Lecture intro to fmri

Image Contrast• Signal comes principally from water • Contrast comes from ‘relaxation’ processes • T1 - how quickly the system gets back to equilibrium • T2 - how quickly the signal deca

Introduction to Functional MRI

Daniel Bulte

Centre for Functional MRI of the Brain University of Oxford

• fMRI as a tool in neuroscience

• What is MRI?

• Neuronal firing, blood flow and BOLD

• A basic fMRI experiment

• What can fMRI be used for?

• Integrating modalities

Functional MRI

Gall and Spurzheim (1810)

Brodmann’s anatomical areas

Functional Localisation

• Brain injury and surgical cases

• Direct electrode stimulation/recording

• PET - blood flow and metabolism

• EEG - scalp potentials from neuronal firing

• fMRI - blood oxygenation

• MEG - magnetic field from neuronal firing

MRI vs FMRI

• In most MRI we only

worry about protons

(1H)

Nuclear Spin

spin

magnetic moment

M M=0

Spin Transitions

Low energy High energy

The Larmor Equation

ω = γ B

128 MHz at 3 Tesla

From Frequencies to Images

• Magnetic field gradients

• Vary the field by position

• Decode the frequencies to

give spatial information

Inside an MRI Scanner

Image Contrast

• Signal comes principally from water

• Contrast comes from ‘relaxation’ processes

• T1 - how quickly the system gets back to equilibrium

• T2 - how quickly the signal decays

• T2* - signal decay due to neighbouring

magnetic effects

T1 Weighted Image

SPGR, TR=14ms, TE=5ms, flip=20º

T2 Weighted Image

SE, TR=4000ms, TE=100ms

T2* Effects

Short TR

Short TE

Long TE

Long TR

Imaging Pathology

T1W T2W PDW

Neuronal Activity, Blood Flow

and the BOLD Effect

Electrical and Metabolic Activity

• large changes in membrane potential leading to action

potential

cell body

spiking axon

Electrical and Metabolic Activity

Pellerin and Magistretti, 1994

glucose

capillary astrocyte

basal pathway

Lactate

phosphorylation

• chemical transmission across synapses

neuro-Electrical and Metabolic Activity

• The biochemical reactions that transmit

neural information via action potentials and neurotransmitters all require energy

• This energy is provided in the form of ATP

• ATP is produced from glucose by oxidative phosporylation and the Kreb’s cycle

Electrical and Metabolic Activity

Glucose

Acetyl CoA

CoA

TCA Cycle

Oxidative Phosphorylation

H2O

2 CO2

e

-Electrical and Metabolic Activity

• ATP is hydrolysed to ADP, giving up energy

• The production of ATP from ADP is governed

by demand

• Rate of oxygen consumption by oxidative

phosphorylation is a good measure of neural activity

• The oxygen is supplied by the blood

• Since oxygen is not very soluble in water it is bound to haemoglobin

• Haemoglobin is an organic molecule with an iron atom bound in the centre

Haemoglobin

Oxy-haemoglobin

Deoxy-haemoglobin

Paramagnetic Diamagnetic

(same as tissue)

BOLD Contrast

• Haemoglobin acts as an endogenous

intravascular contrast agent

• As the level of oxygenation changes, so too does the contrast in the images

Variation with O2 Saturation

Bandettini and Wong Int J Imaging Systems and Technology 6:133 (1995)

BOLD Contrast

• Neuronal processes create oxygen demand

• Oxygen level in the blood affects the local magnetic field

• T2* weighted images are sensitive to this - lower oxygenation = lower signal

• Upon ‘activation’ signal goes down?

Metabolism and Haemodynamics

A Basic fMRI Experiment

The Tools

• High field MRI system

– At higher the BOLD effect is

stronger and signal is higher

• Fast imaging sequence

– Echo Planer Imaging (EPI)

• T2* weighted images

– EPI is a T2* weighted sequence (generally)

The Tools

• Stimulus presentation

– Visual/Auditory/Somatosenory presentation – Response (button box etc.)

– Software

• Must be MR safe

– Not ferromagnetic

– Potential heating effects

– Introducing noise to the scanner room

A Very Simple Experiment

A Very Simple Experiment

A Very Simple Experiment

Basic Analysis

• Simple subtraction (on-off)

• Shows the regions where the signal increased

on activation

• Areas active in both tasks cancel out

• Statistical inference (t-test) to show where

there was significant signal change

A bit more advanced

• Pre-processing to reduce effects of

movement during scan

• Add some information about the expected delay of the haemodynamic response

• Correlation analysis (General Linear Model)

• Statistical inference based on spatial size as well as strength of signal increase

Statistical Parametric Maps

Un-thresholded Thresholded Overlaid

Group Comparisons

• Register images to a standard

space

• Voxel-by-voxel group statistics

• Region of interest analysis

• Inference on a population or to

increase statistical power

What can fMRI be used for?

A few examples

Audio Visual fMRI

Visual

Auditory

❚ Hemifield angular

measurements separate posterior fusiform from V4

❚ Hemifield eccentricity

measurements show similar divisions

❚ Reveal primarily foveal

representations

Visual field mapping

Cohen et al (2002)

1 Regardless of hemi-field, words engage the region

more than consonant letter strings

Visual hemi-field Left RightLocation invariance

fMRI studies of hand movement

rest move rest move rest move rest

fMRI reveals different activity

patterns in disease

control stroke

PMCv

M1 PMCdr PMCdc

hand (right)

Aff

hand (left)

Changes in the FMRI activity during affected hand movements after therapy

Aff hand

(L)

Changes in FMRI activity are specific to

patients who improve

Correlation vs Causation

• Imaging techniques show us

state and a task

• Imaging techniques cannot tell us

whether activation in a particular

performance of a task

• To do this we need to use

Integrating Modalities

MRI can’t do everything (quite)

Catani et al, NeuroImage 2002

Insular fibres

Temporal fibres

“Virtual dissection” of fibre tracts

provides biochemical, physical

and molecular information

Why do MRS?

Rapid modulation of GABA during

motor learning

▲- learning; ■ - random ,◊ - control

• The visible GABA signal in the sensorimotor cortex

decreased rapidly during motor learning, but not

during random tracking movements;

• GABA modulation was specific to learning and not a

general result of performance of motor task. Floyer et al

EEG in the Scanner

Simultaneous EEG/fMRI: 32 channel EEG, Brainamps, 3T

EEG in the Scanner

• Data analysis methods that account for the

temporal and spatial effects of BOLD contrast