MRI Brain Segmentation:
Typical Data

Basic description

Magnetic resonance (MR) provides a 3-dimensional description of internal structures by non-invasively measuring: Proton density (PD), spin-lattice relaxation time (T1), and spin-spin relaxation time (T2). A detailed description of how these parameters are measured (the physics of MRI) is described elsewhere. Different types of tissues in the brain show up differently in these images:

example PD image example T1 image example T2 image

Proton Density T1 T2

These images can be acquired coronally, sagittally, or axially, but all of these are fundamentally 3D data.

spacer Acquisition Planes spacer

example coronal image example axial image example sagittal image

Coronal Axial Sagittal

Two Voxels For this reason, each pixel in an image is more properly referred to as a voxel. In a particular slice image, each voxel has a brightness value that corresponds to a measurement of the tissue weighted by the MR parameters (PD, T1, or T2) averaged over a small 3D region.

In addition to the type of tissue at a given location, the exact intensities in the resulting image are also determined by the neighboring tissues, the radio frequency pulse sequence used to acquire the scan, how well the MR scanner is calibrated, motion of the tissue or fluid, and a number of other possible artifacts. When medical experts examine a brain scan, they use the relative values of these inexact intensities along with their knowledge of brain anatomy to combine a series of slices into 3D shapes which correspond to structures inside the patient's head.

When taking a MRI brain scan, there is a tradeoff between many factors: type of image to acquire (PD/T1/T2), cost, time, resolution, slice thickness, distance between slices, signal-to-noise, etc. Time and considerations such as claustrophobia are especially important considerations regarding the patient.

Go [back up]
Please email Christian if you have additions, questions, comments or suggestions.