Statistical Parametric Statistical Parametric MappingMapping

Lecture 2 - Chapter 8

Quantitative Measurements Using fMRI

BOLD, CBF, CMRO2

Textbook: Functional MRI an introduction to methods, Peter Jezzard, Paul Matthews, and Stephen Smith

Many thanks to those that share their MRI slides online

Cerebral blood vessels

• Capillary beds extend into gray matter

• Arteries enter cortical surface perpendicularly

Layer 1Layer 1

Layer 6Layer 6

Neuron’s General Structure

• input - dendrites & soma• processing - throughout• output - axon

Structural variety of neurons

~50,000 neurons per cubic mm~6,000 synapses per neuron~10 billion neurons & ~60 trillion synapses in cortex

Signal Pathway in BOLD fMRI

Brain activity

OxyhemoglobinDeoxyhemoglobin

Magnetic Susceptibility

Glucose, O2 consumption blood volume, blood flow

T2*, T2

fMRI Signal

Oxy – diamagneticDeoxy – paramagnetic

M = H(susceptibility constant = )

Deoxy> tissue

Oxy~ tissue

T2* fMRI Signal

HbO2 – OxyhemoglobinHbr - Deoxyhemoglobin

From Neural Activity to fMRI Signal

Neural activity Signalling Vascular response

Vascular tone (reactivity)Autoregulation

Metabolic signalling

BOLD signal

glia

arteriole

venule

B0 field

Synaptic signalling

Blood flow,oxygenationand volume

Complex relationship between change in neural activity and change in blood flow (CBF), oxygen consumption (CMRO2) and volume (CBV).

dendriteEnd bouton

fMRI and Electrophysiology

Logothetis et al, Nature 2001

a. 24 sec stimulationb. 12 sec stimulationc. 4 sec stimulation

LFP – local field potentials reflect dendritic currentsMUA – multiunit activitySDF – single unit activity (?)

Haemodynamic Response

Buxton R et al. Neuroimage 2004

balloon model

%

-1

initial dip undershoot

fMRI Bold Response Model

time

BO

LD

res

pons

e, %

initialdip

positiveBOLD response

post stimulusundershootovershoot

1

2

3

0

stimulus

Figure 8.1. from textbook.

• Initial dip 0.5-1sec• Overshoot peak 5-8 sec• + BOLD response 2-3%• Final undershoot variable

Deoxyhemoglobin

BOLD signal

A BOLD Block Design Visual Study

time

voxel response

0 20 40 60 80 100 120 140 160 180

-1.5

0

3

4.5

Time [s]

S

ign

al [

%]

stimulusoff

on

image acquisition

time

-2

2

6

14

10

t value

1.5 predicted response

correlation

0

Bruce Pike, BIC at MNI.

Non-Linearity of BOLD Response

BOLD response vs. length of stimulation

BOLD response during rapidly-repeated stimulation

ts

Block designs use stimulus and rest periods are that are long relative to BOLD response.

Graded BOLD Response

Figure 8.2. from textbook. N=12 subjects.

• Graded change in signal for a) BOLD and b) perfusion (CBF).• 3 minute visual pattern stimulation with different luminance levels.• Note max BOLD change of 2-3 % and max CBF change of 40-50 %.

Model of Overshoot/Undershoot

Figure 8.3. from textbook.

• Models of waveform for a) BOLD and b) perfusion (CBF) change.• Constant stimulation 50-250 sec.• Overshoot more pronounced in BOLD waveform

• slow adjustment of CBV (Mandeville et al., 1999)• Undershoot might be due to same effect

Perfusion vs. Volume Change

Figure 8.4. from textbook.

• 30 second stimulation• 3-second intervals• CBF rapid• CBV slow

Mandeville et al., 1999

In rat experiments TC for CBV similar to that for BOLD overshoot.

Measurement of Cerebral Blood Flowwith PET or MRI (Arterial Spin Labeling - ASL)

• Uses magnetically labeled arterial blood water as an endogenous flow tracer

• Potentially provide quantifiable CBF in classical units (mL/min per 100 gm of tissue)

Detre et al., 1992

arterial labeling

control labeling

imaging slice

T1 relaxation

arterial spin labeling

O infusion

or inhalation

15

α decay

PET or SPECT Steady State Method

MRI PERFUSION Steady State Method

+

511 keV

511 keV

PET

Method

O-15 H20

Arterial Spin Labeling

• ASL IMAGE = IMAGEunlabeled – IMAGElabeled

• Mostly use inversion (IR) labeling• Labeled blood water extracted from capillaries• T1 of blood is long compared to tissues • Flow (perfusion) not dependent on local susceptibility

www.fmrib.ox.ac.uk/~karla/

inversionslab

imagingplane

excitation

inversion

xy

z (=B0)

bloodblood

white matter = low perfusion

Gray matter = high perfusion

Hypercapnia, Perfusion, & BOLDResponses

Figure 8.5. from textbook. Perfusion (CBF) and BOLD changes.

• Response with graded hypercapnia (GHC thin line) and graded visual stimulation (GVS). Four levels in this study.

• BOLD response similar to CBF response to hypercapnia• BOLD response attenuated relative to CBF during aerobic stimulation

CMRO2 – Cerebral Metabolic Rate of Oxygen ConsumptionHypercapnia (increased CO2) increases CBF w/o increasing oxygen demand (CMRO2).

ASL interleaved with BOLD

Figure 8.8. from textbook.

α /1

0

/10

0,2

2 /1

−

⎟⎠

⎞⎜⎝

⎛⎟⎠⎞

⎜⎝⎛ Δ

−=CBF

CBF

M

BOLDBOLD

CMRO

CMRO

Acquisition of CBF and BOLD data supports calculation of CMRO2 using model equation.

Flow/Metabolism Coupling and the BOLD Signal

Figure 8.9. from textbook.

• BOLD vs Perfusion (CBF)• graded hypercapnia (dark circles)• graded visual stimuli (different shapes)

• CMRO2 vs Perfusion (CBF)• perfusion has somewhat linear relationship with CMRO2 • derived from data in “a”

Model Based Images

Figure 8.10. from textbook.

a. M from model equation – predicts max BOLD signal potentialb. BOLD – visual stimulation flashing checkerboardc. CBF (perfusion)d. CMRO2 (oxygen compution rate)

Localization of Functional Contrast

Perfusion Activation

BOLD Activation

PerfusionPerfusion

BOLD*BOLD*

*1.5T/Gradient Echo*1.5T/Gradient Echo

drainingdrainingveinvein

ASL Perfusion fMRI vs. BOLDImproved Intersubject Variability vs. BOLD

Aguirre et al., NeuroImage 2002

Single SubjectSingle Subject Group (Random Effects)Group (Random Effects)

behavior neural function

metabolism

blood flow

BOLD fMRI

biophysics***

ASL CBF MRI

***BOLD contrast includes contributions from biophysical effects such as magnetic field strength homogeneity and orientation of vascular structures.

Physiological Basis of fMRI

disease

ASL fMRI measures changes in CBF directly, and hence measured signal changes may be more directly coupled to neural activity