Hi Cecilia,
For there to be an integer wrap-around problem, the holes would tend to be in bright areas. While the hole on the right side (about the size of the crosshair hole) seems like it could be from that, the others do not so much. But that is just a little eyeball test.
At any rate, you should be able to add -ushort2float to the initial Dimon command (it would then apply it in the subsequent to3d command used to create the datasets). It would be good to see what difference that makes.
- rick