Kalina Christ Off
Kalina Christ Off
Kalina Christ Off
Neuroimaging Tutorial
Kalina Christoff, Ph.D. University of British Columbia
3. Statistical Analysis
Kalina Christoff
Neuroimaging Tutorial
Neuroimaging modalities
Functional Magnetic Resonance Imaging (fMRI) Positron Emission Tomography (PET) Electroencephalography (EEG) Transcranial Magnetic Stimulation (TMS)
fMRI
MRI vs fMRI
Kalina Christoff
Neuroimaging Tutorial
MRI vs fMRI
Spatial resolution
voxel
1 mm
4 mm
8 mm
Temporal resolution
TR: Time of Repetition HRF: Hemodynamic Response Function
TR = 3 sec
TR = 1.5 sec
TR = 0.75 sec
Kalina Christoff
Neuroimaging Tutorial
The vascular system supplies blood containing oxyhemoglobin to active regions of the brain The influx of oxygenated blood to regions that are active reduces the local concentration of deoxyhemoglobin which increases BOLD signals
Kalina Christoff
Neuroimaging Tutorial
Kalina Christoff
Neuroimaging Tutorial
No radioactivity involved Uses magnetic field (measured in Tesla, e.g. 3T Scanner) Works at the level of protons Measures protons protons spin relaxation time Can be made sensitive to water content or oxygenation level
Hair grip
Kalina Christoff
Neuroimaging Tutorial
Cannot bring metal into the scanner Environment is generally quite noisy Head motion can be severely detrimental
Head Motion
Kalina Christoff
Neuroimaging Tutorial
TaskTask-correlated motion Watch out for: activations that trace the outline of the brain
Preventing Motion
Restraining the head Asking participants to remain still PrePre-training using a mock scanner
Ti m
Kalina Christoff
Neuroimaging Tutorial
Motion correction
Original timetime-series
time
Corrected timetime-series
Spatial normalization
Another step of data prepre-processing Allows us to combine data across individuals Matches each individual brain to a template
Kalina Christoff
Neuroimaging Tutorial
Original image
Spatially normalised
Spatial Normalisation
Template image
Deformation field
space
1. Data Acquisition
2. Data Preprocessing
3. Statistical Analysis
Kalina Christoff
10
Neuroimaging Tutorial
voxel
task
Similar to simple averaging across individuals Results are not generalizable to the population
(outliers can strongly influence averages)
Kalina Christoff
11
Neuroimaging Tutorial
Displaying results
The chosen statistical threshold determines what is displayed (and classified) as activated
Displaying results
too lenient acceptable preferable
Kalina Christoff
12
Neuroimaging Tutorial
Is there any evidence for motion artifacts (false activations)? Was an appropriate template used, if normalization was applied? applied? Was random effects analysis performed at the group level? Was the statistical threshold used acceptable?
Kalina Christoff
13