Siemens .IMA - single average and multiple averages

MRS Software Osprey
1

Hi @Helge and @admin,

I have two questions about using Siemens .IMA files.

For one data set, I have a single .IMA file per subject. When loading the data, I get the following error. In debug mode, out.seq is correctly assigned STEAM, so I’m wondering if seq should be out.seq?

Unrecognized function or variable ‘seq’.

Error in io_loadspec_dicom (line 325)
if strcmp(seq,‘PRESS’) || strcmp(seq,‘STEAM’) || strcmp(seq,‘SLASER’)

Error in osp_LoadDICOM (line 67)
raw = io_loadspec_dicom(MRSCont.files{metab_ll,kk});

Error in OspreyLoad (line 185)
[MRSCont] = osp_LoadDICOM(MRSCont);

Error in osp_onLoad (line 33)
MRSCont = OspreyLoad(MRSCont);

Error while evaluating UIControl Callback.

For another data set, I have fMRS with individual averages saved as .IMA (90 .IMA files per subject). Can Osprey be used to average these data? It seems to be looking for a single .IMA per subject, but I’m not sure.

Thanks!

2

Hi Meredith,

Thanks for this bug report - yeah, seq in these final statements ought to be replaced with seqtype or out.seq @Helge.

About your second question - is your job file pointing towards a folder of IMAs instead of a single IMA? In that case, it should automatically load all of them, interpret them as single transients, and align/average them down the line. Can you try using the seqtype bugfix and report back?

Thanks,
Georg

3

Hi Georg,

I tried the seqtype fix for the single .IMA and ran into this error:

Invalid field name: ‘’.

Error in read_dcm_header (line 56)
dcmHeader.(C{1}) = value;

Error in io_loadspec_dicom (line 32)
DicomHeader = read_dcm_header(filesInFolder{1});

Error in osp_LoadDICOM (line 80)
raw_ref = io_loadspec_dicom(MRSCont.files_ref{ref_ll,kk});

Error in OspreyLoad (line 185)
[MRSCont] = osp_LoadDICOM(MRSCont);

Error in osp_onLoad (line 33)
MRSCont = OspreyLoad(MRSCont);

Error while evaluating UIControl Callback.

For the fMRS, I am pointing to a folder.

4

OK this seems odd, looking at the error message it seems like it’s not finding any pair-value lines in the file that is specified in files_ref.

  1. What are the files and files_ref entries in the job file, and
  2. What is your file/folder structure looking like (you will probably have to place the single IMA files for the files and files_ref in separate folders)?
5

These are my entries in the job file:

files = {‘/Users/mar0046/Desktop/REID_K01_SUB387/ima/SVS_ST_VAPOR_643_BASELINE_32AVG_0011’};
files_ref = {‘/Users/mar0046/Desktop/REID_K01_SUB387/ima/SVS_ST_VAPOR_643_BASELINE_RFOFF_0012’};

6

I tried a different subject, and it loaded okay for the single averaged .IMA :thinking:

7

Can you confirm that there are absolutely no other files other than DICOM in these folders?

8

Confirmed. Only the .IMA, and there does not appear to be any hidden files.

9

Can you e-mail me the file (de-identified) that causes this problem?

10

Sent!

For the multi-IMA fMRS data, I successfully loaded all the averages for one subject. When I click on “Process data” I get this error:

Unrecognized field name “averaged”.

Error in OspreyProcess (line 191)
if raw.averages > 1 && raw.flags.averaged == 0

Error in osp_onProc (line 35)
MRSCont = OspreyProcess(MRSCont);

Error while evaluating UIControl Callback.

Thanks!

11

Hi Meredith,

I’ve just committed a bugfix to the develop branch of our Osprey repository that fixed the processing for your single-IMA datasets. - give this one a try (it’ll make it into the next master release too). I have no idea why this error occurs, but essentially our DICOM parser stumbles over a corrupted line that is only present in one of the four files you shared (super-odd). Effectively, it now just skips parsing a line that it can’t make sense of.

The fix also contains a blind shot at dealing with the multi-IMA issue you reported in your last post, so I’m not 100% sure this will work because I didn’t have a dataset to test on, but please have a pop at it.

Cheers,
Georg

12

I’ll give it a try this week! Thanks!

13

Hi Georg,

Everything is running. The multi-IMA loaded and successfully went through the “process data” step. Thanks for the fixes!

Meredith

1 Like
14

Hi Georg,

A related update. I was assisting a student who is using Gannet for segmentation, and he ran into the same error with the DICOM header. I did a quick fix on read_dcm_header in Gannet using your Osprey fix, and then he ran into a second similar error (Invalid field name: ‘0’.) that was also fixed by adding the try/catch to “case 2” at line 63.

Meredith

2 Likes
15

thanks - tagging @mmikkel

16

Hi @Meredith,

Would you mind emailing me me those fixes you applied for Gannet? My email is mam4041[at]med[dot]cornell[dot]edu

17

Sent! I don’t know what is causing the differences in the headers, and it seems to be sporadic.

1 Like
18

Hi @admin and @Helge,

I got the Invalid field name: ’ ’ error again with another Siemens IMA spectrum. In Osprey’s read_dcm_header.m script, I added the same change you previously made (try-catch for case 1 at line 55) to case 2 at line 63, and it seems to have fixed the problem.

1 Like
19

Thanks so much for this fix! I’m constantly amazed by how differently a supposedly standardized file format can behave after all. :wink:

20

Yes! :smile: So far, we’ve only run into issues with case 1 and case 2 for two different labs/studies/voxels but seemingly random subjects. All are Siemens 7T STEAM, but I haven’t identified any other pattern as to why it happens. :woman_shrugging:

1 Like