During testing of D906, I noticed that the MinPosition and MaxPosition as displayed in the Statistics View are not always completely correct when done for an image + mask. After further testing, this seems to depend on the underlying image.
When testing with Pic3D, everything works fine. When using brain.nrrd, with a rotated geometry, the coronal component of the position is always one too small. This was tested with segmentations, as well as planar figures from the Measurement View.