Functional Magnetic Resonance Imaging. All subatomic particles possess a property called ‘spin’ i.e. like a planet rotating on it’s axis Magnetic.

PSYC 4380 Rhythms of the BrainCycle 4: Methods

Functional Magnetic Resonance Imaging

Physics behind MRI

All subatomic particles possess a property called ‘spin’ i.e. like a planet rotating on it’s axis

Magnetic fields can perturb and align these axes of rotation

Physics behind MRI

The central component of an MRI scanner is a very powerful magnet

The earth’s magnetic field is 1/20,000 T

Scanner magnets are typically ~3T (60,000x stronger than earth’s field)

MRI – Basic Principles

Use powerful magnet to align hydrogen atoms in biological tissue

Transmit radio-frequency (RF) pulses to perturb the rotational axes of protons

Record RFs emitted by protons as they return the orientation imposed by large magnet, and use this to calculate H+ density

MRI – Basic Principles

H+ density varies in different types of biological tissue, and MRI has sufficient sensitivity to distinguish different tissue types

MRI – Basic Principles

The proton-emitted RF pulse can be measured in a number of ways:

1. Measure the rate at which protons are realigned with the magnetic field (‘T1’)

2. Measure the drop-off in emitted RF pulses from protons as they realign (‘T2’)

T1 and T2 values at 1.5 T

Tissue T1 (ms)

T2 (ms)

Grey Matte

r920 100

White Matte

r790 92

Magnetic Properties of HB Hemoglobin’s magnetic properties

depend on whether it is oxygenated (HbO2) or deoxygenated (Hb)

HbO2 is ‘dimagnetic’ and has no net effect on the magnetic field

Hb is ‘paramagnetic’ and thus increases the strength of the local magnetic field when aligned by the scanner magnet

BOLD Signals

BOLD = Blood Oxygen Level Dependent

The change in Hb’s magnetic properties as a function of it’s oxygenation allow us to measure changes in blood oxygenation

Always a relative measure (A-B)

BOLD Signals

What is the relationship of the BOLD signal to electrophysiological signals?

from Logothetis, J Neurosci 2003

BOLD Signals

The selectivity of spiking & BOLD signals are not identical

So what is thebest electricalcorrelate of BOLD?

BOLD Signals

from Logothetis et al., Nature 2001

Simultaneous recording of action potentials (MUA), local field potentials (LFPs), and BOLD suggest BOLD is more like LFPs than MUA

BOLD Signals

Simultaneous fMRI and electrical recordings reveal widespread BOLD activation for local excitatory stimulation

from Lee et al., Nature 2010

BOLD Signals

negative BOLD & neural inhibition

from Lee et al., Nature 2010

BOLD Signals

What can you infer from the absence of a BOLD response?

Positive BOLD response?

Negative BOLD response?

Common Pitfalls in fMRI Analysis & Interpretation

Group averaged data- smoothing -> blurring of distinct loci of activation- heterogeneity of activation location across subjects can obscure consistent within-subject results

Common coordinates- reported centroid of an activated cluster can be misleading about it’s extent

[A > C, B !>C ] -> [A > B]- A and B could have same means, different variances

“Activation relative to A or deactivation relative to B”- Can never really tell

Multiple Hypothesis Testing - lack of / under-correcting can lead to spurious type I errors - parametric over-correction can lead to many type II errors