Incorrect DVH for very small ROIs

We are using dicompyler-core for calculating DVH from DICOM files. We are seeing a weird behaviour for very small ROIs (<0.5 cm^3) where the last dX values before d100 are zero even tough d100 isn't:

d96 = 7621.00 d97 = 0.00 d98 = 0.00 d99 = 0.00 d100 = 7620.00

Guessing it has to do with how the algorithm handles a low voxel count.

Thank you for submitting this issue. I will try to reproduce, but if possible could you send an anonymized dataset?