Thanks for the link, That is a strange looking mandibulectomy! The bar looks like it is protruding from the skin!
That volume is not ideal for capturing the small details of the sinus bones so it is no wonder that there are lots of holes in the eye orbits. If you open the volume, then go to the Volumes module, then expand the volume information area you can see the image spacing values in there. The first two values are 0.556mm x 0.556mm which is the size of the voxels in the axial plane. This suggests that this was an axial CT scan with a field of view of about 285mm with a 512 matrix (0.556mm x 512 = 285mm).
The spacing in the z direction (Superior-Inferior) is 2mm which indicates that the slice spacing was 2mm.
If you go down to the display area in the volume module you can untick the interpolation box which allows you to visualise the voxels more easily in the 2D slice views.
Most of the scans I deal with for imaging orbital bones have a field of view of around 200mm so in plane voxel size of around 0.35 - 0.4mm and a slice thickness of 0.5mm - 1mm. This gives much better results that this volume.
You will need more that the fill holes tool to salvage a low resolution volume like this. You may see some improvement with Unsharp mask image filtering as described in the link above. You could also experiment with the shrink wrap segment effect that @lassoan mentioned.
Yes it is rather a complicated process. Read as much as you can in the linked topic above and follow some of the links to the masters thesis etc and post any remaining questions that you have on here.