An Introduction to Functional Magnetic Resonance Imaging (FMRI) and Its Application to Psychiatry Kristen A. McKiernan, Ph.D. Michael C. Stevens Ph.D.

An Introduction to Functional Magnetic Resonance Imaging (FMRI)and Its Application to Psychiatry

Kristen A. McKiernan, Ph.D.Michael C. Stevens Ph.D.

Neuropsychiatry Research CenterThe Institute of Living

September 26, 2002

Kristen MichaelThe basic principles of FMRI Clinical applicationsHow do we get brain images Experimental approachResearch methods Example: An Oddball Task Collecting data Application to clinical groups Analyzing data Where can we go from here

Presentation Overview

Typical FMRI Experimental Setup

The basic principles of FMRI

Necessary Equipment

Magnet Gradient Coil RF Coil

Source: Joe Gati, photos

RF Coil

4T magnet

gradient coil(inside)

Bo

A Moment on Magnet SafetyThe magnetic field strength of these magnets is EXTREMELY powerfulIt is VERY important to keep metallic objects far away from the scanner area

To avoid injuries:Screen subjects (and researchers) carefullyMake sure anyone who will be near the magnet understands the importance of safetyand knows the safety procedures

Source: www.howstuffworks.com Source: http://www.simplyphysics.com/flying_objects.html

The MAGNET is used to align protons in the direction of the magnetic field (Bo)Outside magnetic field

Inside magnetic field

• spins tend to align parallel or anti-parallel to B0

• net magnetization (M) along B0

• spins precess with random phase• only 0.0003% of protons/T align with field

x 80,000 =

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

Robarts Research Institute 4T

Magnetic field is very strong and is continuously ON

Source: www.spacedaily.com

1 Tesla (T) = 10,000 Gauss

Earth’s magnetic field = 0.5 Gauss

B0M

Hydrogen nuclei

The GRADIENT COILS are used make small adjustments so that themagnetic field (Bo) is as homogeneous as possible

The gradients generate small magnetic fields in 3 directions: x y z

Putting a body in magnetic field makes it non-uniform, so we adjust the 3 orthogonal weak magnets to make thee magnetic field as homogenous as possible (i.e., equal strength across the field)

Gradient coil

The RADIO FREQUENCY (RF) Coil is used to apply a “pulse” of radiofrequency waves that “excite” the protons

This means that the direction of magnetization is temporarily altered

+ RF pulse =(90o flip angle) 2-4 ms duration

Bo M Bo M

Spins absorb energy, become excited and “flip”.Time to get back to Bovaries for different tissuesWe can measure this

Why do this?? Can’t detect M if aligned along BoWhen M is in the transverse plane, it induces a voltage in the coil – the RF signalMeasuring this signal produces the raw MRI data that we analyze

EquilibriumExcitation

Resonance frequency of42.58 MHz/T for 1H

Source: http://wsrv.clas.virginia.edu/~rjh9u/hemoglob.html, Jorge Jovicich

Hemoglogin (Hgb): - four globin chains - each globin chain contains a heme group - at center of each heme group is an iron atom (Fe) - each heme group can attach an oxygen atom (O2) - oxy-Hgb (four O2) is diamagnetic no B effects - deoxy-Hgb is paramagnetic if [deoxy-Hgb] local B

Introducing Hemoglobin – a magnetically susceptible molecule

So, I thought we were talking about BRAIN activity?

Hbr and the MRI Signal

Neural activity in the brain initiates a cascade of events: •Metabolic changes: in glucose and oxygen metabolism•Physiological changes: CBF, CBV, blood oxygenation level

These hemodynamic changes influence MRI signal intensity:CBF brings more H2O molecules into the imaging area (a “slice” of brain tissue) more protons align with Bo CBV brings more O2 into the area – much more than is needed

More HbrO2 means less deoxy-Hbr in the capillaries and veins (and less randomness in magnetic field)

The level of deoxy-Hbr is whataffects the MRI signal

deoxy-Hbr = MRI signal

The BOLD Signal in FMRI

Using the dependence of the MRI signal on the level of O2 in the blood is the most common FMRI technique. This type of MR signal is a Blood Oxygenation Level Dependent contrast

This is what the MRI BOLD signal looks likeIt represents “activity” (function) of brain cells

Research Methods

Two Design Possibilities

• Block Design– Useful for block tasks (PET studies)– Analysis simple to implement

• Event-Related Design– Can replicate single trial studies– Provides information about temporal response

Task1Task1Task2Task2Task1Task1Task2Task2ImagingImaging 30s30s 30s30s 30s30s 30s30s

30 s30 s 30 s30 s

Trial1Trial1 Trial2Trial2

Collecting Data and Preparing for Analyses

Creating an Image

Blood flow Voxel

Different voxels have differenthemodynamic propertiesthus the density of the magneticfield is different in each voxel

These differences, put togetherin space, produce images

(4mm x 4mm x 6mm)

Creating a Time Series(3D+time)

We decide on the thickness of each slice (4-7 mm) and number of slices needed (whole brain or a specific region)Take repeated volumes (50+) to get many samples of each voxel

2-3 secfor aVolume

1 slice

mostinferior slice

mostsuperior slice

IR-MPRAGE T1 Weighted StructuralGradient Echo, Echo Planar Image (EPI)

3D only 3D + time

Two Types of Images from Each Subject

provides detailed anatomical informationcontains functional data used in

statistical analysis

FMRI Data Analysis

Step 1: Subject Level Analysis

1. Model1. Model(1 or more(1 or moreRegressors)Regressors)

oror

RegressionRegressionResultsResults

2. Data2. Data

3. Fitting 3. Fitting the Model the Model to the Data to the Data at each at each voxelvoxel

Analysis Using AFNI software

9 voxels

Step 2: Group Level AnalysesStep 2: Group Level Analyses

To account for individual differences in brain size and anatomy, each subject’s 3D brain volume is “warped” to best fit a standard brain

Once normalized we can refer to specific locations using the Talairach Coordinate System

Subject data (ie statistical results) can then be combined across subjects to get experimental results – these are what you usually see reported

C

D

Kristen MichaelThe basic principles of FMRI Clinical applicationsHow do we get brain images Experimental approachResearch methods Example: An Oddball Task Collecting data Application to clinical groups Analyzing data Where can we go from here

Presentation Overview

“…So what does it mean?”(“…So what?”)

Image Interpretation…Is this all?

Clinical FMRI Applications

• In general, one approach is to compare brain activity between psychiatric groups and normal controls.

• But, that leaves a lot of room…• How do you ask intelligent and meaningful

questions?• The benefits of FMRI over other imaging modalities

primarily involve the combined abilities to quantify both the spatial extent and magnitude of that brain activity evoked by some cognitive process.

Any question you can think of...

– “How does brain function differ between schizophrenic patients and healthy controls?”

– OR “Do schizophrenic patients have a deficit in:• Attention• Working Memory• Language Use (i.e., auditory hallucinations)• Overall patterns of brain function on these tasks (functional

organization of brain activity)– “How do antipsychotic medications affect brain function in

schizophrenic patients (acute and chronic)?”– “Is the the relative effectiveness of certain medications

reflected in the hemodynamic measurement of brain function?”

...FMRI can examine.

– “How do biomarkers, symptom profiles and diagnostic classifications relate to patterns of brain function?”

– “What are the effects on the brain of long-term antipsychotic medication treatment?”

– “How effective is cognitive rehabilitation at improving brain function in schizophrenic patients?”

– “Are there cognitive function biomarkers in first-degree relatives of schizophrenics that speak to etiological factors (i.e., genetics)?”

– “How different is the cognitive function of first-break schizophrenics with those having a chronic illness?”

Experimental Approach to FMRI

• Theory• Hypotheses• Methods• Results• Interpretation

Experimental Approach to FMRI

• Theory - Schizophrenia is associated with brain dysfunction related to attentional orienting.

• Hypotheses - Evoked brain activity on an attentional orienting response will show reduced amplitude of response in brain areas known to subserve attention in healthy normal controls.

The oddball task

• Tones are presented and subject responds to low probability target tones (e.g., 12.5% trials)

• First ERPs ever recorded were to the oddball task – stimulus targets and omissions.

• Sokolov said salient stimuli are very robust elicitors of the orienting response, more robust than novel stimuli

• Historically one of the most well characterized tasks in psychopathology, schizophrenia in particular

• ERP studies have shown P3 component is reduced in schizophrenia and in psychopathy

• ERP studies have shown that the P3 is reduced in nearly every pathological condition – how can this be!

Three Stimulus Visual Oddball Task

T T T T T T T X T T T T T T C T T T T X T T T T T X X

T T T T T T T T T X T T T T T T X T T T X T C T T X T

Infrequent Target - “X” - Requires button press responseInfrequent Distractor - “C” - Ignored (no response)

14 - 9% “X” 9 - 9% “C” 97 - 82% “T”

Cognitive Processes Associated with Three-Stimulus Oddball Task Paradigm (Polich & Kok, 1995)

Somato-Motor Cortex– Preparation and Execution

Frontal and Parietal Cortex– Response Inhibition – Working Memory– Self-Monitoring of Response Accuracy

(including orienting)– Vigilance (sustained attention)

Occipital-Temporal Cortex– Visual Object Recognition– Long-Term Memory

Hemodynamic response to auditory oddball stimuli

Healthy ControlParticipants

Kiehl et al. (2001)

Kiehl et al. (2001)

Control subjects (n=11) Schizophrenic patients (n=11)

Results of group data

PSYCHIATRIC DIAGNOSIS

First episode patient

Database of other first episode patients

Schizophrenia AffectiveBipolar

PSYCHIATRIC TREATMENT

First episode patient(with schizophrenia)

Database of other schizophrenia patients

Risperidone HaloperidolOlanzapine

ADHD Anterior Brain Deactivation(Deactivation for ADHD subjects not seen in Controls)

ADHD SUBJECTS

CONTROL SUBJECTS

In areas of superior frontal gyrus and perhaps some medial frontal gyrus, there is deactivation to targets, which is not seen in controls.

Z = 48 mm Z = 54 mm

Z = 48 mm Z = 54 mm

L/R

Normal Control Response to Targets

Conduct Disorder Response to Targets

-30 mm 0 mm +30 mm +60 mm

R/L

R/L

-30 mm 0 mm +30 mm +60 mm

Difference Map: CD - NC

R/LX19 Y38 Z12Right Insula

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

-4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9

Time Course

% S

ign

al C

han

ge

CD+

CD-

X48 Y36 Z12Left Insula

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

-4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9

Time Course

% S

ign

al C

han

ge

CD+

CD-

What else has been done…?(What else COULD be done?)

• You name it…– Conduct Disorder, ADHD, psychopathy, Alzheimer’s Disease,

Learning Disabilities, stroke, epilepsy, autism, head injury, alcoholism, drug addiction, bipolar illness, OCD, Generalized Anxiety Disorder, PTSD, unipolar depression, etc.

– Memory, attention, language, working memory, motor function, executive-function, visual perception, etc.

• Capitalizes on vast field of previous research and theory.• Used in combination with other imaging and research

modalities.

Acknowledgements and

NRC, IOL External

Godfrey Pearlson, M.D. Robert Cox, PhD

Kent Kiehl, Ph.D. Jody Culham, PhD

Vince Calhoun, Ph.D.

Michael C. Stevens, Ph.D.

Kristen McKiernan, Ph.D.

Jin-Suh Kim, M.D.

thanks to those who provided slides or figuresused in this presentation

These websites can provide additional information on FMRI Robert Cox’s webpage:http://afni.nimh.nih.gov/afni/edu/index.html Jody Culham’s webpage:http://defiant.ssc.uwo.ca/jody_web/fmri4dummies.htm 

Doug Noll’s FMRI Primerhttp://www.bme.umich.edu/~dnoll/primer2.pdf 

Mark Cohen’s Basic MR Physicshttp://porkpie.loni.ucla.edu/BMD_HTML/SharedCode/MiscShared.html General questions related to FMRI:http://www.radiologyresource.org/content/functional_mr.htm Brain images from different clinical patients:http://www.med.harvard.edu/AANLIB/home.html