Hi Chris,

many thanks for your excellent conversion software!

Recently, I bumped into this error where it seems that the slices are combined in the wrong dimension. See the attached image. A dicom viewer (papaya/mango in this case) has no problems with these images. What option am I missing? Or have the images been wrongly exported from the scanner?

Best,
Maarten

The issue here is not that the images are being combined in the wrong dimension, rather the Siemens 2D mosaic is not being re-ordered into a 3D spatial volume. The mosaic format saves all the images from a single volume as a single 2D slice. If my software does not handle these, it is usually a sign that the proprietary CSA header has been damaged. If you send me a sample directly I can provide more feedback.

Hi Chris,

many thanks for wanting to take a look. I've sent you a DICOM directly.

Maarten

Thanks for the sample. The upcoming release of dcm2niix does convert your 2D mosaic to a 3D volume.
  https://github.com/rordenlab/dcm2niix/releases
However, there is a serious issue with your image. Specifically, the CSAImageHeaderInfo (0029,1010) has been stripped. It should contain rich data about your sequence
http://nipy.org/nibabel/dicom/siemens_cs...
but has been removed.
While the latest releases of my software is able to detect that this is a mosaic using NumberOfImagesInMosaic (0019,100A), many of the parameters of this image have been removed, so no software will be able to work out phase encoding direction, effective echo spacing, etc. I suggest go directly to your scanner to get the raw data. You should also look carefully at the providence of your data and identify the tool that has stripped out this vital information. The last tool to save the DICOM data (and therefore the prime suspect) is:
MAROTECH REGISTER 5.0

Hi Chris,

many thanks for resolving this. We receive the data via an intermediary who is apparently stripping the data way too aggressively. This was also what I expected, but I had no idea about the specific header tag that was needed to allow proper conversion.

Thanks again!
Maarten