Neuroimaging

1
Neuroimaging

• What is neuroimaging?
• Types of neuroimaging
• Subtractive methodology
• What can neuroimaging tell us about cognition?
• Summary

2
Understanding neuropsychology

• Lesion studies
• Disconnection studies
• Group studies
• Cognitive neuropsychological methodology
• Modularity assumption allows for specific
  cognitive processes e.g. face recognition to be
  isolated using an experimental approach -gt brain
  imaging studies
• Human brain mapping
• Changes in activity in the brain reveal how
  cognition works e.g. how long it takes to
  recognise a word

3
Phineas Gage lead pipe blasted through frontal
lobe
DLPFC Dorsolateral prefrontal cortex
4
Phrenology bumps on the head
5
Brocas area
6
(No Transcript)
7
Brodmanns map
8
Brief History of Brain Mapping

• In 1929 Berger first recorded brain electrical
  activity (EEG) at the human scalp suggesting it
  could be a tool to investigate mental states.
• In the 50s the introduction of radioactive
  tracers as a quantitative measure to study blood
  flow, and later the use of X-ray CT scan and PET
  provided effective methods to study the brain.
• The introduction of fMRI offered the use of a
  non-invasive technique which could be applied to
  the study of cognitive function in normal
  volunteers using the subtractive methodology.

9
Electro-encephalography (EEG)

• Electric and magnetic activity occurs naturally
  in the brain and seems to encode information
  about brain functions thus allowing us to make
  some inferences about mental states.
• In EEG, a number of electrodes are placed over
  the scalp and these record the activity of
  neurons over a period of time (temporal).
• The EEG trace varies according to the general
  state of the brain (asleep, awake etc).

10
Brain gives off an electrical charge
11
(No Transcript)
12
EEG

• When we are quiet or asleep, EEG waves are
  synchronised, with a wave-like shape, with waves
  being of a particular amplitude and frequency
  (usually alpha waves, 8-12 Hz).
• When we are awake and thinking, the EEG wave
  tends to be de-synchronised, which means there is
  an irregular electrical activity.
• Electrical and magnetic activity in the brain can
  be time-locked to a specific stimulus or event,
  and event-related potentials (ERPs) or
  event-related fields (ERFs) can be measured.

13
Event Related Potentials (ERP)

• ERPs are a sequence of positive and negative
  voltage deflections or components (e.g. N400)
  with specific time delays and wave shapes.
• The evoked response to a single stimulus at the
  scalp is tiny and is extracted from background
  activity by means of the averaging technique
  which enhances the amount of signal and reduces
  the amount of noise to nearly zero.
• Brain activity is revealed by ERPs with a very
  high degree of temporal precision (milliseconds)
  but with a rather poor spatial resolution.

14
ERP is time locked to Stimulus presentation
15
(No Transcript)
16
(No Transcript)
17
Positron Emission Tomography (PET)

• The brain stores no oxygen and little glucose so
  the energy necessary for continuous neural
  activity depends on blood supply.
• Measures of blood flow can be obtained by means
  of a tracer in the blood which can provide
  valuable information on brain activity.
• PET uses radioactive tracers, such as 15O2, which
  have a short half-life (approximately 2 minutes)
  and this allows the experimenter to perform
  several scans on the same subject.

18
(No Transcript)
19
Blood supply to the brain can be observed with an
angiogram
20
Regional cerebral blood flow (rCBF)

• A tracer is injected into a vein and it enters
  the brain in about 30 seconds then - while it is
  decomposing - it gives a picture of regional
  blood flow (rCBF) in specific parts of the brain.
• As the tracer breaks down to its stable form, it
  emits a positron, which will collide with an
  electrode to produce two annihilation photons and
  these are detectable by the imaging device.
• The spatial resolution of PET is determined by
  the distance traveled by the positron before its
  annihilation, and is approximately 2-3 mm.

21
(No Transcript)
22
(No Transcript)
23
Data analysis

• The classical approach to imaging techniques uses
  the so-called ROI (region-of-interest) approach
  regions that would be analysed for changes in
  blood flow or metabolic activity are chosen
  a-priori and then scanned.
• This approach is best applied when the brain
  region is known in advance i.e. the primary
  sensorimotor cortices such as vision/audition.
• A-priori approach not suited to higher functions
  such as language and attention because of
  individual differences in brain localisation.

24
(No Transcript)
25
(No Transcript)
26
Subtraction methodology

• PET studies of cognitive function use a different
  approach called the subtraction methodology.
• In this approach two different brain states are
  compared a control state (usually rest) which is
  subtracted from a task state (usually a decision)
  in order to create a difference image.
• This image tells us which areas in the brain are
  involved in the task state.
• Depends on an assumption of feed-forward
  information processing (see Van Orden, 2001).

27
Subtraction methodology
28
Pure insertion?

• The assumption of "pure insertion".
• This assumption is the main criticism leveled at
  PET studies by comparing a task state to a
  control state we assume that the differences we
  observe represent the processing components
  introduced by the task itself and do not reflect
  the processes of the control state at all.
• This can be overcome by designing the control
  task to contain the same components as the
  experimental task except for critical variable of
  interest (e.g. visual lexical decision).

29
a
Lexical Decision gt False font
R
Controls
Wordsay yes Nonword no
xdjuil
ugytr say yes
LH
d
Deep Dyslexic patient
Orthographic Input lexicon
LH
e
Spoken production gtLexical decision
Controls
Read spoon
Wordsay yes Nonword no
c
LH
Phonological output lexicon
Deep Dyslexic patient
cc 0.18
0.6
LH
30
Averaging data

• PET results are usually averaged across subjects
  in order to enhance the signal-to-noise
  properties of the images.
• Averaging is a controversial procedure but the
  averaging principles that are used in PET studies
  are common to all subjects and the algorithms
  used for both across-subjects and within-subjects
  averaging are very powerful.
• The subtraction and averaging techniques require
  a common reference system for the localisation of
  activated areas in each brain.

31
Talairach co-ordinates

• fMRI researchers refer to various brain regions
  according to Brodmanns areas or classical names
  such as that of Broca's area but the use of
  standardised reference points is preferable.
• A common reference system is provided by the
  Talairach stereotaxic space atlas of the human
  brain and software packages (e.g. SPM) have been
  developed to standardize brain images.
• This methodology allows us to define areas of
  activation, their borders and volume in a
  commonly accepted and comprehensible way.

32
(No Transcript)
33
(No Transcript)
34
Magnetic Resonance Imaging (MRI)

• MRI is a non-invasive technique that uses a
  magnetic field in order to elicit a detectable
  signal that can create a spatial brain image.
• The subject is placed in a uniform magnetic field
  and an appropriate radio-frequency is then
  transmitted through the field for a brief period.
  
• Resonance in atomic nuclei (hydrogen atoms) are
  detected by a receiver coil around the head
• An image of the brain can be formed in which
  tissue of different density are distinguishable
  with a spatial resolution of about 2 millimetres.

35
(No Transcript)
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
Functional MRI (fMRI)

• Several techniques have been developed to adapt
  MRI as a measure of brain structure to measuring
  brain activity using blood flow.
• The most commonly technique is the Blood Oxygen
  Level Dependent (BOLD) method.
• fMRI does not require external tracers but uses
  instead the magnetic properties of an internal
  substance (haemoglobin) in order to provide the
  tracer by which brain activity can be analysed.

40
Oxygen metabolism

• Neural activity causes a large increase of blood
  flow but a small change in oxygen consumption
  leading to an increase in the proportion of
  oxygenated haemoglobin in brain tissue.
• The magnetic properties of oxygenated haemoglobin
  differ from those of deoxygenated haemoglobin so
  activate areas are visible by changes in signal
  intensity measured by MRI.

41
Data analysis

• fMRI data are collected as a rapid sequence of
  scans making it possible to observe the changes
  occurring in brain activity in different areas at
  many different points in time.
• Data analysis in fMRI studies is approached by
  using thresholding techniques an approach which
  proved to be powerful in PET.
• These methods attempt to fix a-priori a threshold
  above which a response can be considered
  statistically significant.

42
Statistical Parametric Mapping (SPM)

• Software packages have been produced for this
  purpose among them SPM (statistical parametric
  mapping) is the best known.
• Statistical parametric maps are images in which
  voxels are distributed according to a probability
  density function based on activity.
• These maps are images of significance whose
  simplest forms are t scores based on repeated
  measurements of rCBF data in two different brain
  states (e.g. task vs control).

43
(No Transcript)
44
(No Transcript)
45
Word stimuli
Associative Semantics
Size Judgment
46
Character stimuli
Associative Semantics
Size Judgment
47
Picture stimuli
Associative Semantics
Size Judgment
48
Bold amplitude signal
change
Fig. 4
49
(No Transcript)
50
(No Transcript)
51
Magnetoencephalography (MEG)

• Magnetoelectroencephalography (MEG) can be
  performed during fMRI scans of the brain by
  deriving event-related magnetic fields (ERFs).
• MEG has the same temporal resolution as EEG but
  also allows the experimenter to localise the
  source of the ERF quite reliably.
• This enables direct intervention and precise
  monitoring of electrical activity in the brain
  with reliable spatial resolution (2 for the price
  of 1).

52
(No Transcript)
53
Problems with MEG

• This spatial resolution is restricted to the gyri
  and sulci of the lateral surface of the brain
  that are oriented parallel (near to) the scalp.
• Therefore MEG cannot tell us anything about the
  activity of the bulk of the human neo-cortex
  under the lateral surface at this stage.
• It is useful for measuring the reaction time
  taken to perform a cognitive task (like reading)
  and for tracking changes in location that occur
  during brain activity over time intervals.

54
(No Transcript)
55
Summary

• Different methods of brain imaging can give a
  converging picture of how the brain works.
• Brain imaging studies may tell us about the
  regions of the brain that are used for specific
  behaviours and possibly cognitive processes.
• Brain imaging studies can be used to test
  hypotheses derived from modular theories.

56
References

• Bub, D.N. (2000). Methodological issues
  confronting pet and fMRI studies of cognitive
  function. Cognitive Neuropsychology 17(5), 467 --
  484
• Poeppel D. (1996a) A critical review of PET
  studies of phonological processing. Brain and
  Language, 55, 317-351.
• Poeppel, D. (1996b). Some remaining questions
  about studying phonological processing with PET
  Response to Demonet, Fiez, Paulesu, Petersen, and
  Zatorre. Brain and Language, 55, 380-385.
• Van Orden, G.C. and Paap, K.R. (1997). Functional
  neuroimages fail to discover pieces of mind in
  the parts of the brain. Philosophy of Science, 64
  (4S)S85-S94.

57
(No Transcript)
58
(No Transcript)
59
fMRI Bold Rapid Overview

• Basal state
• Normal flow
• Basal level of Hbr
• Basal CBV
• Normal MRI signal

• Activated state
• Increased flow
• Decreased Hbr (lower field gradients around blood
  vessels)
• Increased CBV
• Increased MRI signal (from lower field gradients).

60
Computerised Tomography CT an X ray
61
Echo planar imaging (EPI)

• fMRI images are collected sequentially.
• Fast data acquisition techniques such as echo
  planar imaging (EPI) are used to avoid the image
  recorded at one point in the brain reflecting
  activity occurring at a slightly different time
  in another part of the brain.
• One property of fMRI is the opportunity to run
  correlation studies because it is possible to
  look for a temporal correlation between a
  particular input (e.g. a stimulus) and the
  resulting response in one part of the brain.

62
Block Timing Diagram
X2
PS picture semantics PB picture size
judgement CS character semantics CB character
size judgement
63
Problems with subtraction methodology

• True theories of cognition (Bub)
• Modularity
• Feedback effects (interactivity) (Van Orden)

64
(No Transcript)