Structural Imaging Techniques: CT and MRI

Structural Imaging

• Structural imaging methods use the different physical properties of tissues like skull, gray matter, and white matter to construct detailed static maps of the brain's physical structure.
• Computerized Tomography (CT) scans are constructed based on the varying absorption of X-rays in different tissue types, which is related to tissue density.
  • The scanned person receives a small amount of radiation.
  • CT scans are used clinically to diagnose tumors and hemorrhages.
• Magnetic Resonance Imaging (MRI) creates structural images of soft body tissues.
  • It relies on the water content of tissues, as different tissues behave distinctively when stimulated.
  • MRI applies a strong magnetic field to the scanned body part, causing single protons in water molecules to align with the field.
  • A radio frequency pulse knocks the aligned protons by 90 degrees, leading to a detectable change in the magnetic field, which forms the MR signal.
  • The external magnetic field is constant during scanning.
  • MRI scanners use field strengths between 1.5-3 Teslas (T).
  • The Earth's magnetic field is about 0.0001T.
  • The scanner emits radio waves to excite different slices of the brain.
  • A whole brain scan takes about 2 seconds with 3mm slices.

MRI advantages over CT

• MRI does not use radiation, making it safe for multiple scans.
• It has better spatial resolution, allowing observation of individual gyri.
• MRI distinguishes between gray and white matter, unlike CT.
• It can detect BOLD changes associated with neural activity.

Voxel-based Morphometry (VBM) & Diffusion Tensor Imaging (DTI)

• VBM and DTI are used to link brain structure to function by imaging white and gray matter.
• VBM analyzes small-scale differences in the organization and concentration of white and gray matter using structural MRI to detect differences.
  • VBM divides the brain into thousands of small regions (voxels) and estimates the concentration of gray/white matter in them.
• DTI measures white matter connectivity between regions by measuring the amount of water without any consideration of how it is connected.
  • It quantifies the effect of water molecules trapped in axons diffusing in certain directions using fractional anisotropy.

Functional Imaging

• PET and fMRI detect changes in brain metabolism or blood flow during tasks, identifying activated brain regions.
• Unlike EEG & MEG, functional imaging doesn't directly measure neural events but correlates metabolic changes with neural activity.
• Neurons need oxygen and glucose, which are distributed via the circulatory system, with the CNS using ~20% of oxygen intake.
• The brain regulates blood flow to different regions based on need, increasing flow to active areas.

Positron Emission Tomography (PET)

• PET scanning measures brain metabolic activity using a radioactive substance, usually radioactive oxygen-labeled water, injected into the bloodstream.
  • As the tracer converts back to its stable form, it emits positrons, which collide with electrons, creating gamma rays detected by the PET scanner.
• Reconstructed PET images display blood flow distribution.
• PET measures "relative activity, not absolute metabolic activity."
• Usually, a PET experiment involves injecting the tracer during a control condition and one or more experimental conditions.
• PET results are reported as changes in regional cerebral blood flow (rCBF) between conditions.
• PET scanners can resolve metabolic activity in regions (voxels) of 5 - 10mm³.
• PET scans can measure any radioactive agent, leading to the search for biomarkers of neurological disorders like Alzheimer's disease, using agents like Pittsburgh Compound B (PIB).

Functional Magnetic Resonance Imaging (fMRI)

• It exploits the principle that local blood flow increases in active brain areas.
• fMRI focuses on the magnetic properties of deoxyhemoglobin.
• Deoxyhemoglobin is paramagnetic, creating distortions in the local magnetic field.
  • These distortions indicate deoxyhemoglobin concentration, known as the BOLD effect/contrast.
• The BOLD signal evolves over time due to neural activity, known as the hemodynamic response function (HRF), involving three phases:
  • Initial dip: a small rise in deoxyhemoglobin results in a reduced BOLD signal.
  • Overcompensation: Blood flow increases, causing a significant BOLD signal increase.
  • Undershoot: Blood flow and oxygen consumption dip before returning to baseline.
• The HRF is stable within the same brain region for the same participant, but varies across different regions and individuals.
• Changes in blood flow are slower than neural events.
• fMRI's temporal resolution is several seconds due to the sluggishness of the hemodynamic response.

fMRI vs PET

• fMRI scanners are less expensive and easier to maintain.
• fMRI uses no radioactive tracers, reducing costs and risks.
• fMRI allows repeated testing of the same subject.
• fMRI has better spatial resolution.

Study Designs

• Block design: activity is integrated over a 'block' of time while a subject performs a task.
  • PET needs enough time to detect radiation, requiring continuous engagement in a task for >40s.
• Event-related design: the BOLD response is linked to specific events.
• fMRI can use both block designs and event-related designs.
• Experimental and control trials can be randomized.

Limitations of PET and fMRI

• PET & fMRI have poor temporal resolution compared to single-cell recordings or EEG.
• PET is constrained by the decay rate of radioactive agents.
• The temporal resolution of PET is around 30s.

PET and fMRI Safety

• Small risks are associated.
• Ethical issues in functional imaging research include excluding pregnant women and children from PET studies due to radiation.
• People with metal implants are excluded from fMRI experiments.

Analyzing Data from Functional Imaging

• Brain images are superimposed with colored blobs, with the blobs representing statistically significant regions.
• Brain images are mapped onto a standard reference brain (stereotactic normalization).
• Each brain is divided into thousands of small volumes (voxels) with 3D spatial coordinates.
• The standard space is based on an atlas from Talairach and Tournoux.
  • Assigns 3D coordinates to each point in the brain.
• Smooting is where brain activity is spread from the raw activation of a given voxel to neighboring voxels.
  • Facilitates detection of common activity regions across individuals.
• Data are corrected for head movement.
• Statistical analysis tests for differences in mean activity.
• Statistical thresholds are based on spatial smoothness using family-wise error (FWE).
• Statistical thresholds are based on spatial smoothness using False Discovery Rate (FDR).

Interpreting Data from Functional Imaging

• Reasons for an increased BOLD signal include:
  • The region being essential for the task.
  • The region being involved in the specific strategy.
  • The region being involved in a general mechanism.

Study Design

• Echo planar images collect all information needed to produce a brain slice image after a single radio frequency excitation.
  • Offer high temporal sampling rates but with compromised spatial resolution.
• Time to repetition (TR) corresponds to the time between two excitation pulses.
• Regulates experiment's temporal resolution.
• Shorter TRs lead to reduced brain coverage.

Key fMRI Study Design Considerations

• Spatial resolution is represented by the selected voxel unit.
• Increased voxel size is increased tissue detected as active using conventional analysis. but sensitivity to BOLD is prioritized over spatial resolution
• Voxel size impacts the SNR of the images negatively, reducing the sensitivity to the BOLD effect.
• Cortical thickness is a good guide when deciding the right voxel sizes.

High Temporal and Spatial Designs

• Jittering: the use of different delays between the start of sampling of brain images relative to the start of stimulus presentation to the subject.
• Parallel imaging (a.k.a. parallel acquisition schemes) : a technique based on the spatial coding of signals from coil sensitivity profiles. Reduces the acquisition time by a factor set by the experimenter