DWIConvert converts UINT16 as INT16

[neurolabusc]

This bug impacts DICOM conversion of 16-bit UNSIGNED data with voxels where raw intensity exceeds 32767 (e.g. MRI magnitude images using full range of 16-bit datatype)

Consider a DICOM file with the following tags:

(0028,0100) US 16	BitsAllocated
(0028,0101) US 16	BitsStored
(0028,0102) US 15	HighBit
(0028,0103) US 0	PixelRepresentation [0000H = unsigned integer](http://dicomlookup.com/lookup.asp?sw=Tnumber&q=(0028,0103))

NRRD files created report type: short whereas the type field should report type: ushort. The consequence of this is that very bright voxels (>32767) appear to have negative intensity.

Attached is an image from dcm_qa where some voxels exceed 32767. I have also artificially converted a few voxels to have intensity 65535 to emphasize the problem:

uint16DICOM.zip

Discussed with dcm2niix issue 338.