search

PTB-MR / mrpro

MR image reconstruction and processing.
https://ptb-mr.github.io/mrpro/
Apache License 2.0
17 stars 2 forks source link

Wrong matrix sizes for Pulseq data. #144

Open schuenke opened 9 months ago

schuenke commented 9 months ago

Currently, the header information for the recon_matrix and encoding_matrix extracted from the ismrmrd header of siemens raw data files acquired with Pulseq sequences are not correct.

For my example cases, the following information were off:

kdata.header.recon_matrix.y this value seems to be twice as high as it should be. Like an oversampling factor.

kdata.header.encoding_matrix.z this value defaults to 2 instead of 1 for 2D acquisitions.

Interestingly, the y-value is wrong for recon, but correct for encoding and the z-value is correct for recon, but wrong for encoding.

Not sure where we want to catch this. Maybe we can discuss it in the next meeting, because there might be several options along the entire pipeline like trying to fix it in the seq-files, in the converter, in the data loader, in the trajectory calculator etc ...

fzimmermann89 commented 8 months ago

Do we have any progress here?

Is this caused by the converter? Or the sequences? Or pulseq? Or the scanner?

How does recon_matrix get set anyways if using pulseq? At least the encoding_matrix issue is something (I think) should be fixed upstream somewhere. Question is where.. (The recon_matrix might be something that connot be fixed without writing a valid ismrmrd header for each sequence.)

fzimmermann89 commented 2 weeks ago

Aaand another question if we have any news here why the matrix is wrong....

schuenke commented 2 weeks ago

As we are using the MRD header files in most cases (where we set the encoding and recon matrix), this was not really an isssue for us any more. But still should be fixed imo. I will have a look at it again. Probably next week.

But it's most probably a wrong default thats either set by the pulseq interpreter sequence (NOT the seq-files) or by Siemens in general.