MRI Scans
Introduction
Dianne Patterson, Ph.D. dkp@arizona.edu
August 16, 2021
This is a brief description of the MRI sequences (images) you are likely to analyze if you do research. With the exception of the defaced T1w image, the images here are of the author and publicly available (there are no HIPAA issues).
Each image is linked to a dynamic viewer for that image. The viewers were created with Papaya.
An excellent source of technical information about MRI sequences is the MR-TIP website.
Structural (Anatomical) Images
Structural images are 3D volumes that capture information about the tissue in the brain (white matter, grey matter and CSF). The T1w image is the most common choice in research and is used for tissue segmentation, and image registration.
T1 weighted image
Click T1w Image to explore the T1w image using the Papaya Viewer.
You will usually acquire a T1-weighted structural (a.k.a anatomical) image. It takes about 5 minutes and has voxels ~1mm in each dimension (1 mm isotropic). This is considered high resolution because it is much better than typical fMRI or DWI.
Defaced T1 weighted image
Click Defaced T1w Image to explore the defaced T1w image using the Papaya Viewer. In the Papaya viewer, you can see that this participant has a chronic stroke (large irregular CSF-filled region on the left).
HIPAA requires that you de-identify MRI data. Deidentification requires defacing (removing identifying features), after which images can be made public. The nose, mouth and chin have been removed (lower left) to protect the identity of the participant. Defacing is not the same as brain extraction (BET).
T2 weighted image
Click T2w Image to explore the T2w image using the Papaya Viewer.
Researchers sometimes acquire a T2w image is like a negative of the T1w image. This image has 1x1x3 voxels and takes just a few minutes to acquire.
FLAIR image
Click FLAIR Image to explore the FLAIR image using the Papaya Viewer.
A FLAIR, fluid attenuated inversion recovery image, is a T2w image with dark CSF (i.e., the fluid is attenuated). It is useful for identifying white matter hyperintensities. This is especially useful for researchers whose subjects are older or are known to have neurological abnormalities (MS, stroke etc)
Standard Space image
Click MNI Image to explore the labeled standard space image.
To compare and analyze brain images from different individuals, the images must be "warped" to a standard "space" (template). The most common standard spaces were developed by the Montreal Neurological Institute (MNI). Because this image is in a standard space, you can move the mouse to display which brain region the pointer is in (displayed at the bottom of the viewer).
4D Image Sequences
Some images contain more than one 3D volume. These are said to be 4D. For DWI, each volume uses a different gradient to detect information about white matter tracts that travel in that plane. For fMRI, each volume represents a short period of time (e.g., 2-4 sec) and captures changes in blood oxygen. By comparing volumes, it is possible to see changes in oxygenation that are related to cognition.
DWI: 4D image sequence
Click DWI Image to explore the DWI image using the Papaya Viewer. DWI is a 4D sequence consisting of separate volumes.
Diffusion images (a.k.a. DTI [Diffusion Tensor] or DWI [Diffusion Weighted] Images) are used to examine the integrity of white matter. This one has 30 volumes, each with different gradient directions applied. It also includes two B0 (b-zero) images, for a total of 32 volumes. B0 images have no gradients applied, They appear as the first and last image in the sequence (and are much brighter than the images with the gradients). This is probably the minimum you would want for research. If you double the number of directions, you'll also double the acquisition time! This sequence has 2 mm isotropic voxels and takes about 5 minutes to acquire. It was acquired with A-P (anterior to posterior) phase encoding, which is also pretty typical. Repetition time is typically quite long for diffusion-weighted images (10 seconds for the typical B-1000 illustrated here).
fMRI: 4D image sequence
Click fMRI Image to explore the fMRI image using the Papaya Viewer. The fMRI is a 4D sequence consisting of separate volumes.
This resting state fMRI (functional magnetic resonance image) includes 177 volumes, repeated every 2 seconds (i.e., repetition time, TR, is 2 seconds). It took about 6 minutes to acquire. Voxels are 2.5 x 2.5 x 3.5 mm, so they are anisotropic (not the same in every dimension). It is often suggested that resting state scans be longer, but it depends on whether your participants can hold still and avoid going to sleep for a longer scan. As is shown here, often fMRI scans sacrifice some spatial coverage for speed (the top of the head and the cerebellum are cut off)
Field Maps
Field maps take about 1.5 minutes to acquire and are extremely useful for correcting distortions in DTI or even fMRI sequences. A typical field map sequence on our Siemens scanner includes two magnitude images and one phase image (below). These voxels are 3 x 3 x 4 mm.
Note: fmriprep will only find and use the field maps IF there is an IntendedFor key and value in the accompanying JSON files for each field map. You can read more about the issue here.
Magnitude Field map
Click Magnitude Image to explore the magnitude image using the Papaya viewer.
Phase Diff Field map
Click Phase Image to explore the phase image using the Papaya viewer.