Loading data from GE and Siemens

Dear MRS experts,

I tried to load GE p-files or Siemens’ .dat files to JMRUI v5.2. However, the loaded number of data points doesn’t match the actual number. I have tried all the formats under Load As but it still doesn’t work.



Additional information: I saw that the GE and Siemens data contain FIDs for each channel, i.e., not coil combined. Does JMRUI accept data of this type, or should I combine the coils before loading the data in JMRUI? Thanks.

Hi @ruifengd,

Part I:

My first comment is that jMRUI 5.2 is now quite old – version 7 was released last year. I wouldn’t be too surprised if older versions struggle to import data from newer scanner software releases. So if possible, I’d recommend you upgrade to the latest version.

Having said that – here are a couple of pointers based on my (limited) experience with v6.0, which is also now a couple of years old.

In principle, jMRUI supports GE and Siemens formats. However, by default it tries to auto-detect the format – as far as I can tell, by attempting each configured format one at a time until something works. In my experience, it often fails before ever reaching the “correct” format. There are a couple of ways around this – either you can use the “Open As” function and then choose the correct format:

Skjermbilde 2024-02-01 kl. 11.55.31
Skjermbilde 2024-02-01 kl. 11.57.45

…or re-arrange the order and selection of formats it will try:

Skjermbilde 2024-02-01 kl. 11.50.30

Probably best to move the ones you are commonly using to the top of the list, and maybe un-check formats you’re never expecting to use.

Part II:

Yes, and it’ll offer to combine it for you immediately after loading. Here’s how it looks for our GE data:

“Combine and add” will do coil combination and adding of individual transients – for GE data it’ll handle water ref and metabolite transients separately, for most Siemens data these are likely to be in separate scans

Or “Combine” to do coil combination only, keeping all the individual transients separate

Or “Skip” to not combine anything – then you can choose various other processing steps to combine in whatever way you choose.

How far off is the number of data points? At least for the Siemens TWIX data, these may be oversampled and contain some additional samples at the beginning (and end) of the FID… this could explain a factor of two-ish difference from the value specified on the exam card.

Thanks for your reply.

For both the GE p-file and Siemens’ .dat file, 4096 points were sampled. However, JMRUI always loads them as having 1024 points.

Strange. I’d strongly recommend getting hold of a more recent jMRUI version before spending too much time on the problem…

I also ran into some random issues with earlier versions on a 64-bit (or mixed) operating system, and found it to be overall much better behaved when installed on a 32-bit virtual machine.

If you still have difficulties with the latest version, you could try using something like FID-A to convert the data into jMRUI format (io_writejmrui), but I’m not certain the how well the FID-A readers supports the most recent GE formats. Perhaps it helps, perhaps it just adds a different set of challenges.


I can read the files in other software like FSL_MRS. But in order to convert it to text files for JMRUI to read, I don’t know how that format should be like. I know for coil- and repetition-combined data, we just need to put real and imaginary parts to two separate columns. But for uncombined data, what should be the order of FIDs from each channel and repetition?



I’m afraid I don’t know the order, but my guess would be just concatenate the FIDS, with receivers in the inner loop then repetitions. More frequent jMRUI users may be able to correct me :slight_smile:

Is this for a replication/longitudinal study? Otherwise I’m not really sure what you’re trying to achieve; if you can read (and hence, process) the data with FSL-MRS, why do you need to get uncombined data for individual receivers into an outdated version of jMRUI? Maybe do some of the processing with fsl_mrs_preproc and just feed the output to jMRUI?

The latest release of jMRUI promises more flexible coil combination, support for the new GE format, support for the NIfTI-MRS format (same as FSL-MRS, as generated using spec2nii), and some really interesting 2D fitting tools. It’s almost certainly the best solution for a new study.