sMRI Flashcards

Question

What does VBM measure

Answer 1

Voxel-wise grey matter concentration/probability. Reflects relative amounts of GM/WM per single voxel, allowing for construction of whole-brain grey matter volume maps, which are then compared across subjects

Answer 2

1. T1 image acquisition 2. **Segmentation** of image into GM/WM/CSF so that specific tissue type can be extracted and compared quantitatively 3. **spatial normalization** of the images to a standard brain template 4. **smoothing** the extracted tissue type with Gaussian kernel 3. **Voxelwise statistical testing** of individual density maps, to see which voxels/clusters show significant differences in tissue concentration between groups or conditions. 4. Interpret the results: output is a statistical parametric map showing voxels where specific tissue type concetration differs significantly

Answer 3

VBM entails classifying brain into WM, GM, CSF and background, extracting Gray Matter to produce binary GM map, then smoothing the extracted tissue type with a Gaussian kernel to produce map of GM density with values ranging from 0 to 1 representing the amount of gray matter within a local neighborhood as determined by the blurring kernel. ## Footnote SEE THE DIAGRAM IN THE PAPER!!!!!!!!

Answer 4

Strengths: Whole brain, ROI-free analysis, good for studying disease, development and training-related plasticity. Good spatial resolution. Limitations: Noisy, sensitive to registration errors and tissue mixing and motion

Answer 5

DBM analyzes how much the brain had to stretch or compress during normalization. Jacobian determinant describes where tissue expands or contracts.

Answer 6

Expansion = Higher local volume relative to the template. Contraction = Lower local tissue volume/atrophy.

Answer 7

It directly measures how much each area had to stretch or shrink over time.

Answer 8

By dividing the cortex into inner surfaces and outer surfaces, these algorithms measure the distance between two surfaces and thus thickness.

Answer 9

Surface-Based Morphometry (SBM) is a set of methods that analyze the cortex by treating it as a 2-dimensional folded surface, not a 3-dimensional volume.

Answer 10

Volumetric approaches (VBM/DBM) try to match the whole brain volume to a template. This causes problems: Sulci differ across people Gyri shift in position Folding patterns have no one-to-one homology VBM smoothing mixes tissue across sulci Volumetric registration cannot align cortical surfaces accurately SBM solves this by working directly on the cortical sheet, which is topologically consistent across individuals.

Answer 11

1. tissue segmentation and spatial normalization: locate the grey-white matter boundary 2. surface reconstruction: making 3D mesh model of inside and outside surface 3. surface inflation: explanding mesh so sulci can be visualized 4. surface registration: align surfaces of several individuals 5. morphological analysis: measuring parameters like cortical thickness, area, volume, etc. 6. statistical analysis: GLM models used to ind sig differences in morph measures between groups

Answer 12

- difficulty in accurately delineating lesions with blurry boundaries - inflated false positive rates due to violations of statistical assumptions due to individuals' unique anatomical features - doesn't measure internal volume of structures, like VBM

Answer 13

It uses sulcal/gyral landmarks for registration, which are more consistent across individuals than voxelwise intensity patterns, improving spatial correspondence (e.g., Broca’s alignment improves dramatically).

Answer 14

Thickness and area have distinct genetic and developmental trajectories, and disease affects them in dissociable patterns, making SBM highly sensitive to subtle, region-specific changes. eg: Schizophrenia → abnormal thinning in DMN regions Depression → widespread thinning in cingulate/fronto-parietal areas Autism → early increased surface area, atypical folding ADHD → delayed cortical maturation SBM can detect these subtle, spatially specific abnormaliti

Answer 15

Folding patterns, thickness, and area show high heritability; SBM studies (family/twin designs) reveal genetically driven cortical development and X/Y chromosome effects.

Answer 16

Longitudinal studies, growth, atrophy, development

Answer 17

cortical thickness surface area gyrification sulcal depth

Answer 18

VBM is sensitive to local tissue composition changes, while DBM captures global and regional shape differences. Using both reduces interpretational ambiguity.

Answer 19

The cortex’s highly folded geometry makes voxelwise comparisons inaccurate. By dividing the cortex into inner and outer surfaces and measuring the distance to give thickness, SBM reconstructs white–gray and pial surfaces to measure cortical thickness, area, and folding. Uses cortical surfaces instead of 3D voxels → better anatomical alignment.

Answer 20

The distance between the white–gray boundary and the pial surface, computed along the cortical normal vector.

Answer 21

Uses diffusion MRI to look at **large white matter tracts**; it has applications for studying structural connectivity, development and white-matter disorders

Answer 22

The thermally driven diffusion of water molecules, which reveals restrictions imposed by microstructure (axons, membranes, myelin, extracellular space). ## Footnote Gives information about microenvironment

Answer 23

Studies microstructure within singular voxels; how the contents of a voxel are distributed (for white matter!) mostly done using diffusion MRI (dMRI) consists of: - diffusion MRI - magnetization transfer - quantitative susceptibility mapping - quantitative t1/t2 mapping

Answer 24

FA = fractional anisotropy; represents how elongated shape of ellipsoid is (a model of the diffiusion of the water molecules) FA = 0: Isotropic diffusion (CSF, highly disorganized tissue). FA = 1: Purely linear diffusion (perfectly aligned fibers). ## Footnote REstriction of the water molecules gives information about the organization of the tissue inside an imagine voxel

Answer 25

In white mater, water molecules tend to flow along the direction of nerve fibers. Diffusion motion is restricted by denser structures within tissues. If more nerve fibers go in the same direction, water molecules will flow into that direction so the FA will be closer to one. If the fibers are disorganized and go in multiple directions, the FA will be closer to 0. So, the restriction of the water molecule diffusion gives info about the organization of the tissue inside an imagine voxel.

Answer 26

- indirect measurements; water motion not exactly equatable to cell structure and is highly sensitive to motion/artifacts - Not biologically specific; confounds fiber crossings, diameter, dispersion.

Answer 27

It reflects the interaction between free water protons and protons bound to large macromolecules such as proteins.

Answer 28

MT ratio: by applying an off-resonance MT saturation pulse, selectively saturating macromeulcule-bound proton pool. Through cross-relaxation / magnetization exchange, saturation transfers to the free-water proton pool. This reduces the measured MRI signal from the free-water pool. The MT ratio is a proxy for myelination of tissue damage: Higher MTR → greater macromolecular content (e.g., more myelin). Lower MTR → demyelination, inflammation, edema, or tissue damage.

Answer 29

Directly measures the absolute relaxation times to infer tissue properties like myelin, iron and water content.

Answer 30

shorter t1 = more mylein/macro longer t1 = more water shorter t2 = denser tissue longer t2 = increased water content

Answer 31

Uses phase information to measure local magnetic susceptibility, reflecting iron, myelin, calcium content. ## Footnote More direct and specific than T1/T2-based measures

Answer 32

**T1 relaxation:** how quickly proton's magnetization realigns with main magnetic field after relaxation **T2 relaxation:** how quickly the protons lose **phase coherence** with each other due to variations in local tissue environment (gray matter has longer T2 than white matter)

Answer 33

Tissues with dense macromolecules (e.g., fat, myelin-rich white matter) have short T1 because they provide efficient energy exchange, allowing rapid recovery of Mz. Exam trick: Myelin shortens T1 → brighter WM on T1 images.

Answer 34

Dephasing (T2) requires only tiny microenvironmental field differences, whereas T1 requires slower energy exchange with the lattice. Therefore phase coherence is lost faster than longitudinal magnetization regrows. T2 ≤ T1 always.

Answer 35

Through careful choice of repetition time (TR) and echo time (TE): Short TR + short TE → T1-weighted Long TR + long TE → T2-weighted Adjusting TR and TE amplifies differences in recovery (T1) or dephasing (T2) between tissues.

Answer 36

more myelin or macromolecules

Answer 37

increased water content

Answer 38

denser tissue

Answer 39

increased water content

Answer 40

macro/meso: large-scale mpas of structure, but only indirectly reflect tissue composition micro: specific to looking at bio properties (myelin, axons, iron) at tissue-level but remain model-dependent, and require longer, complex scans and carefui interpretation Combination of these scales is future of neuroanatomy.

Answer 41

Motion blurs boundaries, lowering apparent GM intensity during segmentation → the algorithm mislabels GM as WM/CSF, mimicking atrophy. Motion effects mirror true neurodegeneration patterns.

Answer 42

Using comfortable padding and optimized head fixation Using multi-echo MRI sequences: scanner collects several images at diff echo times during each scan then combines these to obtain a more stable and less distorted image Prospective motion correction (navigators, external tracking) Post-hoc: exclusion, motion regressors, outlier replacement

Answer 43

children elderly psychiatric patients neurologically impaired Since these groups differ systematically from controls, motion can perfectly confound group effects.

Answer 44

Variations in the static magnetic field caused by air/tissue boundaries, especially near orbitofrontal and temporal regions (areas near air spaces) ## Footnote This may lead to parts of the brain looking stretched or compressed

Answer 45

Shimming: applying small corrective magnetic fields to make the main magnetic field more uniform, but cannot fully fix distortions in regions where magnetic changes are very sharp (like the frontal sinuses or temporal lobes)

Answer 46

The dura has similar T1/T2/PD properties to gray matter at typical resolutions; segmentation algorithms confuse them, creating artificially thick cortex. ## Footnote This may lead to confused segmentation algorithms: dura contamination

Answer 47

1. inspect software-generated cortical segmentation overlays, visually checking overlay of the image near the CSF and dura. 2. Multiecho sequences. which capture T2 differences, as the dura loses its signal faster than gray matter; if you slightly increase TE, dura becomes darker and easier to tell apart from grey matter.

Answer 48

no one-to-one anatomical mapping of sulcal patterns across individuals. Folding variability of cerebral and celebellar cortices prevents perfect alignment (**registration**) between brains. ## Footnote Even worse with volumetric registration used in VBM

Answer 49

Use surface-based registration instead of volumetric registration → Aligns brains based on cortical folding patterns rather than voxel intensity, giving much better anatomical correspondence (especially for cortex). Visually inspect registrations → Check overlays for mismatched sulci/gyri, tissue boundaries, stretching, or compression artifacts. Model misregistration as informative variability → Treat consistent alignment errors as reflecting real anatomical differences rather than noise (Lerch et al. explicitly note this). Improve alignment using folding-based or sulcal-shape mapping → Directly align cortical features (sulcal depth, curvature), which improves localization of effects (e.g., Broca’s alignment improves drastically).

Answer 50

Integration of genetics, epidemiology, environment, and MRI to understand brain variation across populations, rather than patient/control samples.

sMRI Flashcards

(76 cards)