Summary MRI BMS24 MSc Biomedical Sciences Nijmegen

MR Basics:


Structural MRI  Describes shape, size and integrity of WM and GM in the brain. To indicate brain
structures. It visualise based on water and fat, not direct axons or neurons can be visualized.

Useful for highlighting lesions and pathologies

- FLAIR  suppresses signal from fluids like CSF to highlight lesions
- DWI  highlights areas of restricted water movement (strokes)
- fMRI  detects change in blood flow which is used to study brain activation during tasks or
at rest

T1- weighted  most often used to visualize WM and GM (white, gray)

T2 – weighted  gets signal from CSF (white) and less from GM, WM (black/gray)

Bias

- Hardware related
o Signal to noise ratio  higher is better
o Contrast to noise ratio
- Motion

MRI analysis

1. Brain extraction
2. Segmentation on tissue OR structure
3. Registration alignment  making a common standard brain out of many brain type



Diffusion MRI  measures water/CSF/blood/microstructure directionality and integrity in WM. Gets
info on anatomic connectivity and can go in many directions when scanning

Tensor = 3 dimensional ball that indicates how much water can diffuse in space of 3 dimensions

Strokes can be identified as following: because of the bleeding the tissue dies what makes CSF starts
flowing around  can be identified with diffusion MRI



Image registration:

 All brain images combined into 1 image
 Essential for group study, fMRI or diffusion
 Correction for motion
 Quantifying structural change
 Structural analysis VBM

Brain extraction = elimination of non-brain tissue

- Estimating the transformation and finding the transformation
- Resampling is applying a transformation, what means a new modified image that only
reduces image quality.

, Image spaces

- Reference coordinate system for reporting and describing
- In group study  registered to standard space
- Talairach and Tournoux  based on post mortem brain  is a coordinate system that maps
the location of brain structures
- MNI template  standard image based on non linear group average



FSL tools can move between different spaces

Different types of coordinate systems

- Voxel coordinates  not aligned with anatomy, only with voxel
- Standard space coordinates  axes aligned with anatomy

Spatial transformations: image transformed to become aligned

DOF = degrees of freedom

 Rigid body 6 DOF  3 rotations and 3 translations, within subject motion
 Non-linear 12 million DOF  high quality image, local changes can be made in image
 Affine 12 DOF  Rigid body + 3 scaling and 3 skews/shears, need to initiatlize non linear,
lower quality imaging
 Linear transformation  makes image taller or thinner

Cost function = when are we good, goodness of alignment.

Interpolations: finds intensity values between grid points  how to voxel value in new images

- Nearest neighbour  assume missing voxels are value of a neighbour voxel
- Trilinear  fills missing voxel by using average of voxels around
- Spline
- Sinc
- K-space methods



Physics:

Protons and neurons have their personal spin  nuclei with odd numbers of protons and neutrons
 net magnetic moment

Hydrogen

Properties of hydrogen

1. Resonance
a. Oscillation about magnetic field before settling
b. Time it takes is the resonance frequency
2. Excitation
a. Another magnetic field can get the ion out of its settling phase after resonance
b. Especially if this other magnetic field oscillates at the same resonance frequency =
DEFLECTION