fMRI Methods

Lecture 9 – The brain at rest

The brain never rests!

Brain takes up 20% of the metabolites in the body during rest.

Energy costs

Energy consumption measured with PET (radioactive glucose consumption) during 4 conditions.

Which is rest and which is task?

Energy costs

Brain is always at max power: similar costs in all conditions

Energy costs

Energy cost breakdown for running the cortex.

Most of the energy is spent on neural signaling.

Information encodingGiven that energy consumption is almost constant:

Only relatively small populations of neurons respond strongly at any given time.

Relatively local, transient changes in neural activity represent stimuli evoked responses.

Hence the importance of: Attention and neurovascular coupling

Brain is never static!

Connected neural populations tend to synchronize and oscillate together.

EEG Alpha (10 Hz)

Over occipital electrodes.

Different time scales of synchronization

Close eyes

Slow changes

Similar fluctuations in fMRI signal during rest

Nir Y., Neuroimage (2006)

Structure of rest activity

Spontaneous neural activity during rest and sleep may be random over time, but it’s not random over space.

Large neural populations are synchronized.

What are the characteristics of this activity?

How can we study them? No experimental structure?

Are there differences between rest, sleep, anesthesia?

Local spatial structure

Anesthetized cat primary visual cortex

Orientation columnar map – voltage sensitive dye imaging

Kenet, Nature (2003)

Local spatial structure

Many occurrences of strong correlation between spontaneous and evoked columnar maps.

Local spatial structure

Correlation between spontaneous “snapshots” and particular orientation columnar maps.

Visual cortex randomly moves from one map to another in the absence of stimulation (relatively quick transitions).

Large spatial structure

Cortical areas with similar functionality (e.g. right and left auditory cortex) show strong and selective correlations.

Very slow hemodynamic changes over time.

Functional connectivityAreas that are connected anatomically because of shared

functionality will be active together.

In reality this is just correlation – problematic interpretation

Inter-hemispheric correlations

Strongest correlations are between corresponding locations in the two hemispheres

Default mode system

Three areas that show reduced activity during “external” tasks (e.g. visual, auditory, somatosensory stimulation)

Default mode system at rest

Is the brain separated into two general antagonistic networks?

Measuring default mode system correlations during rest.

“Intrinsic” ROI“Extrinsic”

ROI

Source of fMRI correlations

Several “less interesting “ sources contribute to an fMRI signal. Are they driving correlations during rest?

Source of fMRI correlationsSimultaneous fMRI and electrophysiology:

Shmuel et. al. HBM 2008

Source of fMRI correlations

Significant correlations between neural activity and BOLD during rest…

Source of fMRI correlations

Similar inter-hemispheric correlations in epilepsy patients

Nir Y., Nat. Neurosci. (2010)

Source of fMRI correlations

Correlations during rest correspond to anatomically connected areas that are commonly active during task.

Fox, Nat. Rev. Neurosci. (2007)

Patient lacking corpus callosum

Inter-hemisphericcorrelations disappear after section of corpus callosum.

Spontaneous and evoked activityThere’s a lot of variability in the neural response to the same repeating visual stimulus.

Arieli, Science (1996)

Spontaneous and evoked activity

The spontaneous state right before stimulation predicted the amplitude of the response 42 ms later.

Spontaneous and evoked activity

Changing the spatial response pattern in a predictable way.

Is the evoked response to atrial the sum ofspontaneous and evokedactivity?

Motor response variability

Fox, Nat. Neurosci. (2006)

Right hand button press LH ROI

Functionally correlated RH ROI

Motor response variability

LH responses RH responses

Subtracting out RH spontaneous activity, allowed reduction of trial by trial evoked response variability.

Diagnosis by rest scans

If resting state correlations represent the functionality of the brain, abnormal correlations may represent abnormal functionality and enable an easy form of diagnosis.

Build a distribution of “normal” correlation values and test whether these deviate in particular neurological or psychiatric disorders.

Alzheimer’s disease

Weaker resting state correlations between “default mode” network (precuneus area) and hippocampi.

Fare comparison?

Control Alheimer’s

Greicius, PNAS (2004)

Alzheimer’s disease

Direct comparison:

Voxels that showed stronger correlation with PPC timecourse in controls versus Alzheimer’s patients.

Comma

Patients in reduced states of awareness show decreased inter-hemispheric correlations.

Autism

1-3 year old children with autism or language delay, during natural sleep.

Strong inter-hemispheric correlations are already evident at extremely young ages.

Autism

Decreased inter-hemispheric correlations in language areas

Autism

Inter-hemispheric correlation strength predicts language ability and autism symptom severity.

Rest activity and development

Visual system development (e.g. ocular dominance) before birth (i.e. before visual experience)

Hebbian learning during rest?

In the absence of externally evoked responses?

Summary

Spontaneous brain activity is the largest and least understood component of brain function.

Different temporal and spatial scales of organization.

To the lab!