Transcript The Anatomy of Language Sydney Lamb Rice University, Houston

Ling 411 – 10
Functional Brain Imaging (cont’d)
MEG
REVIEW
Functional Brain Imaging Techniques
 Electroencephalography (EEG)
 Positron Emission Tomography (PET)
 Functional Magnetic Resonance Imaging
(fMRI)
 Magnetoencephalography (MEG)
• Magnetic source imaging (MSI)
 Combines MEG with MRI
Magnetoencephalography (MEG)
 MEG (MagnetoEncephaloGraphy) measures
the magnetic field around the head
 Compare EEG: Measures voltage changes
on the scalp
 MSI (Magnetic Source Imaging) is MEG
coupled to MRI
Intra-Cranial Sources
Dipole (source
current)
Papanicolaou 1998:31
Magnetoencephalography (MEG)
 Records the magnetic flux or the magnetic
fields that arise from the source current
 A current is always associated with a
magnetic field perpendicular to its
direction
 Magnetic flux lines are not distorted as
they pass through the brain tissue because
biological tissues offer practically no
resistance to them (cf. EEG)
A dipole is a small current source
 Dipole generates a magnetic field
 Dendritic current from apical dendrites of
pyramidal neurons
 At least 10,000 neighboring neurons firing
“simultaneously” for MEG to detect
Recording of the Magnetic Flux
 Recorded by special sensors called magnetometers
 A magnetometer is a loop of wire placed parallel to
the head surface
 The strength (density) of the magnetic flux at a
certain point determines the strength of the
current produced in the magnetometer
 If a number of magnetometers are placed at
regular intervals across the head surface, the
shape of the entire distribution by a brain activity
source can be determined (in theory)
Magnetic flux from source currents
Magnetometer
Magnetic flux
Source current
Recording of Magnetic Signals
An MRI Machine
Recording of the Magnetic Flux
 Present day machines have 248 magnetometers
 The magnetic fields that reach the head surface
are extremely small
 Approximately one million times weaker than the
ambient magnetic field of the earth
 Because the magnetic fields are extremely small,
the magnetometers must be superconductive
(have extremely low resistance)
 Resistance in wires is lowered when the wires are
cooled to extremely low temperatures
Recording of the Magnetic Flux
 When the temperature of the wires approaches
absolute zero, the wires become
superconductive
 The magnetometer wires are housed in a
thermally insulated drum (dewar) filled with
liquid helium
 The liquid helium keeps the wires at a
temperature of about 4 degrees Kelvin
 The magnetometers are superconductive at this
temperature
Recording of the Magnetic Flux
 The currents produced in the
magnetometers are also extremely weak and
must be amplified
 Superconductive Quantum Interference
Devices (SQUIDS)
 The magnetometers and their SQUIDS are
kept in a dewar, which is filled with liquid
helium to keep them at an extremely low
temperature
How a MEG Recording is Made
 The MEG machine is
located in a
magnetically shielded
room
•
Subjects cannot wear
any metal because it
affects the recording
 Digitization process
 After digitization, the
task is run and the
recording is made
The Digitization Process
 Needed for co-registration with MRI
•
•
•
MRI scan is done later
Provides images
MSI – Magnetic Source Imaging
•
5 points
•
Subjects must remain extremely still during the
digitization process
 Method
 3 electrodes on forehead
 2 earpieces
 After digitization, the task is run and the
recording is made
Dipolar Distribution of the Magnetic Flux
 In the following figure, one set of concentric
circles represents the magnetic flux exiting the
head and the other represents the re-entering
flux
 This is called a dipolar distribution
 The two points where the recorded flux has the
highest value are called extrema
 The flux density diminishes progressively,
forming iso-field contours
Surface distribution of magnetic signals
Extrema
Dipolar Distribution of the Magnetic Flux

1.
2.
•
•
3.
4.
From the dipolar distributions, we can determine some
characteristics of the source
The source is below the mid-point between the
extrema (points where recorded flux has highest
value)
The source is at a depth proportional to the distance
between the extrema
Extrema that are close together indicate a source close
to the surface of the brain
A source deeper in the brain produces extrema that are
further apart
The source’s strength is reflected in the intensity of
the recorded flux
The orientation of the extrema on the head surface
indicates the orientation of the source
Co-registration of MEG and MRI space
MEG scan co-registered with MRI scan
using fiducial markers
Result of co-registration
Event-related brain responses: EEG & MEG
 Both types of signals come from the same
type of event: active dipoles
• Different directions from the dipoles
• Detected by different devices
 With EEG
• ERP – event-related potential
 With MEG
• ERF – event-related (magnetic) field
• Addition from 100 or more trials for each
tested condition needed to get measurable data
The inverse problem
 A problem for EEG and MEG
 Locating the dipole(s) based on signals reaching
surface of scalp
 Problem: Multiple solutions are possible
• Cf. solving x + y = 24
 Computer uses iterative procedure to come up
with best fit
 The problem is compounded by the fact that the
brain is a parallel processor
• Many dipoles at each temporal sampling point
Testing Reliability of MSI
 Necessary in early stages of research
• Does MEG give reliable localization results?
 Compare with results of Wada test
• Excellent correlations found
• (But this tests only very crude localization)
 Compare with results of intraoperative mapping
• MSI and mapping by cortical stimulation
demonstrate similar localization abilities –
excellent correlation
MSI before neurosurgery
 MSI is preferred because mapping by
cortical stimulation increases the
patients’ susceptibility to infections as a
result of lengthened surgery durations
 MSI can be performed prior to the
scheduled surgery so that the surgeons
can plan the best way to remove the
damaged area while avoiding language
areas as best they can
Temporal Resolution of MEG
 Excellent – unlike fMRI and PET
 The temporal order of activation of areas in a
pattern can be discerned
 The time course of the activation can be followed
 MEG has potential to detect the activation of
several brain regions as they become active from
moment to moment during a complex function such
as recognition
Temporal Resolution of MEG
 Only with MEG can we detect the
activation of several brain regions as
they become active from moment to
moment during a complex function such
as recognition
 But it is (at present state of the art)
virtually impossible to achieve precision
Time course of activation
 We can follow the activation of a source across time
 The magnetic fields recorded in MEG are evoked
 Activation at each point in time is recorded
(millisecond sensitivity)
 Sources of early components of Evoked Fields
circumscribe the modality-specific sensory areas
 Sources of late components circumscribe different
sets of brain regions (mostly association cortex)
• These activation patterns are function- (or task-)
specific
Spatial limitation of MEG
 Magnetic flux is perpendicular to direction
of electrical current flow
 Flux is therefore relatively easy to detect
if dendrites are parallel to surface of skull
• i.e., for pyramidal neurons along the sides of
sulci
 But hard or impossible to detect if vertical
• i.e., for pyramidal neurons at tops of gyri or at
bottoms of sulci
The challenge of MSI
 The cortex is a parallel processor
• Hundreds or thousands of dipoles can be active
simultaneously
 Multiple dipoles make comprehensive
inverse dipole modeling virtually impossible
 Hence, compromises are necessary
• Sample larger time spans (up to 500 ms)
• Sample larger areas (up to several sq cm)
Some MEG/MSI Findings
Speech recognition: MEG results
Hemispheric Asymmetry
Wernicke's Area
Variability in location of Wernicke’s area
(different subjects)
From MEG lab, UT Houston
Wernicke’s area in bilinguals
From MEG lab, UT Houston
Localization of phonemes:
The claim of Obleser et al.
 Different locations (in
temporal lobe) for
different vowels
 The anterior-posterior axis
corresponds to the
backness of a vowel – the
more back the vowel, the
more posterior the source
location
 The superior-inferior axis
corresponds to the height
of a vowel (inverse
relationship) – the higher
the vowel, the more
inferior the source location
of that vowel
From: Ladefoged, P. (2001). Vowels and Consonants:
An Introduction to the Sounds of Languages. Malden,
Massachusetts: Blackwell Publishers, Inc.
Distinguishing features of vowels
 Tongue height
corresponds to F1 (first
formant)
 Front-back dimension
corresponds to F2 (2nd)
 The formants are
detected in auditory
processing (upper
temporal lobe)
 Tongue positions are
controlled by motor
cortex (frontal lobe) and
monitored in parietal lobe
Tongue positions
From: Ladefoged, P. (2001). Vowels and Consonants:
An Introduction to the Sounds of Languages. Malden,
Massachusetts: Blackwell Publishers, Inc.
MEG and localization of phonemes
 Wernicke’s area may be
organized phonemotopically
 The anterior-posterior axis
corresponds to the
backness of a vowel – the
more back the vowel, the
more posterior the source
location
 The superior-inferior axis
corresponds to the height
of a vowel (inverse
relationship) – the higher
the vowel, the more
inferior the source location
of that vowel
From: Ladefoged, P. (2001). Vowels and Consonants: An
Introduction to the Sounds of Languages. Malden,
MEG and localization of phonemes
 Results: The relative positions of neural
representations for vowels in Wernicke’s area
correlate with the relative positions of the vowels
in articulatory space
•
•
•
•
Obleser, Elbert, Lahiri, & Eulitz, 2003
Obleser, Lahiri, & Eulitz, 2004
Obleser, Elbert, & Eulitz, 2004
Eulitz, Obleser, & Lahiri, 2004
•
•
Finding supported by different lab!
Shestakova, Brattico, Soloviev, Klucharec, & Huotilainen,
2004!
 Can this finding be replicated?
Shestakova et al. experiment (2004)
 Done in Helsinki, Russian vowels [i a u]
• Obleser et al. in Germany, German vowels [i a u]
 Results similar to those of Obleser et al.
• Higher cortical location for [a]
• Front-back cortical location corresponds to
articulatory positions
 They go two steps further:
• Input from different speakers (all male)
• Similar findings in both LH and RH
An MEG study from Max Planck Institute
Naming animals from visual (picture) input
LH
RH
More information on MEG
 The University of Texas Health Science
Center at Houston Division of Clinical
Neurosciences MEG Lab:
• http://www.uth.tmc.edu/clinicalneuro/
 Papanicolaou, A. (1998). Fundamentals of
Functional Brain Imaging: A Guide to the
Methods and their Applications to
Psychology and Behavioral
Neuroscience.Lisse: Swets & Zeitlinger.
Imaging methods compared
A practical consideration: Cost
 Most expensive: MEG
• About $2 million for the machine
• $1 million for magnetically
shielded room
 Next most expensive: PET
 Next: fMRI
 Cheapest: EEG
Temporal resolution – summary
 PET: 40 seconds and up
 fMRI: 10 seconds or more
 MEG and EEG: instantaneous
• Theoretically it is possible to do ms by ms
•
•
tracking, to follow time course of activation
Commonly used sampling rate for MEG: 4 ms
Practically, such tracking is difficult or
impossible
 The inverse problem
 Too many dipoles at each point in time
Spatial Resolution
 EEG: Poor
 PET: Fair – 4-5 mm
 fMRI: Fair – 4-5 mm
• MRI: Good – 1 mm or less
 MEG: Fairly good – 3-4 mm or less
• Under good conditions
Sensitivity of Imaging Methods
 All of the methods have limited sensitivity
 MEG
• 10,000 dendrites in close proximity have to be
active to detect signal
 PET and fMRI
• Similar limitations
 Any activation that involves fewer numbers
goes undetected
The Territory of Neurolinguistics:
An Intellectual Territory
with three dimensions
Dimension 1: Size
Dimension 2: Static – Dynamic
D
y
n
a
m
i
c
major structural
change
small structural change
function/operation
anatomical structures
Static
1
10
100
- - - - - nm - - - - - -
1
10
100
- - - - - μm - - - - -
1
mm
1
10
- - cm - -
Tiny sizes – nm to μm range
Synaptic Structure
Small sizes – μm range
 Pyramidal cell
• Diameter of cell body: 30-50 μm
• Diameter of axon: up to 10 μm
• Diameter of apical dendrite: up to 10 μm
 Cortical minicolumn
• Diameter: 30-50 μm
 Layers of the cortex
• Average thickness of one layer: 500 μm
Middle range – the Cortex
From top
to bottom,
about 3
mm
Representation, Processing, Change
(the second dimension)
 Static
• Representation of linguistic information
 Large scale (LARGE-SCALE REPRESENTATION)
 Small scale (SMALL-SCALE REPRESENTATION)
 Dynamic
• Linguistic information processing (PROCESSING)
• Learning and adapting (CHANGE OF STRUCTURE)
Understanding Representation
 The large scale (sq cm and up)
• How organized?
• What components?
• Where located?
• How interconnected?
 The middle scale (sq mm and below)
• Minicolumns
• Maxicolumns
• Clusters of columns
• Interconnections of columns
• Internal structure of minicolumns
 The small scale
• Internal structure of neurons
Representation at the large scale
 Principles of organization
 Linguistic subsystems
• Broca’s area –
•
•
•
 Phonological production
 Syntax(?)
Wernicke’s area – phonological recognition
Conceptual areas
Etc.
 Interconnections of subsystems
 Functional webs
LARGE-SCALE REPRESENTATION
What we know so far –
Principles of Organization I
“Wernicke’s Principle”
 Each local area does a small job
 Large jobs are done by multiple small areas
working together, by means of interconnecting
fiber bundles
 The basic principle of connectionism
 Consequences
•
•
Distributed representation and local representation
Distributed processing
LARGE-SCALE REPRESENTATION
What we know so far –
Principles of Organization II
 Genetically determined primary areas
• Motor – frontal lobe
• Perceptual – posterior cortex
 Somatic – parietal
 Visual – occipital
 Auditory – temporal
 Hierarchy
 Proximity
 Plasticity
LARGE-SCALE REPRESENTATION
The Proximity Principle
 Neighboring areas for closely related
functions
• The closer the function the closer the proximity
 Intermediate areas for intermediate
functions
 Consequences
• Members of same category will be in same area
• Competitors will be neighbors in the same area
end