I suspect this is an issue with your DICOM images. I suspect the Type 1 (Required) tags Patient Position 0020,0032 is missing or corrupted (see http://dicom.nema.org/medical/dicom/current/output/chtml/part03/sect_C.7.6.2.html for details). 

You can see this by running dcm2niix in verbose mode ("-v 1") and looking at the slice positions reported, e.g.

dcm2niix -v 1 ~/DICOM
Chris Rorden's dcm2niiX version v1.0.20200331 (JP2:OpenJPEG) (JP-LS:CharLS) Clang11.0.0 (64-bit MacOS)
Found 2 DICOM file(s)
Patient Position 0020,0032 (#,@,X,Y,Z) 1 2398 -63 -660.32 598.576
DICOM file: /DICOM/MR.1.3.12.2.1107.5.2.32.35131.2014031012593442716690029
patient position (0020,0032) -63 -660.32 598.576
orient (0020,0037) 0 1 0 0 0 -1
acq 1 img 1 ser 21 dim 384x384x1x1 mm 3.25x3.25x3.6 offset 88840 loc 0 valid 1 ph 0 mag 1 nDTI 0 3d 0 bits 16 littleEndian 1 echo 1 coilCRC 798937909 TE 30 TR 3000
Patient Position 0020,0032 (#,@,X,Y,Z) 1 2396 -63 -660.32 598.576
DICOM file: /DICOM/MR.1.3.12.2.1107.5.2.32.35131.2014031012593723427590139
patient position (0020,0032) -63 -660.32 598.576
orient (0020,0037) 0 1 0 0 0 -1
acq 2 img 2 ser 21 dim 384x384x1x1 mm 3.25x3.25x3.6 offset 88838 loc 0 valid 1 ph 0 mag 1 nDTI 0 3d 0 bits 16 littleEndian 1 echo 1 coilCRC 798937909 TE 30 TR 3000
Slices stacked despite varying acquisition numbers (if this is not desired recompile with 'mySegmentByAcq')
Converting /DICOM/MR.1.3.12.2.1107.5.2.32.35131.2014031012593442716690029
Slice timing range appears reasonable (range 0..2440, TR=3000 ms)
Convert 2 DICOM as /Users/chris/Neuro/issue400/issue400_sag_int_36sl_20140310133834_21 (64x64x36x2)
Conversion required 0.023337 seconds (0.015922 for core code).

Alternatively, you could view the full DICOM header using a tool like dcmdump (e.g. "dcmdump /DICOM/0001.dcm") and inspect these tags. I think you want to look at how your images were created and modified. In particular, some crude tools for DICOM anonymization remove these crucial details. If you can source a copy before these tags were corrupted/removed you should not have a problem converting the images.