It seems that the “default” data ordering for reading in MRSI data has been reversed in jMRUI 5.2 compared with 5.1. Previously it was necessary to use a Custom file to invert the MRSI matrix top-bottom to have the correct voxel ordering. This seems no longer to be the case. As a consequence when following the processes that we have used for years the data matrix is inverted top-to-bottom, at least for reading in 2d Siemens Numaris 4 rda files, compared with the previous version. Left-right is still OK. It is now wrong to use a Custom file, and the default ordering is correct. I suspect the same is true for Philips spar/sdat files. I have no information currently about dicom or formats from other vendors.
For our data this change was not immediately obvious, and it took a while to discover why data were not quite as expected (!)