BOLD Contrast:BOLD Contrast:Functional Imaging with MRIFunctional Imaging with MRI

Mark A. Elliott, PhD

Department of Radiology

University of Pennsylvania

Overview

1. Mechanisms of functional imaging with MRI

2. Methodology of fMRI

3. Issues for animal studies

4. Spatial and temporal sensitivy of fMRI

Methods for Imaging Neural ActivityMethods for Imaging Neural Activity

electrical activity- excitatory- inhibitory- soma action potential

metabolic response

- glucose consumption- oxygen consumption

hemodynamic response- blood flow- blood volume- blood oxygenation

FDG PET

H215O PET

fMRIfMRIEEG

MEG

fNIRelectrophysiology

- ATP tightly regulated

Perfusion MRI

Vascular Sensitivity ofVascular Sensitivity offMRI and fNIR fMRI and fNIR

IIIIIIII

IIIIIVIV

Perfusion MRIPerfusion MRI

fNIRfNIRIIIIII

IIIIII IVIV

Vessel Size

Intravascular

Extravascular

Venous Arterial

fMRIfMRI

fMRI vs fNIRfMRI vs fNIR

fMRI fNIR

Spatial Resolution 8-27 mm3 “Blobs” 1-10 cm3

Temporal Resolution Slow (1-2 sec)Fast (50 Hz)Fast (50 Hz)important?important?

Measurement parameterMix of blood volume, blood

flow, and O2 metabolism [Hb] and [HbO]

Vascular Response

Mechanisms of fMRI Signal:Mechanisms of fMRI Signal:BOLD ContrastBOLD Contrast

• Hemodynamic response is a surrogate marker for neural activity

• BOLD = Blood Oxygenation Level-Dependent

• BOLD signal is a complex interaction of CBF + CBV + CMRO2:

CBF >> CMRO2 less deoxyhemoglobin with activation

CBF is monitored indirectly

– “Tracer” is primarily venous

– “Tracer” is endogenous

spatial dimension

Neural Activity

CMR02

CBF

BOLD ( CBF - CMR02)

“Flooding the garden to feedthe thirsty flower” - ???

Magnetic Susceptibility Affects Magnetic Susceptibility Affects Background Magnetic FieldBackground Magnetic Field

B H r0

r 1M

: permeabilityr : relative permeabilityM: magnetic susceptibility

For biological tissues, | M | << 1Diagmagnetic: M < 0Paramagnetic: M > 0

M1

M2

B1

B2

B1 B2

The interface between regions with different M behaves like a magnetized dipole, perturbing the local B field.

M creates larger B

BOLDBOLD Contrast: Changes in Contrast: Changes in Magnetic Susceptibility of BloodMagnetic Susceptibility of Blood

Blood and brain tissue are diamagnetic.Hb0 is diamagnetic.Hb is strongly paramagnetic.HbO is paramagnetic.

Increased Neuronal activity:• blood flow increases ≈ 30%• 02 consumption increases ≈ 5%• [Hb0] • [Hb]

Decrease in [Hb] reduces the Decrease in [Hb] reduces the MM between blood and brain tissue between blood and brain tissue

Magnetic field becomes more uniform Magnetic field becomes more uniform MRI signal affected MRI signal affected

Hemoglobin Saturation AffectsHemoglobin Saturation AffectsMagnetic Field HomogenietyMagnetic Field Homogeniety

NormoxiaNormoxiaHypoxiaHypoxia

from Bandettini and Wong, 1995

from Ogawa, 1990 Rat brain, 7T

Field Map vs. Hemoglobin Saturation

Summary: BOLD Contrast in fMRISummary: BOLD Contrast in fMRI

• BOLD = Blood Oxygenation Level-DependentBOLD = Blood Oxygenation Level-Dependent

• Oversupply of CBF raises [HbO] in regions of increased CMROOversupply of CBF raises [HbO] in regions of increased CMRO22

• Susceptibility mismatch between blood and tissue is Susceptibility mismatch between blood and tissue is reducedreduced

• Magnetic field becomes Magnetic field becomes moremore homogeneous homogeneous

• Temporal T2* contrast generated in T2* sensitive MRITemporal T2* contrast generated in T2* sensitive MRI

Broca’s areaBroca’s area

Wernicke’s areaWernicke’s area

Verbal Fluency TaskVerbal Fluency Task

fMRI Methodology: AcqusitionfMRI Methodology: Acqusition

EPI

structuralT1 weighted

~ 5 min

functionalT2* weighted

~ 2 sec/volume

. . . .

~ 300 images

~ 10 min

Temporal series of EPIs

1x1x1 mm voxels

3x3x3 mm voxels

time

fMRI Methodology: StimulusfMRI Methodology: Stimulus

On

Off

OnOff

Blocked Design

Event Related

ISI

Variable ISI allows for more stimuli per time.• Increased statistical power in analysis.

Event related designs can have either fixed or variable inter-stimulus interval (ISI)

Fixed ISIISI

Variable ISI

Blocked Design, Event-Related Design, and ISI

. . . .

Stimulus

SignalProcessing

Stimulus

Signal

“Activation”

Processing

fMRI Methodology: AnalysisfMRI Methodology: Analysis

“Non-Activation”

T2*-weightedSnapshot

Image

AverageDifference

Image

StatisticalSignificance

Image

ThresholdedStatistical

Image

Overlay onAnatomic

Image

Brain Activation MapsBrain Activation MapsStatistical Parametric MappingStatistical Parametric Mapping

OFFOFF

ONON

courtesy J. Detre

task

signal

Hemodynamic Response FunctionHemodynamic Response Function

FWHMAmplitude

Onset Time

Peak Contrast

Time to Peak

The “HRFHRF” - the theoretical impulse response of BOLD contrast to brief neuronal activity

StimulusStimulus

fMRI Model: HRF Linear SystemfMRI Model: HRF Linear System

stimulus (stimulus (xx))

HRF (HRF (hh))

signal (signal (yy))

Linear Model Assumption y = y = hh xx

Expected signal (y) is convolution of the

stimulus signal (x) with the HRF (h)

Signal is predicted for any arbitrary sequence of stimuli

Applications of fMRIApplications of fMRI

• Cognitive Neuroscience

– Localization of sensorimotor and cognitive function

– Brain-behavior correlations

• Clinical Neuroscience

– Presurgical mapping

– Differential diagnosis of cognitive disorders

– Recovery of function/neuroplasticityPhotic Stimulation

Implications for Animal fMRIImplications for Animal fMRI

• Pharmacological effects on neuronal metabolism and hemodynamic Pharmacological effects on neuronal metabolism and hemodynamic responseresponse

• Small voxel sizes reduce SNRSmall voxel sizes reduce SNR

• Smaller volumes enable higher field magnets (7 and 9.4T)Smaller volumes enable higher field magnets (7 and 9.4T)

• Passive stimulus delivery (training possible in some models)Passive stimulus delivery (training possible in some models)

T2* Signal Loss in the Pre-Frontal CortexT2* Signal Loss in the Pre-Frontal Cortex

F

SE

F = frontal sinusE = ethmoidal sinusS = sphenoidal sinus

Air is highly paramagnetic (like Hb)Air-tissue interface has “static” M

Background signal “drop-out”B0

1

2

Bn = Normal componentBn = Tangential component

B0

Normal component is unchanged by B1n = B2n

Tangential component is altered by B1n = 1 / 2 B2n

Signal Dropout in T2* Weighted Images

TE=4msTE=12ms

TE=20ms

TE=28ms

TE=36ms

TE=44ms

TE=52ms

TE=60msIncreasing TE

Spatial Extent of BOLDSpatial Extent of BOLD

Hb Saturation (%, approx.) resting activearterioles 90 90capillaries 80 90veins 60 90

Positive T2* contrast derived from CBF > CMR02

Venous compartment experiences largest Hb (and T2*)

Draining veins are less spatially specific to site of neural activity

Neural Activity

CMR02

CBF

BOLD ( CBF - CMR02)

draining veins microvessels

Extravascular BOLD Signal

from Principles of Functional MRI, Seong-Gi Kim

microvesselmicrovessel macrovesselmacrovessel

B0 “inhomogeneity” from vessel extends into extravascular (EV) space

water diffusion

EV Magnetic Field Gradient

Diffusion of water molecules through B0 gradients

• Large vessels: static dephasing, T2* effect• Small vessels: dynamic dephasing, T2 and T2* effect

Spin-echo fMRI less sensitive to large vessel (venous)extravascular space

Echo Time and Field Strength Echo Time and Field Strength Effects on BOLD ContrastEffects on BOLD Contrast

• BOLD contrast increase with echo time (TE)• SNR decreases with echo time

• Optimal CNR when TE resting T2*

• BOLD contrast increases with magnetic field• SNR increases with magnetic field

from Stroman et al, Proc. ISMRM, Glasgow (2001)

%

Sign

al%

Si

gnal

TE (msec)

TE (msec)

1.5T

3T

Field Strength EffectField Strength Effecton BOLD Spatial Sensitivityon BOLD Spatial Sensitivity

from S.P. Lee et al, (2003)

Rat brain, 9.4T

• T2* of blood shortens quadratically with B0

• Field dependence of T2,blood T2,tissue

- Decreased venous contribution

Diffusion weighted BOLDIntravascular BOLD component

model simulation