The title bar of MRIcroGL reports 
   X×Y×Z (i×j×k) = v
Where XYZ is the crosshair location in world space, the ijk is the crosshair position in image space, and v is the voxel intensity at that location.

The world space location is estimated using the orientation information of the NIfTI header (with precedence for the matrix-based sform). When an image is converted from DICOM to NIfTI, the world space origin is the isocenter (for MR scans) or table center (for CT scans). After normalization the world space origin should be the anterior commissure.

The software does not make any assumptions about the precision of the data or the spatial transform. MRIcroGL assumes the spatial transform is truthful. It is simple a matrix transformation of the image space to world space. NIfTI assumes Talairach-Tournoux coordinates (RAS, such that XYZ are L->R, P->A, I->S). For more details:

  https://brainder.org/2012/09/23/the-nift...
  https://en.wikipedia.org/wiki/Transforma...
  https://bids-specification.readthedocs.i...
  https://www.nitrc.org/plugins/mwiki/inde...