Structural neuroimaging part 2 Flashcards
What is 1. T1 weighted structural MRI?
Very used with fMRI because during tasks, fMRI doesnt have good res and contrast so use this to see structure and better localise source of brain activity
How to acquire: select TR that maximises tissue types but minimise T2 decay
How do we do a quality check before analysis?
• Image artefacts due to removable ( e.g. hairpin) or fixed ( e.g. orthodontic) material
• Finding anatomical abnormalities can be incidental in research study
• Routine pulse sequences = robust to acquisition problems (If trying new sequence or unusual settings, have to make sure quality is good)
What is normalization?
Goal: put all brains in common reference frame despite inter individual variability (make brains comparable)
When look at participants, assume that a pt in one brain is the same as the pt in another but that’s not true unless you normalize bc everyone’s brain has diff shape and size, have to make the brains align
Brains can have diffs in overall size and in size and shape of diff parts so even if participants ae in same positions, pts might not correspond to each other
What is Volume based normalization?
• Preserves 3D volumetric structure
• Translations and rotations (“rigid transformation”) but a lot of variability remains
• More complex transformations to compensate for individual variability
• Reference templates
• Talairach atlas, MNI templates, population specific
• Large inter subject variability –> blurry
What is Segmentation based normalization? (6)
• Separation into tissues
• Based on signal value of voxels (diff tissues are diff colours)
• Using knowledge about where tissue is typically located
• Manual adjustments
• Resulting in tissue probability maps
• Less blurry (better normalisation results)
What is Surface based normalization?
• First: segmentation
• Then: render cortical surface
• Outer edge cortex
• Edge between white matter and grey matter
• 3D point –> 2D position = voxels –> vertices
• Vertex value = weighted combination of signal value of neighboring voxels
• Normalizing the surfaces into common reference frame
How do we visualise the cortical surface?
• Inflated brain
• Shows all activity (including from sulci)
• Straightforward relation to original 3D layout
• Flattening (cuts necessary)
• Flatmaps
• Landmarks e.g. calcarine sulcus and layout of visual cortex
• But… no sub cortical activity
How do we use brain morphometry to analyse images? (2 methods)
• Voxel based morphometry (VBM)
• Normalization: nonlinear registration to group template (info on deformations of voxels = jacobian map ), normalization of group to standard template
• Segmentation results in tissue probability –> multiply with jacobian
• Smoothing before comparison across subjects
• Surface based morphometry
• Cortical (thickness) surface reconstruction per subject
• Alignment across subjects
• Further normalization
What are the statistical approaches for T1 weighted structural MRI?
• Significant differences in morphometry between two groups of participants
• Voxel wise or vertex wise analysis
• Student’s t test on each voxel/vertex
• Correction for multiple comparisons
• Certain problems need to be considered
• Normalization problems
• Total brain volume
What is Voxel based lesion symptom mapping (VSLM)?
On anatomical image, 1st analyse location of lesion (see if each voxel is in lesion)
• Voxel in or outside lesion, correlation with behavioral symptoms?
• Patients with lesion in a voxel more deficit compared to patient without?
Gives info on which brain regions support diff fctions
What is Diffusion tensor imaging (DTI)?
• Pattern of action potentials
• From where and to where?
• Paths = axons
• Noninvasive imaging: DTI
• Large bundles of 1000s of axons
• Axons that start and end in each other’s vicinity, stay together –> white matter pathways or tracts
What is Diffusion weighted imaging (DWI)?
• Diffusion : water molecules move from parts with higher concentration to parts with lower concentration
• Cell membrane barrier –> anisotropy in diffusion
• Based on spin echo EPI and pair of diffusion gradients in particular direction in 3D space
• De-phasing and re phasing (if protons have not moved since 1 st gradient)
• Diffusion along the axons in a tract –> MRI signal loss due to incomplete re phasing when encoding direction aligns with axons
What is Pulsed gradient spin echo (PGSE)?
Spin echo sequence: have exci pulse folowed by 180° refocusing pulse (allows p+ to get back in phase; slow ones catch up to faster ones), p+ get rephased
1st gradient after 1st pulse and 2nd after opposite pulse
What is a Diffusion tensor?
Gradient is in 1 direction, if p+ moves 90° to gradient, wont change MRI signal
Have to use multiple diffusion gradients in diff directions to catch all the mvmts
• Quantify amount of diffusion in each possible direction in 3D space
• Diffusion described by tensor
Use matrix to see mvmt in each voxel
Each vector has direction and magnitude and indicates amount of mvmt in that direction
What 3 metrics do we use for DTI data analysis?
• Mean diffusion (MD): overall amount of diffusion in that voxel, regardless of direction, If have higher overall diffusion, memb is less dense (correlates)
• Fractional anisotropy (FA):
0 = isotropic diffusion (goes in all directions =lly)
1 = diffusion only along main axis (doesnt go in any other directions at all)
Indication of integrity of WM (if memb is degraded, get lower FA bc diffuse in all directions)
• Axial and radial diffusivity (AD and RD):
AD = diffusion along main axis
RD = other directions
