coupled to glucose uptake through the

neuron–astrocyte system (e.g. [66,67]).

The neuron–astrocyte system is compatible with the

notion that neuronal signals, which are mediated by fast

neurotransmitters or by ions or molecules that are tran-

siently released in the extracellular space upon neural

excitation, can trigger receptor-mediated glycogenolysis

in astrocytes in anticipation of or at least parallel to

regional activation (for a review see [68]).

Atwell and Laughlin [69��] proposed that the greater part

of energy expenditure is attributable to the postsynaptic

effects of glutamate (about 34% of the energy in rodents

and 74% in humans is attributable to excitatory postsy-

naptic currents). They formed this proposal on the basis

of computations of the number of vesicles released per

action potential, the number of postsynaptic receptors

activated per vesicle released, the metabolic conse-

quences of activating a single receptor and changing

ion fluxes, and neurotransmitter recycling [54��,69��].Yet others have suggested that most of the brain’s energy

is consumed for the generation and propagation of action

potentials [70�]; according to these estimates, the cost of

an action potential would permit only 1% of the neurons

in any area to be active concurrently. This prediction

seems to be at odds with experimental and theoretical

studies that suggest that the cortical neurons operate on a

high-input regime with a well-controlled balance be-

tween excitation and inhibition [71–74].

In summary, the neurovascular coupling may be med-

iated by fast neurotransmitters, other signaling molecules

(nitric oxide, prostaglandins etc.) or by any other mechan-

ism able to sense the increasing or decreasing energy

demand rapidly enough. In the former case, research

suggests that a glutamate-evoked Ca2þ influx in postsy-

naptic neurons activates the signaling-relevant molecules

that in turn produce a vasodilation. This vasodilation

reflects both the activity of neurons presynaptic to the

cells that have released the signaling molecules and the

level of depolarization of the postsynaptic cell, which will

typically alter the Mg2þ-block of NMDA receptors and

the resulting Ca2þ influx [54��]. In the latter case, over-

whelming evidence suggests that energy demand will be

driven to a lesser degree by the need to recycle neuro-

transmitters and to a greater degree by the processes of

restoring perturbed gradients through postsynaptic cur-

rents. Both pre- and postsynaptic currents are dominant

elements of the local field potentials, which — as men-

tioned above — were in fact found to correlate best with

the hemodynamic changes in the cerebral and cerebellar

cortex.

The study of networks with MRIConnectivity studies with paramagnetic tracers

Neuroanatomical cortico–cortical and cortico–subcortical

connections have been examined mainly by using degen-

eration methods and anterograde and retrograde tracer

techniques (e.g. [23,75]). Although such studies have

demonstrated the value of the information gained from

the investigation of the topographic connections among

different brain areas, they do require fixed processed

tissue for data analysis, and therefore cannot be applied

to an animal participating in longitudinal studies, in which

consecutive studies examining an entire circuit are car-

ried out in the same subject.

MRI visible tracers that are infused into a specific brain

region and are transported anterogradely or retrogradely

along axons may enable us to study neuronal connectivity

in the living animal. Such paramagnetic tracer studies

may also be used to validate and further develop non-

invasive fiber tracking techniques, such as diffusion ten-

sor MRI, which permits the study of connectivity even in

the human brain.

Manganese (Mn2þ) is an interesting example of an MRI-

visible contrast agent. The axonal transport of its radio-

active isotope (54Mn2þ) was first studied using histologi-

cal methods [76,77]. Although these studies were carried

out with the goal of understanding the regional specificity

of Mn2þ distribution, they indicated the usefulness of

Mn2þ as an anterograde neuronal tract tracer. Mn2þ

distribution and transport has been also studied with

MRI in rats and mice [78,79]. Injection of manganese

(manganum chloride, MnCl2) in to a nostril or an eye

yields a clear signal enhancement in the olfactory and

visual pathways [78,79]. Furthermore, the possibility that

the transport of manganese may pass across synapses was

suggested by some studies [77,78]. Pautler et al. [78]

indicated that Mn2þ must have traversed a synapse to

explain the enhancements detected in the olfactory cor-

tex of the mouse following the injection of its olfactory

bulb. By contrast, Watanabe and co-workers [79] reported

that the signal enhancement they observed in their rat

study was confined to regions known to receive direct

projections from the retina, and concluded that it did not

constitute evidence for trans-synaptic crossing of Mn2þ.

An example of trans-synaptic transfer of manganese is

displayed in Figure 4 (N Logothetis, unpublished data).

8 New technologies


Following intravitreal (into the vitreal body of the eye

ball) manganese injection a series of anatomical scans

were acquired that illustrate the transfer of the sub-

stance along the retinogeniculate–striate pathway. Sig-

nal enhancement is clear along the optic nerve and

tract, the dorsal LGN, superior colliculus, optical radia-

tion and striate cortex. Similar results were obtained in

a recent study that provided a detailed account of both

the specificity and the trans-synaptic transfer of man-

ganese in the basal ganglia of the monkey. Injections

were made into the striatum [80��]. Its projections were

confirmed histologically by injecting wheat germ ag-

glutinin conjugated to horseradish peroxidase (WGA-

HRP) into the same sites that had previously been

injected with MnCl2. The size and location of the

projection foci in the striatal targets were comparable

to those found in both the magnetic resonance and

the histology images. By injecting WGA-HRP at the

same sites as MnCl2, we also confirmed for each animal

the absence of a direct connection from the injection

sites to various brain structures (e.g. thalamic nuclei).

In this study, manganese was actually found in several

structures receiving no direct projections from the

injected sites.

MR imaging and electrical microstimulation

Our knowledge of connectivity and functional organiza-

tion could profit a great deal from the combination of MRI

with electrical microstimulation. Microstimulation is

established as an important neurobiological tool for the

study of areal representation and the functional properties

of CNS output structures. A new method was recently

developed that combines this technique with fMRI for

the detailed study of neural connectivity in the alive

animal. Specially constructed microelectrodes were used

to directly stimulate a selected subcortical or cortical area

while simultaneously measuring changes in brain activity,

which was indexed by the BOLD signal [81]. The exact

location of the stimulation site was determined by using

anatomical scans, as well as by the study of the physio-

logical properties of neurons. Electrical stimulation was

delivered using a biphasic pulse generator attached to a

constant-current stimulus isolation unit. The compensa-

tion circuit, designed to minimize interference generated

Figure 4

OT

SC

dLGN

ONOT

OR

(a)

(b)

(c) (e)

(d) (f)


dLGN

Intravitreal manganese injection. 0.049 ml of MnCl2 1.2M concentration was injected over a 5 min time period. Following the injection a series of

anatomical scans were acquired (MDEFT, TE ¼ 4 ms, TR ¼ 21:3 ms, FOV ¼ 12:8 � 12:8 � 8 cm3, matrix ¼ 256 � 256 � 160, voxel size ¼ 0:5 mmisotropic, segments ¼ 4). (a) Injection site (right eye). (b) Lateral (upper) and top (lower) view of the retinogeniculate–striate pathway revealed with

the Mn2þ transfer. (c) Horizontal slice at the level of the optic tract and dLGN showing signal enhancement. (d) Coronal slices with the tracer in

the lateral geniculate bodies. (e) Chiasm and superior colliculus. (f) Nissl stained histological section showing dLGN for comparison with d.

Abbreviations: ON, optic nerve; OR, optical radiation; OT, optic tract; SC, superior colliculus.



by the switching gradients during recording, was always

active, minimizing the gradient-induced currents in the

range of the stimulation current. Local microstimulation

of the striate cortex yielded both local BOLD signals and

activation of areas V2, V3 and MT. Microstimulation of

dorsal LGN resulted in the activation of both striate

cortex and areas V2, V3, and MT. Figure 5 shows an

example of V1 activation after microstimulation of LGN.

The findings show that microstimulation combined with

fMRI can be an exquisite tool for finding and studying

target areas of electrophysiological interest.

ConclusionsThe suitability of MRI for functional brain mapping has

become firmly established over the past decade or so.

BOLD fMRI has been successfully implemented in

awake human subjects as well as in animals such as rats,

cats, and monkeys. The use of high magnetic fields

improves both signal specificity and spatial resolution.

MRI studies, in which voxels may contain as few as 600–

800 cortical neurons, can help us to understand how

neural networks are organized, and how small cell assem-

blies contribute to the activation patterns revealed in

fMRI. The combination of this technique with electro-

physiology has fully confirmed the longstanding assump-

tion that the regional activations measured in MR

neuroimaging do indeed reflect local increases in neural

activity. In addition, it has been demonstrated that fMRI

responses mostly reflect the input of a given cortical area

and its local intracortical processing, including the activity

of excitatory and inhibitory interneurons. Finally, MRI

visible tracers and microstimulation appear to be ideal for

the study of connectivity in the living animal.

AcknowledgementsThis research was supported by the Max Planck Society. I would like tothank my collaborators M Augath, Y Murayama, A Oeltermann and J Paulsfor their contribution in the experiments, S Weber for machining work,and D Blaurock for proof reading the final version of the text.

References and recommended readingPapers of particular interest, published within the annual period ofreview, have been highlighted as:

� of special interest��of outstanding interest

1. Ogawa S, Lee TM, Kay AR, Tank DW: Brain magnetic resonanceimaging with contrast dependent on blood oxygenation.Proc Natl Acad Sci USA 1990, 87:9868-9872.

2. Fox PT, Raichle ME: Focal physiological uncoupling of cerebralblood flow and oxidative metabolism during somatosensorystimulation in human subjects. Proc Natl Acad Sci USA 1986,83:1140-1144.

3. Fox PT, Raichle ME, Mintun MA, Dence C: Nonoxidative glucoseconsumption during focal physiologic neural activity. Science1988, 241:462-464.

4. Bandettini PA, Wong EC, Hinks RS, Tikofsky RS, Hyde JS: Timecourse EPI of human brain function during task activation.Magn Reson Med 1992, 25:390-397.

5. Kwong KK, Belliveau JW, Chesler DA, Goldberg IE, Weisskoff RM,Poncelet BP, Kennedy DN, Hoppel BE, Cohen MS, Turner R:Dynamic magnetic resonance imaging of human brain activityduring primary sensory stimulation. Proc Natl Acad Sci USA1992, 89:5675-5679.

6. Ogawa S, Tank DW, Menon R, Ellermann JM, Kim SG, Merkle H,Ugurbil K: Intrinsic signal changes accompanying sensorystimulation: functional brain mapping with magneticresonance imaging. Proc Natl Acad Sci USA 1992, 89:5951-5955.

7.�

Kourtzi Z, Tolias AS, Altmann CF, Augath M, Logothetis NK:Integration of local features into global shapes: monkey andhuman fMRI studies. Neuron 2003, 37:333-346.

In this study, an adaptation paradigm was used, in which stimulusselectivity was deduced by changes in the course of adaptation of apattern of randomly oriented elements. The authors report strongerincreases of activity when local orientation changes in the adaptingstimulus resulted in a co-linear shape rather than a different randompattern. This selectivity to co-linear shapes was observed not only inhigher visual areas that are implicated in shape processing but alsoin early visual areas in which selectivity depended on the receptive fieldsize.

8.�

Rainer G, Augath M, Trinath T, Logothetis NK: The effect of imagescrambling on visual cortical BOLD activity in the anesthetizedmonkey. Neuroimage 2002, 16:607-616.

This is a study of BOLD signal changes that are associated with scram-bling natural images into different numbers of segments in visuallymodulated regions of the monkey brain. The authors suggest that theBOLD signal might reflect average activation of local orientation detectorsin V1, and sensitivity to more global object representations in higher visualareas. Scrambling causes substantial changes to the spatial frequencycontent of images.

9.�

Sereno ME, Trinath T, Augath M, Logothetis NK:Three-dimensional shape representation in monkey cortex.Neuron 2002, 33:635-652.

The authors use fMRI in anesthetized monkeys to study how the primatevisual system constructs representations of 3-D shapes from a variety ofcues. Computer-generated 3-D objects defined by shading, randomdots, structure elements or silhouettes are presented either staticallyor dynamically (rotating). Results suggest that 3-D shape representations

Figure 5

75uA/500Hz

25uA/500Hz

150uA/500Hz

30uA/500Hz


Imaging and electrical microstimulation. The top left panel shows

activation of LGN and striate cortex following visual stimulation. The top

right panel shows activation of striate cortex after electrical

microstimulation of LGN. The two bottom panels indicate the electrode

position in a coronal (left) and parasaggital (right) view.

10 New technologies


are highly localized despite being widely distributed in occipital, temporal,parietal and frontal cortices, and may involve common brain regionsregardless of the shape cue.

10. Vanduffel W, Fize D, Peuskens H, Denys K, Sunaert S, Todd JT,Orban GA: Extracting 3D from motion: differences in human andmonkey intraparietal cortex. Science 2002, 298:413-415.

11. Logothetis NK, Guggenberger H, Peled S, Pauls J: Functionalimaging of the monkey brain. Nat Neurosci 1999, 2:555-562.

12. Dubowitz DJ, Chen DY, Atkinson DJ, Scadeng M, Martinez A,Andersen MB, Andersen RA, Bradley WGJR: Direct comparisonof visual cortex activation in human and non-human primatesusing functional magnetic resonance imaging. J NeurosciMethods 2001, 107:71-80.

13. Ogawa S, Menon RS, Tank DW, Kim S-G, Merkle H, Ellermann JM,Ugurbil K: Functional brain mapping by blood oxygenation leveldependent contrast magnetic resonance imaging. Biophys J1993, 64:800-812.

14. Weisskoff RM, Zuo CS, Boxerman JL, Rosen BR: Microscopicsusceptibility variation and transverse relaxation: theory andexperiment. Magn Reson Med 1994, 31:601-610.

15. Menon RS, Ogawa S, Tank DW, Ugurbil K: 4 Tesla gradientrecalled echo characteristics of photic stimulation inducedsignal changes in the human primary visual cortex. Magn ResonMed 1993, 30:380-386.

16. Boxerman JL, Bandettini PA, Kwong KK, Baker JR, Davis TL,Rosen BR, Weisskoff RM: The intravascular contribution to fMRIsignal change: Monte Carlo modeling and diffusion-weightedstudies in vivo. Magn Reson Med 1995, 34:4-10.

17. Hoogenraad FG, Pouwels PJ, Hofman MB, Reichenbach JR,Sprenger M, Haacke EM: Quantitative differentiationbetween BOLD models in fMRI. Magn Reson Med 2001,45:233-246.

18. Zhong J, Kennan RP, Fulbright RK, Gore JC: Quantification ofintravascular and extravascular contributions to BOLD effectsinduced by alteration in oxygenation or intravascular contrastagents. Magn Reson Med 1998, 40:526-536.

19. Gati JS, Menon RS, Rutt BK: Field strength dependence offunctional MRI signals. In Functional MRI. Edited by Moonen CT,Bandettini PA. Berlin: Springer; 2000:277-282.

20. Gati JS, Menon RS, Ugurbil K, Rutt BK: Experimentaldetermination of the BOLD field strength dependence invessels and tissue. Magn Reson Med 1997, 38:296-302.

21. Duong TQ, Yacoub E, Adriany G, Hu X, Ugurbil K, Kim SG:Microvascular BOLD contribution at 4 and 7 T in the humanbrain: gradient echo and spin echo fMRI with suppression ofblood effects. Magn Reson Med 2003, 49:1019-1027.

22. Ugurbil K, Toth L, Kim DS: How accurate is magnetic resonanceimaging of brain function? Trends Neurosci 2003, 26:108-114.

23. Felleman DJ, Van Essen DC: Distributed hierarchical processingin the primate cerebral cortex. Cereb Cortex 1991, 1:1-47.

24. Wandell BA: Computational neuroimaging of human visualcortex. Annu Rev Neurosci 1999, 22:145-173.

25.�

Brewer AA, Press W, Logothetis NK, Wandell B: Visual areasin Macaque cortex measured using functional MRI.J Neurosci 2002, 22:10416-10426.

The authors used functional MRI signals to locate the positions andmeasure the topography of visual areas in an anaesthetized macaquemonkey. Strong fMRI signals were recorded in striate cortex and theareas V1, V2, V4 and MT revealing the retinotopical organization of theseareas. Somewhat weaker signals were observed in other cortical regions,and these provide some insight into the overall organization of visualcortex in more anterior regions of cortex. On the basis of concordance ofthe fMRI signals with the expected properties of visual areas and topo-graphy, the authors conclude that this experimental protocol producesfMRI signals that measure local neural activity. These methods alsoprovide a good foundation for further studies of topography and reorga-nization in the early cortical pathways.

26.�

Logothetis N, Merkle H, Augath M, Trinath T, Ugurbil K: Ultra-highresolution fMRI in monkeys with implanted RF coils. Neuron2002, 35:227-242.

The authors present the first application of implanted RF coils in themonkey. Voxel sizes as small as 0.016 ml with high signal-to-noise andcontrast-to-noise ratios were obtained, revealing both structural andfunctional cortical architecture in great detail. Spatial specificity wasdemonstrated by the lamina-specific activation in experiments compar-ing responses to moving stimuli with those elicited by flickering stimuli.The implications of this technique for combined fMRI/electrophysiologyexperiments and its limitations in terms of spatial coverage are discussed.

27. Logothetis NK: The neural basis of the blood-oxygen-level-dependent functional magnetic resonance imaging signal.Philos Trans R Soc Lond B Biol Sci 2002, 357:1003-1037.

28. Hess A, Stiller D, Kaulisch T, Heil P, Scheich H: New insights intothe hemodynamic blood oxygenation level-dependentresponse through combination of functional magneticresonance imaging and optical recording in gerbil barrelcortex. J Neurosci 2000, 20:3328-3338.

29. Menon V, Ford JM, Lim KO, Glover GH, Pfefferbaum A: Combinedevent-related fMRI and EEG evidence for temporal- parietalcortex activation during target detection. Neuroreport 1997,8:3029-3037.

30. Bonhoeffer T, Grinvald A: Optical imaging based on intrinsicsignals. In Brain Mapping, The Methods. Edited by Toga AW,Mazziotta JC. New York: Academic Press; 1996:55-97.

31.��

Logothetis NK, Pauls J, Augath M, Trinath T, Oeltermann A:Neurophysiological investigation of the basis of the fMRIsignal. Nature 2001, 412:150-157.

The authors present the first simultaneous physiological and fMRI record-ings. Local field potentials, single- and multi-unit spiking activity werecompared with high spatio-temporal BOLD fMRI responses from thevisual cortex of monkeys. The largest magnitude changes were observedin LFPs, which at recording sites characterized by transient responseswere the only signal that significantly correlated with the hemodynamicresponse. Linear systems analysis on a trial-by-trial basis showed that theimpulse response of the neuro-vascular system is animal- and site-specific, and that LFPs yield a better estimate of BOLD than the multiunitresponses. These findings suggest that the BOLD contrast mechanismreflects the input and intra-cortical processing of a given area, rather thanits spiking output.

32. Henze DA, Borhegyi Z, Csicsvari J, Mamiya A, Harris KD,Buzsaki G: Intracellular features predicted by extracellularrecordings in the hippocampus in vivo. J Neurophysiol 2000,84:390-400.

33. Logothetis NK: The underpinnings of the BOLD functionalmagnetic resonance imaging signal. J Neurosci 2003,23:3963-3971.

34. Logothetis NK: On the neural basis of the BOLD fMRI signal.Philos Trans R Soc Lond A 2002, 357:1003-1037.

35. Mitzdorf U: Properties of the evoked potential generators:current source-density analysis of visually evoked potentials inthe cat cortex. Int J Neurosci 1987, 33:33-59.

36. Kamondi A, Acsady L, Wang XJ, Buzsaki G: Theta oscillations insomata and dendrites of hippocampal pyramidal cells in vivo:activity-dependent phase-precession of action potentials.Hippocampus 1998, 8:244-261.

37. Harada Y, Takahashi T: The calcium component of the actionpotential in spinal motoneurons of the rat. J Physiol 1983,335:89-100.

38. Gustafsson B: Afterpotentials and transduction properties indifferent types of central neurones. Arch Ital Biol 1984,122:17-30.

39. Higashi H, Tanaka E, Inokuchi H, Nishi S: Ionic mechanismsunderlying the depolarizing and hyperpolarizing afterpotentialsof single spike in guinea-pig cingulate cortical neurons.Neuroscience 1993, 55:129-138.

40. Kobayashi M, Inoue T, Matsuo R, Masuda Y, Hidaka O, Kang Y,Morimoto T: Role of calcium conductances on spikeafterpotentials in rat trigeminal motoneurons. J Neurophysiol1997, 77:3273-3283.

41. Buzsaki G, Bickford RG, Ponomareff G, Thal LJ, Mandel R,Gage FH: Nucleus basalis and thalamic control of neocorticalactivity in the freely moving rat. J Neurosci 1988, 8:4007-4026.



42. Lauritzen M, Gold L: Brain function and neurophysiologicalcorrelates of signals used in functional neuroimaging.J Neurosci 2003, 23:3972-3980.

43.�

Tolias AS, Smirnakis SM, Augath MA, Trinath T, Logothetis NK:Motion processing in the macaque: revisited with functionalmagnetic resonance imaging. J Neurosci 2001, 21:8594-8601.

In this study, an adaptation paradigm was used to study the activity ofstriate and early extrastriate cortex with fMRI. The representation ofdirectional selectivity was found to be different from estimates formedon the basis of single neuron recordings.

44. Grill-Spector K, Kushnir T, Edelman S, Avidan G, Itzchak Y,Malach R: Differential processing of objects under variousviewing conditions in the human lateral occipital complex.Neuron 1999, 24:187-203.

45. Desimone R, Schein SJ: Visual properties of neurons in area V4of the macaque: sensitivity to stimulus form. J Neurophysiol1987, 57:835-868.

46. Felleman DJ, Van Essen DC: The connections of area V4 ofmacaque extrastriate cortex. Abstr Soc Neurosci 1983, 9:153.

47. Luck SJ, Chelazzi L, Hillyard SA, Desimone R: Neural mechanismsof spatial selective attention in areas V1, V2, and V4 of macaquevisual cortex. J Neurophysiol 1997, 77:24-42.

48. McAdams CJ, Maunsell JH: Effects of attention on orientation-tuning functions of single neurons in macaque cortical area V4.J Neurosci 1999, 19:431-441.

49. Shulman GL, Corbetta M, Buckner RL, Raichle ME, Fiez JA,Miezin FM, Petersen SE: Top-down modulation of early sensorycortex. Cereb Cortex 1997, 7:193-206.

50. Roy CS, Sherrington CS: On the regulation of the blood supply ofthe brain. J Physiol 1890, 11:85-108.

51. Sokoloff L: Relation between physiological function and energymetabolism in the central nervous system. J Neurochem 1977,29:13-26.

52. Raichle ME: A brief history of human functional brain mapping.In The systems. Edited by Toga AW, Mazziotta JC. San Diego:Academic Press; 2000:33-75.

53. Fox PT, Mintun MA, Raichle ME, Miezin FM, Allman JM,Van Essen DC: Mapping human visual cortex with positronemission tomography. Nature 1986, 323:806-809.

54.��

Attwell D, Iadecola C: The neural basis of functional brainimaging signals. Trends Neurosci 2002, 25:621-625.

The authors present a review of the literature that suggests that hemo-dynamic responses are driven by neurotransmitter-related signaling, andnot directly by the local energy needs of the brain.

55. Duvernoy HM, Delon S, Vannson JL: Cortical blood vessels of thehuman brain. Brain Res Bull 1981, 7:519-579.

56. Schuz A, Palm G: Density of neurons and synapses in thecerebral cortex of the mouse. J Comp Neurol 1989,286:442-455.

57. O’Kusky J, Colonnier M: A laminar analysis of the number ofneurons, glia, and synapses in the adult cortex (area 17) of adultmacaque monkeys. J Comp Neurol 1982, 210:278-290.

58. Zheng D, LaMantia A-S, Purves D: Specialized vascularization ofthe primate visual cortex. J Neurosci 1991, 11:2622-2629.

59. Ames A: CNS energy metabolism as related to function.Brain Res Brain Res Rev 2000, 34:42-68.

60. Sokoloff L: Function-related changes in energy metabolism inthe nervous system: localization and mechanisms. Keio J Med1993, 42:95-103.

61. Kadekaro M, Crane AM, Sokoloff L: Differential effects ofelectrical stimulation of sciatic nerve on metabolic activity inspinal cord and dorsal root ganglion in the rat. Proc Natl AcadSci USA 1985, 82:6010-6013.

62. Kadekaro M, Vance WH, Terrell ML, Gary HJ, Eisenberg HM,Sokoloff L: Effects of antidromic stimulation of the ventral rooton glucose utilization in the ventral horn of the spinal cord inthe rat. Proc Natl Acad Sci USA 1987, 84:5492-5495.

63. Nudo RJ, Masterton RB: Stimulation-induced [14]C2-deoxyglucose labeling of synaptic activity in the centralauditory system. J Comp Neurol 1986, 245:553-565.

64. Jueptner M, Weiller C: Does measurement of regional cerebralblood flow reflect synaptic activity - implications for PET andfMRI. Neuroimage 1995, 2:148-156.

65. Volterra A, Magistretti PJ, Haydon PG: The Tripartite Synapse:Glia in Synaptic Transmission. Oxford: Oxford UniversityPress; 2002.

66. Takahashi S, Driscoll BF, Law MJ, Sokoloff L: Role of sodiumand potassium ions in regulation of glucose metabolismin cultured astroglia. Proc Natl Acad Sci USA 1995,92:4616-4620.

67. Pellerin L, Magistretti PJ: Glutamate uptake into astrocytesstimulates aerobic glycolysis: a mechanism coupling neuronalactivity to glucose utilization. Proc Natl Acad Sci USA 1994,91:10625-10629.

68. Magistretti PJ, Pellerin L: Cellular mechanisms of brain energymetabolism and their relevance to functional brain imaging.Philos Trans R Soc Lond B Biol Sci 1999, 354:1155-1163.

69.��

Attwell D, Laughlin SB: An energy budget for signaling in thegrey matter of the brain. J Cereb Blood Flow Metab 2001,21:1133-1145.

The authors used anatomical and physiological data to analyze theenergy expenditure on different components of excitatory signaling inthe gray matter of rodent brain. The authors suggest that functionalmagnetic resonance imaging signals are likely to be dominated bychanges in energy usage that are associated with synaptic currentsand action potential propagation.

70.�

Lennie P: The cost of cortical computation. Curr Biol 2003,13:493-497.

The author presents an estimation of the cost of individual cellular eventsfor the human brain. On the basis of this energy budget, the authorsuggests that less than 1% of neurons in cortex can be active concur-rently. It is suggested that the high cost of spikes requires the brain notonly to use representational codes that rely on very few active neurons butalso to allocate its energy resources flexibly among cortical regionsaccording to task demand. This constraint may explain the need formechanisms of selective attention.

71. Scannell JW, Young MP: Neuronal population activity andfunctional imaging. Proc R Soc Lond B Biol Sci 1999,266:875-881.

72. Tagamets MA, Horwitz B: Integrating electrophysiological andanatomical experimental data to create a large-scale modelthat simulates a delayed match-to-sample human brainimaging study. Cereb Cortex 1998, 8:310-320.

73. Douglas RJ, Martin KAC: A functional microcircuit for cat visualcortex. J Physiol 1991, 440:735-769.

74. Douglas RJ, Koch C, Mahowald M, Martin KA, Suarez HH:Recurrent excitation in neocortical circuits. Science 1995,269:981-985.

75. Saint-Cyr JA, Ungerleider LG, Desimone R: Organization of visualcortical inputs to the striatum and subsequent outputs to thepallido-nigral complex in the monkey. J Comp Neurol 1990,298:129-156.

76. Sloot WN, Gramsbergen JB: Axonal transport of manganese andits relevance to selective neurotoxicity in the rat basal ganglia.Brain Research 1994, 657:124-132.

77. Takeda A, Kodama Y, Ishiwatari S, Okada S: Manganesetransport in the neural circuit of rat CNS. Brain Res Bull 1998,45:149-152.

78. Pautler RG, Silva AC, Koretsky AP: In vivo neuronal tract tracingusing manganese-enhanced magnetic resonance imaging.Magn Reson Med 1998, 40:740-748.

79. Watanabe T, Michaelis T, Frahm J: Mapping of retinal projectionsin the living rat using high-resolution 3D gradient-echoMRI with Mn2R-induced contrast. Magn Reson Med 2001,46:424-429.

80.��

Saleem KS, Pauls JM, Augath M, Trinath T, Prause BA,Hashikawa T, Logothetis NK: Magnetic resonance imaging of

12 New technologies


neuronal connections in the Macaque monkey. Neuron 2002,34:685-700.

The authors present the first application of manganese-enhanced MRI inmonkeys. Combined injections of manganese chloride and WGA-HRPwere performed in order to evaluate the specificity of the manganesechloride by tracing the neuronal connections of the basal ganglia of themonkey. Mn2þ and WGA-HRP yielded remarkably similar and highlyspecific projection patterns. By showing the sequential transport ofMn2þ from striatum to pallidum-substantia nigra and then to thalamus,

the authors demonstrated unequivocally MRI visualization of transportacross at least one synapse in the CNS of the primate.

81. Logothetis NK, Pauls J, Oeltermann A, Augath M, Trinath T:Studying connectivity with electrical microstimulation & fMRI.Abstr Soc Neurosci 2001, 27:821-830.

82. Hochmann J, Kellerhals H: Proton NMR on deoxyhaemoglobin.Use of a modified DEFT technique. J Magn Reson 1980,38:23-39.



MR imaging in the non-human primate: studies of function and of ...

Documents

Transcript of MR imaging in the non-human primate: studies of function and of ...