A P P E N D I X

F
Brain Imaging Techniques:
Invasiveness and Spatial and
Temporal Resolution
Alberto Zani, Gabriele Biella, and Alice Mado Proverbio

concerned, both are far superior to their

INTRODUCTION forebear, the X-ray technique.
Functional imaging includes a wide
In this appendix we provide some esti- range of techniques, which are listed here
mates of the spatial and temporal resolu- in increasing order of spatial and temporal
tion, as well as of the invasiveness, of the resolution: (1) 2-deoxyglucose cerebral
most frequently used neuroimaging tech- blood flow (2-deoxyglucose CBF), a fore-
niques. Before illustrating these estimates, runner of hemodynamic techniques rarely
it is important to remember that neuro- used nowadays; (2) single-photon emission
imaging techniques may be subdivided computed tomography (SPECT); (3)
into two broad categories according to positron emission tomography (PET); (4)
their different aims: imaging of brain functional magnetic resonance imaging
anatomy (structural imaging), or of brain (fMRI); (4) electroencephalography and
function (functional imaging). Structural event-related potentials (EEG–ERP); (5)
imaging is used to examine the static out- magnetoencephalography (MEG); and last,
lines of brain structures in both physiologi- but not the least, (6) microelectrode single-
cal and pathological situations. Functional unit recording, which involves recording
imaging, on the other hand, is used to gain electrophysiological signals from inside or
knowledge on (1) which structures are outside the membrane of a single neuron
activated during a specific cognitive task, body by means of microelectrodes.
at sensory and/or cognitive levels, (2) the Although the categorization of neuro-
interactions between the structures that are imaging techniques into the two aforemen-
activated, and (3) the way the functional tioned approaches is generally sound, it
activation of the brain is reorganized in should be mentioned that all these tech-
individuals affected by neurological dis- niques have some limitations and attempts
eases, strokes, or head injuries. are being made to overcome these. Some
Generally speaking, the structural cate- very recently devised techniques combine
gory of neuroimaging techniques includes both structural and functional imaging
two well-known neuroradiological tech- information and cannot be classified into
niques: computerized axial tomography either category. A rather interesting
(CAT) and magnetic resonance imaging example of these combined techniques is
(MRI). As far as brain imaging capacity is the so-called CAT–PET.

The Cognitive Electrophysiology of Mind and Brain 417 Copyright 2002, Elsevier Science (USA). All rights reserved.
418 F. IMAGING TECHNIQUES

techniques mentioned above. The spatial

SPATIAL AND resolution, expressed in millimeters, is
TEMPORAL RESOLUTION reported on the ordinate axis; the temporal
resolution––here indicated in seconds on a
The accuracy with which the imaging logarithmic scale—is depicted on the
techniques are able to provide definite abscissa. The height and width of the
images of the anatomy of the centers of the forms with which the different techniques
central nervous system (CNS), and/or the are represented in the figure indicate the
activation of these centers, in order to be known range of spatial and temporal reso-
able to localize them reliably, is defined as lutions, respectively, for each of the tech-
spatial resolution. Conversely, the speed niques. The increasing saturation of the
with which the techniques can keep on gray hue of the different shapes represents
scanning the CNS anatomy and physiol- the increasing degree of invasiveness of the
ogy, taking into account all intrinsic limita- techniques.
tions, i.e., the minimum time that must Structural techniques currently have the
necessarily pass between the collection of a most accurate spatial resolution or, in other
measure of one CNS activation and the words, localization capacity. Both CAT and
successive one, is described as temporal PET techniques have a spatial resolution
resolution. that is vastly superior to that of previous
Figure 1 is a graphical representation of techniques. Indeed, their spatial resolution
an estimate of the normal spatial and tem- has become so good that it is now in the
poral resolution for each of the imaging order of millimeters.

FIGURE 1 Invasiveness, spatial resolution, and temporal resolution of the main imaging techniques used in
humans to investigate function and structure of the brain. The different sizes of the shapes representing the dif-
ferent techniques vary as a function of the level of both spatial resolution (in milimeters) and temporal resolution
(in seconds). Note that the increasing level of saturation of the gray color represents the increasing levels of inva-
siveness of the techniques.

APPENDIXES
 SPATIAL AND TEMPORAL RESOLUTION 419

In PET, for example, the activated parts of Given that the final aim of research on
the brain selectively take up a radioactive the mind and brain should be to construct
tracer previously administered intraven- a model of functional relations between
ously to a patient or healthy volunteer. The the pathways and centers of the brain from
gamma rays deriving from the emission of which mental life comes, besides simple
positrons by these activated structures allow localization of these to particular areas
a functional map of cerebral activation to be of the brain, it is important to have a
built; as illustrated in Fig. 1, such a map has temporal resolution of milliseconds for the
a precision ranging between about 2 and 5 processes involved. The only imaging
mm. fMRI, on the other hand, can reflect techniques that have such a good temporal
structural variations caused by increased resolution are the techniques used sys-
local blood flow and dilatation of cerebral tematically or on single cells that measure
tissues with a mean precision of 3 mm, the electromagnetic activity of the brain
although the range is from 2 to 4 mm. directly. As illustrated by Fig. 1, the maxi-
Notwithstanding their high spatial reso- mal temporal resolution, as well as spatial
lution, none of the functional imaging resolution, is provided by single-unit
techniques, with the exception of MEG, can recordings. Thanks to these it is possible to
provide functional images that are also carry out neurofunctional investigations
accurate in temporal terms. In fact, the with a temporal resolution below the order
temporal resolution with which they can of milliseconds (<10–3), and with a spatial
provide accurate images of ongoing func- resolution under 1 mm. It is, however,
tional activation of the brain is rather poor. unthinkable to use this technique for func-
This resolution can reach the order of a tional imaging of the human brain because
tenth of second (~100–150 msec) with the of its invasiveness; it would require neuro-
most technologically advanced type of surgery to implant the microelectrodes.
fMRI––that is, 3 or 4 tesla echo-planar Unlike single-unit recordings, scalp
fMRI (or event-related fMRI)—but still recordings of voltages (EEGs and ERPs)
remains in the order of seconds with less that mirror the intracranial currents origi-
powerful equipment. The temporal resolu- nating from neuronal sources in the brain
tion of PET is tens of seconds or even cortex, and spreading by volume conduc-
minutes. The significance of this technical tion throughout the brain and the scalp,
limitation to research can readily be can be used as tools for human research.
appreciated by considering that an action Indeed, while having the advantage of
potential originating in the pyramidal 1-msec temporal resolution, or quite close
motor neurones of the premotor cortex to this level, the recording method is com-
propagating along the efferent pathways pletely noninvasive. However, because of
takes about 150 msec to reach the muscle its irregularities, the skull is not a homoge-
bundles of the forearm, causing flexion of neous conductor. The volume currents that
the terminal phalanx of the index finger, in come in contact with the electrodes over
order, for example, to press a button for the scalp are distorted by irregularities
measuring reaction times. Or consider that such that the technique does not have
we can identify an object that enters our sufficient spatial resolution to be able
visual field within a few hundreds of milli- locate the real intracranial sources of the
seconds (~180–220 msec). It is clear that the currents.
velocity with which the above-mentioned This difficulty translates into a spatial
neural processes occur means that their resolution that cannot be relied on for
subprocesses escape measurement tech- localization purposes. In the best cases,
niques because of the interval between suc- localization of the electric dipole ranges
cessive sampling. between a minimum of 6 mm and a maxi-

APPENDIXES
420 F. IMAGING TECHNIQUES

mum of 11 mm or even beyond, depending involves parallel recording of these para-

on a whole series of recording conditions meters in a single experimental paradigm.
and modeling parameters (see Fig. 1). Even Although it is certainly true that the differ-
then, there are some extreme cases, such as ent approaches are very useful individually,
the so-called far-field potentials, in which the only integration of the techniques with dif-
scalp site at which the largest signal is ferent spatial and temporal resolutions can
measured is actually far away from the provide truly valuable information on the
source area; for example, sensory-evoked neurofunctional mechanisms of mental
responses of the auditory cortex to uni- processes, and on their temporal course of
lateral stimuli, despite originating on the activation.
dorsolateral side of the contralateral brain
hemisphere close to the ear, produce their
largest amplitudes at the top of the scalp. INVASIVENESS
Although having the same temporal res-
olution as the EEG, magnetoencephal- Techniques to investigate the function
ography is only minimally influenced by and structure of the nervous system can be
the nonuniform conductivity of the brain, classified as invasive, semiinvasive, and
skull, and scalp, and if the head is modeled noninvasive. The difference between the
using spherical geometry the recorded invasive and semiinvasive techniques is
magnetic field can be considered com- that the former implies a surgical lesion
pletely independent of the conductivity of (mechanical) and the latter involves a
the head. Referring once again to Fig. 1 it physicochemical stress (with radioopaque
can be seen that only MEG has high levels or radioactive substances).
of both spatial and temporal resolution. In Techiques that can be defined as nonin-
detail, the spatial resolution of this tech- vasive are MEG and EEG, both of which
nique ranges from a minimum of 1.5 mm involve simple recordings of electrical
to a maximum of 4 mm for the cortical potentials or natural magnetic fields pro-
areas of the brain. Unfortunately, this reso- duced by brain activity without experi-
lution decreases dramatically to some cen- mentally introduced chemicostructural
timeters for subcortical regions. This perturbations of the cerebral regions.
technique could, therefore, be the tech- Semiinvasive techniques provide a
nique of choice for localizing activity in the means to observe dynamic and structural
superficial areas of the brain, i.e., the cere- properties through images that are recon-
bral cortex, which is responsible for mental structed on the basis of a principle of both
processes in general. Unfortunately, this active and passive detectability. In the first
technique is still too expensive to become case, the products of rapidly metabolized
as widely used as the ERPs. compounds or radioactive substances with
In order to overcome the various limita- a short half-life are detected. The com-
tions of each of the techniques presented pounds are injected intravenously and,
here, strategies are being developed to use being distributed preferentially to certain
combined methods; for example, the images regions of the brain, show in the various
obtained by MRI, fMRI, or PET can be com- areas of the brain a differential distribution
bined with those from microelectrodes, or that is proportional to the state of activa-
those from MEG and ERPs. Combination tion of the area (PET, SPECT, and, when
methods can be based on a direct or an tracers are used, also fMRI and MRI). In
indirect approach. The former uses hemo- the case of passive detection (e.g., CAT),
dynamic images to obtain a real structural the X radiations are delivered by the
basis of the estimates of functional activa- machine and hit the cerebral tissues
tion acquired separately, whereas the latter ab externo (with the usual radiographic

APPENDIXES
 SUGGESTED READING 421

principle). Even MRI and fMRI, without Binder, J. R., and Rao, S. M. (1994). Human brain
injection of markers, induce changes in mapping with functional magnetic resonance
imaging. In “Localization and Neuroimaging in
cerebral or spinal structures. In fact, apply- Neuropsychology” (A. Kertesz, ed.), pp. 185–212.
ing a magnetic field causes a change, albeit Academic Press, Orlando.
transitory, in the characteristics of the De Yoe, E. A., Bandettini, P., Neitz, J., Miller, D., and
spatiotemporal molecular organization, Winans, P. (1994). Functional magnetic resonance
with realignment of the electromagnetic imaging (fMRI) of the human brain. J. Neurosci
Meth 54, 171–187.
dipoles of charged molecules. Frith, C. D., and Friston, K. J. (1997). Studying brain
Mechanically invasive techniques belong function with neuroimaging. In “Cognitive
to the realm of functional neurosurgery; Neuroscience” (M. D. Rugg, ed.), pp. 169–195.
this involves extra- or intradural record- Psychology Press, Taylor & Francis Group, Hove,
ings, or even recordings from deep struc- East Sussex, UK.
Josephs, O., Turner, R., and Friston, K.J. (1997). Event-
tures (in the case of therapeutic placement related fMRI. Hum. Brain Mapping 5, 243–248.
of stimulatory electrodes). These techniques Perani, D., and Cappa, S. (1999). Neuroimaging
obviously involve surgery on specific nerv- methods in neuropsychology. In “Handbook of
ous tissue and perforation or opening of the Clinical and Experimental Neuropsychology” (G.
skull. Denes, and L. Pizzamiglio, eds.), pp. 69–94.
Psychology Press, Taylor & Francis Group, Hove
Mechanical or physicochemical invasive- East Sussex, UK.
ness implies a different degree of danger Posner, M. I., and Raichle, M. E. (1994). “Images of
for the patient. Naturally this creates a Mind.” W. H. Freeman, New York.
boundary; the economic costs and benefits Reiman, E. M., Lane, R. D., Van Petten, C., and
of the results of the diagnostic or surgical Bandettini, P. A. (2000). Positron emission tomog-
raphy and functional magnetic resonance imaging.
intervention must be compared to not car- In “Handbook of Psychophysiology” (J. T.
rying out the procedure. On average, given Cacioppo, L. G. Tassinary, and G. G. Berntson,
precise conditions and the noninstrumental eds.), 2nd Ed., pp. 85–118. Cambridge University
diagnosis, the indication for an invasive Press, Cambridge, Massachusetts.
intervention is highly controlled and Rosen, B. R., Buckner, R. L., and Dale, A. M. (1998).
Event-related MRI: Past, present, and future. Proc.
justifiable. Natl Acad. Sci USA 95, 773–780.
Rugg, M. D. (1999). Functional neuroimaging in
cognitive neuroscience. In “The Neurocognition of
Suggested Reading Language” (C. M. Brown and P. Hagoort, eds.),
pp. 15–36. Oxford University Press, Oxford and
Single-Unit Recordings New York.
Huguenard, J., and McCormick, D. (1994). “Electro-
physiology of the Neuron.” Oxford University Electromagnetic Imaging
Press, New York. Del Gratta, C., and Romani, G. L. (1999). MEG:
Llinás, R. (1988). The intrinsic electrophysiological Principles, methods, and applications. Biomed.
properties of mammalian neurons: Insights into Technik 44 (Suppl. 2), 11–23.
central nervous system function. Science 242, Hari, R., and Lounasmaa, O. V. (1989). Recording and
1654–1664. interpretation of cerebral magnetic fields. Science
Nicholls, J., Nartin, A., and Wallace, B. (1993). “From 244, 432–436.
Neuron to Brain.” 3rd Ed. Sinauer, Sunderland, Hillyard, S. A., and Picton, T. W. (1987). Electro-
Massachusetts. physiology of cognition. In “Handbook of
Physiology, Sect. 1, Vol. 5, Higher Functions of the
Hemodynamic Functional Imaging Brain” (F. Plum, ed.), pp. 519–584. American
Bandettini, P. A., Rasmus M. B., and Donahue, K. M. Physiological Society, Bethesda, MD.
(2000). Functional MRI. Background, method- Hillyard, S. A. (1993). Electrical and magnetic brain
ology, limits, and interpretation. In “Handbook of recordings: Contributions to cognitive neuro-
Psychophysiology” (J. T. Cacioppo, L. G. Tassinary, science. Curr. Opin. Neurobiol. 3, 217–224.
and G. G. Berntson, eds.), 2nd Ed., pp. 978–1014. Kutas, M., and Dale, A. (1997). Electrical and mag-
Cambridge University Press, Cambridge, netic reading of mental functions. In “Cognitive
Massachusetts. Neuroscience” (M. D. Rugg, ed.), pp. 197–242.

APPENDIXES
422 F. IMAGING TECHNIQUES

Psychology Press, Taylor & Francis Group, Hove Vaughan, H. G. (1988). Topographic analysis of brain
East Sussex, UK. electrical activity. In “The London Symposia (EEG
Kutas, M., Federmeier, K. D., and Sereno, M. I. (1999). Suppl. 39)” (R. J. Ellingson, N. M. F. Murray, and
Current approaches to mapping language in A. M. Halliday, eds.), pp. 137–142. Elsevier,
electromagnetic space. In “The Neurocognition of Amsterdam.
Language” (C. M. Brown and P. Hagoort, eds.),
pp. 317–392. Oxford University Press, Oxford and Combining Techniques
New York. Dale, A. M., and Sereno, M. I. (1993). Improved local-
Näätänen, R., Ilmoniemi, J., and Alho, K. (1994). ization of cortical activity by combining EEG and
Magnetoencephalography in studies of cognitive MEG with MRI cortical surface reconstruction: A
brain function. Trends Neurosci. 17, 389–395. linear approach. J. Cogn. Neurosci. 5, 162–176.
Regan, D. (1989). “Human Brain Electrophysiology: Halgren, E., and Dale, A. M. (1999). Combining of
Evoked Potentials and Evoked Magnetic Fields in electromagnetic and hemodynamic signals to
Science and Medicine.” Elsevier, Amsterdam. derive spatiotemporal brain activation patterns:
Scherg, M. (1992). Functional imaging and localization Theory and results. Biomed. Technik 44 (Suppl. 2),
of electromagnetic brain activity. Brain Topogr. 5, 53–60.
103–111. Luck, S. J. (1999). Direct and indirect integration of
Scherg, M., and Ebersole, J. S. (1993). Models of brain event-related potentials, functional magnetic
sources. Brain Topogr. 5, 419–423. resonance images, and single-unit recordings.
Swick, D., Kutas, M., and Neville, H. J. (1994). Hum. Brain Mapping 8, 115–120.
Localizing the neural generators of event-related Wieringa, H. J. (1993). “MEG, EEG and the Integration
brain potentials. In “Localization and Neuro- with Magnetic Resonance Images”. Doctoral
imaging in Neuropsychology” (A. Kertesz, ed.), Thesis.” CIP-Gegevens Koninklijke Bibliotheek,
pp. 73–121. Academic Press, Orlando. Den Haag, Nederlands.

APPENDIXES