Transcript Functional Magnetic Resonance Imaging (fMRI)

What we can do and what we
cannot do with fMRI
NIKOS K. LOGOTHETIS
Goals of Article
 Provide an overview of fMRI in its current state
 Examine relationship between haemodynamic signals and neurons
 Explain how constraints on neuroimaging data interpretation
fMRI Research
 Since first publication in 1991, over 19,000 papers have been published referencing fMRI
 Since 2007 approximately eight fMRI related papers are published a day
 43% - cognitive anatomy associated with task or stimulus
 22% - ROI studies examing physiological properties
 8% - Neuropsychology
 5% - Properties of fMRI
 Remainder – Plasticity, drug action, experimental designs and analysis methods
Advantages of fMRI
 Noninvasive procedure
 Ever-increasing availability
 Relatively high spatiotemporal resolution
 Provides data for entire brain while subject
carries out a task
CREDIT: LOTFI B. MERABET/MASSACHUSETTS EYE AND EAR INFIRMARY
Shortcomings
 Haemodynamic-based (surrogate signal)
 Spatial specificity and temporal response are subject to physical and biological constraints
 Reflects neuronal mass activity
A Brief Overview of fMRI
 Measures tissue perfusion, blood-volume
changes, or changes in concentration of O2
 B0 : Main static field
B1 : Field generated by excitation pulses
 Blood-oxygen-level-dependent (BOLD) is most
common in human neuroimaging
Use subtraction methods to analyze data
 Spatial specificity increases with magnetic field
strength
Studies look at activity of functional subunits
and instances of joint or conditional activation
 87% of studies use conventional gradient-echo
echoplanar imaging (GE-EPI). Others use variants
of spin-echo echoplanar imaging (SE-EPI)
 Use voxel-based analysis (MVPA on rise)
Neurons and Voxels
 Approximately 90,000-100,000 neurons
under 1 square mm of cortical surface
 Primary visual cortex has twice amount
 fMRI resolution
 9-16 square mm
 5-7 mm slice thickness
 Average voxel size: 55 cubic mm
 Each voxel contains approximately 5.5
million neurons
 With 100 billion neurons in the brain, that’s
approximately 18,000 voxels (130,000?)
Contents of a Single Voxel
Differences in gradient and spin techniques
What do Activation Maps Represent?
 Does activation of an area mean it is involved  Microcircuits act as drivers, transmitting stimulus
in the task at hand?
info, or modulators, adjusting sensitivity/specificity
 Input-Local Processing-Output Model and
brain connectivity is mostly bidirectional
 Cognitive capacities are more likely to induce large
changes in the fMRI signal than the sensory signals
 Excitation-Inhibition Networks (EIN)
 Changes in E-I balance affect metabolic demand
and subsequently increase local cerebral blood flow,
however this does not necessarily mean excitatory
activity
 Can be thought of as microcircuits
 Final response determined by the sum of all components
 Net excitation or inhibition might occur when afferents drive
overall E-I in opposite directions
 Responses to large sustained input may vary
Excitation-Inhibition and fMRI Data
Does Inhibition cause a metabolic increase?
 fMRI studies using dichotomous systems and electrophysiological evidence
 Inhibitory neuron density is 10-15 times lower than excitatory and ATP is not directly
expended in the uptake of Cl2
 Modeling inhibition is unlikely straightforward and subject to ROI
Blood Oxygen Level Dependent (BOLD)
 Most common technique for human neuroimaging
 Sense differences in magnetic sensitivity of Oxyhemoglobin and Deoxyhemoglobin
 Types of EIN activity
 Single-unit (isolated neurons near electrode tip)
 Multiple-unit (small neuron populations 100-300mm radius)
 Perisynaptic (within 0.5-3mm radius of electrode tip)
 Multiple unit and local field potentials (LFP) separation by frequency band
 LFP shortcoming is its ambiguity
Conclusion
 fMRI signals cannot easily differentiate between function-specific processing and
neuromodulation, bottom-up and top-down signals, and may potential confuse excitation and
inhibition
Despite its shortcomings, fMRI is still one of the best tools for relevant research
 Suggested: Multimodual approach is necessary to properly study function and dysfunction
The neural Signature of satiation is associated
with ghrelin response and triglyceride metabolism
XUE SUN, MARIA VELDHUIZEN, AMANDA WRAY, IVAN DE ARAUJO,
ROBERT SHERWIN, RAJITA SINHA, DANA SMALL
Purpose of Study
 Use fMRI to measure brain response to palatable food under different caloric intake conditions
 Look for internal indicators of metabolic state
 Look at differences between perception of food availability
Experimental Design
Measurements from Blood Samples
Example Data Collected
Conclusion
 Satiation-induced brain response to palatable and energy dense food are influenced by
circulating ghrelin and triglycerides
 Fixed portions and volitional meal termination may influence the brains response
 Suggested: examine relationship between control over meal termination and self-control in
making food-related decisions
24/11
Why BOLD-fMRI????
Magnetic Resonance Imaging (MRI): clearly identify the olfactory
bulb (OB) layers.
Functional MRI (fMRI): map the activity patterns evoked by
odorants.
Blood oxygenation level dependent- fMRI (BOLD-fMRI):reveal
dynamics of neuronal responses and properties of adaptation.
25/11
Materials & Methods
Animals
18 Adult Sprague Dawley rats (250-230g)
Tecniques
- BOLD-fMRI
- Local field potential (LFP)
Odorants used
Iso-amyl acetate (IAA)
Octanal (OCT)
26/11
RESULTS (Fig.1)
OB layers reordered simultaneously
LFP recordings
Transient sharp potential
Site for potential reversing
LFP
Spikes from MCL
Objective
>300 Hz
MRI T2-W
Adapted from Henry Lütcke et al., Front. Neural Circuits 2010
Recording sites
GL=Glomerular layer
MCL= Mitral cell layer
GCL= Glomerular cell layer
High signals in the mitral cell layer (MCL)
27/11
RESULTS (Fig.2)
Activity maps induced by odorants stimulation
Threee dimensional t-value
High [IAA] = 50%
Low [IAA] = 5%
NDP (Normalize dot product)
Low [OCT] = 5%
**p<0.01
Different activated patterns are evoked by different odorants.
28/11
RESULTS (Fig.3)
Layer-depend BOLD responses in the OB
Global activity maps
Intensity of signal
n=9
n=9
n=8
Absolute intensity of the signals in a given layer was affected significantly by odor type and
concentration.
29/11
RESULTS (Fig.4)
Local Field signals after IAA exposure
Traces
Normalized time-frequence graphs
All neural activity is high in GCL after odor stimulation.
30/11
RESULTS (Fig.4)
LFP signals after multiple odorants perception
Neuronal oscillations
Frequency bands
**p<0.01
Confirmation of the strongest responses GCL level.
31/11
RESULTS (Fig. 5)
Relationship of temporal response profiles of BOLD and LFP
signals in the OB
Correlation analysis
Software R
p<0.01
Similar responses from BOLD and LFP analysis.
32/11
RESULTS (Fig. 6)
Temporal features through LFP and BOLD tecniques
BOLD vs LFP bands
LFP signals
Worst correlation
33/11
CONCLUSIONS
BOLD signals (Indirect measure)
• local neuronal activity
Effected by
Hemodynamics
Metabolism
Neuroanatomy
Well correlated with
LFPsignals (direct measure)
34/11
35/16
Aim of the study
Observation of different areas of the brain involved in odor
perception using fMRI with BOLD effect on awake animals.
36/16
Materials & Methods (1)
- Animals
Adult ♂ Sprague Dawley rats
n= 58
- Nose poke assay
Behavioral motivation to odors
Pour odorants (1ml of 1% of solution (odorant/water):
Benzbenzaldehyde (almond odor),
Iso-amyl acetate (banana odor)
Methyl benzoate (rosy odor)
Limonene (citrus odor)
Elevated circular wooden platform
Adapted from: http://www.med-associates.com/
37/16
Materials & Methods (2)
- Imaging on awake animals
Blood oxygenation level dependent functional magnetic resonace imaging (BOLD-fMRI)
Adapted from: Ferris et al., Front.Neuro. 2014
http://www.youtube.com/watch?v=W5Jup13isqw
Pour odorants:
Benzbenzaldehyde (almond odor)
Iso-amyl acetate (banana odor)
Methyl benzoate (rosy odor)
Limonene (citrus odor)
Commercial odorants:
Penaut oil
Standard rat chow
38/16
RESULTS(Fig. 1)
BOLD activation pattern for the entire brain
Benzhaldehyde and penaut oil increases change in BOLD signal.
39/16
RESULTS
Neuronal activation described by 3D volume renderings (BOLDfMRI) in:
 Primary olfactory system (POS)
 Involved in taste and smell.
 Associate disorders: i.e. Parkinson's disease and Alzheimer's disease.
 Cortical circuit of Papez
 Involved in emotional reactions and in part releated to memory.
 Associate disorders: i.e. Alzehimer’s desease and Parkinson’s desease.
 Hippocampus
 Involved in short-term and long-term memory, spatial navigation.
 Associate disorders: i.e. Alzehimer’s disease and Temporal lobe epilepsy’s desease.
 Amygdaloid system
 Involved in processing of memory, decision-making and emotional reactions.
 Associate disorders: i.e. autism and epilepsy’s desease.
40/16
RESULTS (Fig. 2)
Odorants produce activation in the POS
n=9/odorant
almond odor
banana odor
rosy odor
citrus odor
Benzaldheyde increases significantly neuronal activity.
41/16
RESULTS (Fig. 2)
BOLD signals intensity in the POS
Positive BOLD response
Neurons start working and using up blood
For each odrant: gradual increase of BOLD signals intensity.
42/16
RESULTS (Fig.3)
Odorants produce activation in cortical circuit of Papez (1)
almond odor
banana odor
citrus odor
rosy odor
Roubustly activation to the smell of almond (Benzaldehyde).
43/16
RESULTS (Fig. 3)
Odorants produce activation in cortical circuit of Papez (2)
The activation is significantly increases by the smell of almond.
44/16
RESULTS (Fig. 4)
Odorants produce activation in the Hippocampus
almond odor
banana odor
citrus odor
rosy odor
CA1, CA3, DG and subiculum are significatly activated by almond smell.
45/16
RESULTS (Fig. 5)
Odorants produce activation in amygdala
Anterior, lateral and posterior nuclei are partially activated by odorants.
46/16
RESULTS (Fig. 6)
Nose poke assay: Odor preference
Rats prefer the odorant benzaldheyde (almond).
47/16
DISCUSSION
• Benzaldehyde (almond) smell extends beyond the olfactory
bulb
» Activation of the anterior nuclei of the Thalamus
cornerstone of the neural circuitry of emotion
• Rats have genes that code for odorants receptors in the
olfactory bulb
» M71 is a benzaldehyde-sensitive odorant receptor
48/16
CONCLUSION
BOLD-fMRI can be used
- On awake animals (avoiding in this way anesthesia artifacts)
- To idenfy neural circuits that have an innate preference for food
odorants
- Activation of lymbic pathways, hippocampus and regions of the cortical
Papez circuit in response to the odorants associated with high
calorie/protein sources
49/16