Three volumetric T1-weighted gradient-echo scans of the OBT
stereotactic frame mounted on a plastic phantom filled with CuS0
solution were acquired transversely, sagittally and coronally
respectively. The imaging parameters were identical for the three
scans (voxel size = 
mm, echo time 10.3 ms, repetition
time = 17.8 ms). The frequency encoding direction was x for the
transverse acquisition and z for the other two. The left, right,
superior and anterior plates were mounted on the frame.
A Matlab script was written to extract the centroids of the channel sections semi-automatically on each slice of the three data sets. The section of one particular channel is seen when the volume is viewed in a plane perpendicular to that channel. Centroids were found by computing the average of the pixel positions, weighted by the pixel values, in a square neighbourhood of the maximum pixel of the marker section (the mean background values were subtracted from the pixel values and any resulting negative pixels were set to zero).