newseg

This page describes our new high-resolution MRI technique designed to help us segment brain images into different tissue types. Specifically, this allows us to determine the distribution of each individual’s gray matter, white matter, cerebrospinal fluid (CSF), bone, non-brain soft tissue and air based purely on MRI scans. This is a tricky endeavor, as air, CSF and bone are all dark in traditional MRI scans, making these tissues difficult to distinguish. An alternative is to use CT scans, but these expose people to X-rays. There are many possible uses for this technique. For example, this information can help source localizing scalp-recorded recordings of the brain’s electrical signal. ALternatively, this information can provide a more accurate idea of how each individual will be influenced in a closed head injury.

We acquire two MRI scans from each indivdual. The left column below shows the scans from a single person. We collect both a T1-weighted MPRAGE scan and a T2-weighted SPACE scan. The benefit is that these two modalities show different tissue contrasts – for example bone, CSF and air are all dark on the T1 scan, whereas on the T2 scan the CSF is bright while the air and bone are dark. The two scans are collected with identical slice orientation and with the same field of view (each scan has a resolution of 320×320x208 voxles with a 0.85mm isotropic resolution).

We collected data from 25 young adults and used the ANTS software to generate a template T1-weighted image (middle column in image). This allows us to account for our scanner’s image brightness, but also ensures that we have information about the neck (most MRI templates only show the skull). This image is than normalized to scanner space and we then develop scanner-specific Bayesian tissue maps.

This allows us to use SPM8’s “new segment” function, that includes multiple channels (T1 and T2 weighted images) and all six tissue types. The image on the right column shows the results from a single individual (showing gray matter, white matter, bone and CSF tissue types).

Click here to watch a video demonstrating the results.

Here is another video (also from a single participant).