Functional neuro-imaging methodsPET H2O15 -blood flow

CO215 , CO15 , O-O15 -blood volume

FDG -metabolism

Specific neuro-transmitters

Optical Signal Imaging -blood flow

-blood volume

fMRI Perfusion with contrast agents

Perfusion without contrast agents

Blood Oxygenated Level dependence (BOLD)

Connectivity

EEG MEG

PET Optical Imaging fMRI

Spatial resolution

Temporal resolution

Field ofview

4mm

1min

The wholebrain

The wholebrain

~50 <1mm

few ms <100ms

~2cm, cortex only

MRI Timeline

1946 MR phenomenon – Bloch & Purcell

1952 Nobel Prize – Bloch & Purcell

1960 NMR development as an analytical tool

1972 Computerized Tomography

1973 Backprojectionj MRI – Lautembur

1975 Fourier Imaging – Ernst

1980 MRI demonstration – Edelstein

1986 Gradient Echo Imaging

MRI Microscopy

Functional MRI demonstration

1988 Angiography – Dumoulin

1989 Echo-Planar Imaging – Mensfield

Perfusion imaging

1991 Nobel Price – Ernst

1994 Xe Imaginng Hyperpolarized (Xe-129)

2004 Nobel Price – Lautembur & Mensfield

Necessary Equipment

Magnet Gradient Coil RF Coil

Source: Joe Gati, photos

RF Coil

4T magnet

gradient coil(inside)

x 80,000 =

4 Tesla = 4 x 10,000 0.5 = 80,000X Earth’s magnetic field

Robarts Research Institute 4T

The Big MagnetVery strong

Continuously on

Source: www.spacedaily.com

1 Tesla (T) = 10,000 Gauss

Earth’s magnetic field = 0.5 Gauss

Main field = B0

B0

Microscopic Property Responsible for MRI

The human body contains ~63% hydrogen atoms

Single voxel cells Water molecules

Each hydrogen molecules can be thought of as a small magnetic field, and will cause the nucleus to produce an NMR signal

Spins

Spin is a fundamental property of nature like electrical chargeor mass. Spin comes in multiples of 1/2 and can be + or -. Protons, electrons, and neutrons possess spin. Individual unpaired electrons, and neutrons each possesses a spin of 1/2.

When placed in a magnetic field of strength H, a particle with a net spin can absorb

a photon, of frequency o. The frequency o depends on the gyromagnetic ratio, 

of the particle.

=- Ho The Larmor frequency

 For hydrogen = 42.58 MHz/T

Low energy state High energy state

E=hhEnergy for transition

Energy transitions

E=-1/2hoE=+1/2ho

o =- Ho

The Larmor frequency

N-/N+ = exp(-E/kT) – Boltzmann distribution

The magnetization vector

Two magnetic moments – procession

000 HM

dt

dM

The Bloch eq. (without relaxation)

Ho

Mo

DT- MRI Fiber Tract VisualizationDT- MRI Fiber Tract Visualization

Coronal Sagittal Axial

Diffusion ellipsoids in coronal slice of human brain

Diffusion ellipsoids in coronal slice of human brain

In 1890, Roy and Sherrington concluded that

“…the chemical products of cerebral metabolism contained in the lymph which bathes the walls of the arterioles of the brain can cause variations of the caliber of the cerebral vessels: that in this re-action the brain possesses an intrinsic mechanism by which its vascular supply can be varied locally in correspondence with the local variations of functional activity.”

This “neurovascular coupling” is the base of functional neuro-imaging.

E = mc2

???

The First “Brain Imaging Experiment”

“[In Mosso’s experiments] the subject to be observed lay on a delicately balanced table which could tip downward either at the head or at the foot if the weight of either end were increased. The moment emotional or intellectual activity began in the subject, down went the balance at the head-end, in consequence of the redistribution of blood in his system.”

-- William James, Principles of Psychology (1890)

Angelo MossoItalian physiologist

(1846-1910)

… and probably the cheapest one too!

Blood Oxygenation Level Dependent (BOLD) fMRI contrast

Based on modulation of blood susceptibility with activation.

Measures: Amount of deoxy-hemoglobin in the tissue. This indirectly is proportional to the ratio of oxygenated/deoxygenated hemoglobin (diamagnetic and paramagnetic)

Gradient echo - GEFI, GE-EPI

Spin Echo - FSE, SE-EPI

• changes in T2*

• changes in T2

T2* vs. T2

• Extravascular (~33%) contribution in T2* (none in T2 )

• T2* changes caused by signal dephasing

diffusion

out

in

• T2 changes caused by either: i) diffusion through the gradients surrounding the erythrocytes or by ii) exchange of water between regions of different susceptibility (erythrocyte-plasma)

B0 i

nhom

ogen

eity

Distance from blood vessel

Statistical Mapsuperimposed on

anatomical MRI image

~2s

Functional images

Time

Condition 1

Condition 2 ...

~ 5 min

Time

fMRISignal

(% change)

ROI Time Course

Condition

Activation Statistics

Region of interest (ROI)

The BOLD signal

neuronal activity

energy demand

need for oxygen/glucose

hemodynamic response

• Coupling between oxygen/glucose utilization and hemodynamic response• Control mechanism• Blood flow vs. blood volume ?

1 2 3

The BOLD signal - 2

Phase 1: Diamagnetic oxy-hemoglobin minus oxygen > paramagnetic deoxy

Effective T2* relaxation ; smaller T2* signal decreased

Localized to the activation neurons.

Phase 2: Over flow of arterial blood – uncoupling between flow and oxygen

Less paramagnetic material > inefficient relaxation > T2* increase

Signal increases

Phase 3: 1. Flow returns to rest, still high rate of oxygen utilization 2. Flow returns to rest, blood volume return slower

Increase dexoy > efficient relaxation > short T2* >> signal decrease

The fMRI BOLD signal The fMRI BOLD signal

• Aid in neurosurgery planning

• Used for brain mapping

fMRI mapping of the brain’s language areas replaces invasive pre-surgery electrocortical

stimulation mapping which requires a patient to be awake.

Medical imaging can now display changes in brain activity caused by normal thought processes, disease, or therapeutic drugs

Human brain circuitry for imagining one's hand in the posture of another's hand.

• The basis of cognitive research: e.g., relation

between perception cognition behavior mood

and health

• Used in neuropathology research

Dementia Patient Volunteer

Brain Iron Distribution as a Potential Biomarker for Neurodegenerative Diseases (NDD) (By short T2 mapping)

• Allow research on brain function, architecture and

organization

• Used to understand brain network

Neural architecture of emotion perception and affects-related cognition

Generation of cortical transient clusters during activation

Flow vs. volume

Increase in blood flow > less deoxy Hb > signal increases

Increase in blood volume > more deoxy Hb > signal decreases

• Increase/decrease in blood flow FASTER than blood volume

• In normal conditions, in the cortex - flow is dominant (~70%)

CBV=0.5CBF 0.5

Temporal resolution

fMRI speed:

• GEFI produces an image in few sec (1-2sec).

• EPI produces an image in few tens of ms (100-1000ms)

Brief Stimuli

• The shortest stimuli that gives fMRI signal is ~35ms

• The shortest distinguished time-difference between two stimuli is 200ms

Temporal resolution –2

Hemodynamic Delays

• A delay of ~1-2sec for the initial deep• A delay of ~5-8sec for the positive main BOLD signal

Neither of these effects seems to add up to the delay observed

• Some delay due to “plumbing” is expected

•The coupling between neuronal activity and blood flow is not direct it includes a cascade of events of chemical signaling of several messengers.

Temporal resolution -3

• The brain venous network is not uniform. Particularly, deep brain nuclei do not have the cortical venous structure.

• The yet unknown main mechanism of the BOLD delay

Limits the possibility to follow neuronal information processing relates to the serial order of events in

different brain areas.

Spatial resolution

Cortical Columns are radially oriented clusters of neurons processing similar task

• Functional connectivity within the columns is rich, between columns is much weaker

• Size of cortical column is 100-300 in diameter and ~3mm in length

Are cortical columns the brain elementary processing units?

What is the size element of neural data processing ?What is the size element of neural data processing ?

Spatial resolution -2

Typical fMRI voxel is 2-4mm, covering tens of columns

Practical needed resolution – ~2 mm for definition of area of activityand ~300 for detailed structure

Relationship of microvasculature to cortical elements:Relationship of microvasculature to cortical elements:

Cortical columns seems to be arranged about a single common artery and vein that presumably perfussed the column.

Using optical imaging it was shown that activity in single column affects vascular signal for several mm around

Linearity of BOLD responseDale & Buckner, 1997

Linearity:“Do things add up?”

red = 2 - 1

green = 3 - 2

Sync each trial response to start of trial

Not quite linear but good enough

Data Analysis

• Single voxel time course Vs. Cluster analysis

• Model driven Vs. Model free statistics

• Strength of activation Vs. Volume of activation

• Choice of threshold Signal and effect -to-noise ratio

Model based analysis

Knowledge of the neuronal response is assumed – • linear or non-linear with the stimulus• impulse response• hemodynamics delay• assuming no spatial dependence

Neuronal response, even strong, that do not mach themodel will be invisible !!!

Model based analysis - problems

• In pathological conditions the basis assumption for the BOLD signal might change. (i.e., the fraction of flow and volume)

• the model have to be modified –• spatial dependent model is needed but the dependency is unknown (i.e., tumor)

• For long stimulus or short rest intervals between stimulus – adaptation have to be included – how ?

Model-free statistics

• Pharmaceutical MRI - in most cases model is unknown

• In pathological conditions - BOLD basic assumption might change

When it is necessary:

• Non cortical areas

Possible approaches

Cluster analysis PCA ICA

Threshold

How sure we are that this is the area of activation ?

fMR

I S

igna

l

Area of activation

Threshold level