Phantom construction was based on a high-resolution (1mm isotropic voxels)
low-noise data set that was created by registering 27 scans (T1-weighted
gradient echo acquistions with TR/TE/FA=18ms/10ms/30deg) of the same
individual in stereotaxic space where they were subsampled and intensity
averaged [15]. The volume
contains
voxels and covers the brain completely,
extending from the top of the scalp to the base of foramen magnum (see
Fig. 1). As a direct result of the high signal-to-noise
ratio (SNR), this single subject average dataset, dubbed CJH27, exhibits
fine anatomical details such as the claustrum, subthalamic nuclei, gray
matter bridges between the caudate and putamen, and vessels passing through
the lentiform nucleus; details normally obscured by noise in single images.
The 3D, high-resolution, anatomically accurate human brain phantom was
created by manually correcting an automatically classified and preprocessed
version of the MRI volume. The following sections describe the process in
more detail.