Structural MRI

Question 1

What is structural MRI dataset?

Answer 1

It is a matrix with dimensions x,y,z

Question 2

What is a voxel? What does it represent?

Answer 2

A voxel is the single entry in a 3D image (cf. pixel). It represents a single cell in that matrix

Question 3

What is an isotropic voxel?

Answer 3

A voxel that measures the same in each direction

Question 4

What (anatomical) planes are used in MRI?

Answer 4

• axial (horizontal, transverse, divides into superior and inferior sections)
• coronal (frontal, divides into dorsal and ventral sections)
• sagittal (longitudinal, divides into right and left sections)

Question 5

What is the most common resolution in contemporary MRI?

Answer 5

contemporary acquisition resolution is ~1 mm3

Question 6

What two approaches are there in classic lesion analysis?

Answer 6

• ‘lesion-defined’ approach
• ‘behavior-defined’ approach

Question 7

What are the princples of ‘lesion-defined’ approach? What are its pros and cons?

Answer 7

behavioral performance of a group of patients with a common area of injury сompared to that of a control group / another patient group
* + good for assessing the functional roles of particular regions of interest
* - loss of information if ROI contains multiple subregions that each contribute to behavior
* - regions outside the ROI that are part of a distributed functional network may be overlooked

Question 8

What are the princples of ‘behaviour-defined’ approach? What are its pros and cons?

Answer 8

patients are grouped according to whether or not they show a specific behavioral deficit
lesion reconstructions are overlaid to find common areas and compared to lesion overlays from patients without the deficit
* + effective in identifying brain regions that may contribute to a cognitive skill
* - when behavioral data are continuous → cut-off needed → information reflecting varying degrees of performance can be lost

Question 9

What is an alternative to classic analysis? What are its principles and pros?

Answer 9

Voxel-wise lesion-symptom mapping:
* mass-univariate (voxel-by-voxel) statistical analysis
* for each voxel, patients are categorized according to whether they did or did not have a lesion affecting that voxel
* behavioral scores are then compared for these two groups, yielding a t-statistic for each voxel

Pros:
* does not require patients to be grouped by either lesion site or behavioral cutoff a priori
* makes use of continuous behavioral and lesion information

Question 10

What are existing approaches in segmentation?

Answer 10

Manual segmentation
* ITK-Snap (hand-selected regions)
Automated segmentation
* FSL FIRST (subcortical)
* FreeSurfer (subcortical+global)
* FSL SIENA(X) (global tissue volumes)

Question 11

What are contemporary applications of structural MRI?

Answer 11

• Lesion-Symptom Mapping
• Volumetry
• Voxel-based morphometry
• Cortical thickness

Question 12

What are the pros and cons of manual segmentation?

Answer 12

Pros:
* Remains the gold standard
* Ideal for delineating structures with intricate anatomy/multiple subregions
* Well-suited for smaller studies with focused hypothesis
* Biologically and anatomically meaningful

Cons:
* Labor-intense: impossible for large studies (> 1.000 scans)
* Requires expert anatomical knowledge
* Requires at least two blinded tracers to avoid bias
* Intra-rater variability
* Inter-rater variability
* Inter-protocol variability

Question 13

What are the pros and cons of automated segmentation?

Answer 13

Pros:
* Replaces manual segmentation for most applications
* Substantially faster (large datasets)
* Higher reliability (no intra-rater and inter-rater variability) — you measure the same result in different trials
* Validity — you measure what you want to measure
* Standardized
* Agreement between manual and automated approaches is continuously improving

Cons (evidence from single studies):
* Freesurfer can overestimate total hippocampal volumes
* Problems with accurately detecting boundaries between hippocampus and neighboring structures
* Differences in segmentation outcomes with regard to age effects and hemispheric asymmetry

Question 14

What is whole-brain volumetry with FSL SIENA(X) particularly useful for?

Answer 14

• both single-time-point (cross-sectional) and two-time-point (longitudinal) analysis of brain change
• particularly useful for the estimation of atrophy (volumetric loss of brain tissue)

Question 15

The volume of which brain tissues can be assessed with FSL SIENA(X)

Answer 15

• Whole-brain volume
• Peripheral gray matter volume
• Gray matter volume
• Ventricular CSF volume
• White matter volume

Question 16

What are new methodologies in structural MRI?

Answer 16

Existing methods are based on notion of size (volume, thickness), new methologies are trying to move to shape measuring

=> Shape-related measures of brain morphology:
* local gyrification index (GI): how much of the cortex is buried within sulcal folds
* regional fractal dimensionality (FD): measures structural complexity, how irregular the shape of an object is
* brain shape development in infants

Question 17

What are the features of cortical surface models?

Answer 17

• Triangle mesh (“finite element”) cortex split in ~ 300 000 triangles
• Point of triangle intersections: vertex
• Surface of the triangles: face
• XYZ coordinates at each vertex
• area, distance, curvature, thickness, …

Question 18

What are the pros and cons of cortical thickness estimation with FreeSurfer?

Answer 18

Pros:
* automated, continuous, whole cortex
* processing and measurement respect cortical topology
* direct, biologically meaningful measure in millimeters
* surface-registration may increase sensitivity

Cons:
* heavy post-processing (6-24 hours/scan)
* dependent on classification
* manual corrections often necessary
* limited to (neo)cortex

Question 19

What models of the human cortex are there?

Answer 19

• MRI →
• Cortical surface models with all gyri and sulci →
• inflated cortex image (sulci are brought to the surface and can be mapped) →
• flattened, completely normalised cortex image

Question 20

What is characteristical about the human cerebal cortex?

Answer 20

• it is an outer layer of gray matter
• it is 1-5mm thick
• it is highly folded
• it is 2-dimensional, but embedded in 3 dimensions

Question 21

How is cortical thickness estimated?

Answer 21

It is the shortest distance between white matter surface and pial surfaces

Question 22

What diseases lead to cortical thinning?

Answer 22

Frontotemporal dementia (FTD): brain disorders that are characterized by a degeneration of frontal and temporal lobes
* Behavioral changes
* Speech and language impediments
* Parkinson-like motor symptoms

Amyotrophic lateral sclerosis

Question 23

What diseases are characterised by change in the brain volume?

Answer 23

• Neuromyelitis optica: WM volume is decreased
• Anti-NMDA-receptor encephalitis: cortical brain loss, reduced hippocampus volume

Ofc, also, AD, PD, HD, MS

Question 24

What is voxel-based morphometry? How is it implemented?

Answer 24

• voxel-wise analysis of the local concentration of gray matter
• characterizes local differences in gray matter topography, while discounting large-scale differences in anatomy
• align images globally and compare GM likelihood at each voxel

Question 25

What are necessary steps in MRI processing?

Answer 25

Segmentation (into GM, WM, CSF):
* often combined with intensity normalization
* classification commonly combination of intensity-based clustering informed by spatial priors
* removing non-brain classes
* intensity distribution (from T1 scans) based clustering informed by spatial priors

Normalisation (from native space to standard space): linear vs. non-linear; is based on a template

Smoothing:
* makes each voxel more similar to its neighbors (weighted means)
* smoothing kernels are 4-16 mm full width at half maximum (FWHM)
* smoothing helps satisfy the assumption of Gaussianity of residuals
* helps accommodate inter-individual differences in local anatomy
* makes the analysis sensitive to the kernel size, such that very small differences are disregarded

Question 26

What are some common types of interpretational uncertainty?

Answer 26

1. not detecting a folding
2. wrong estimation of thickness
3. misclassification: segmentation of brain tissue types
4. misregistration