We found a mismatch between the number of points we set and the number of points read when using Ospery to read 7T Siemens data. We set the acquisition points to 2048, but the final read result was 4192. The pointsToLeftshift value was 64, and Ospery detected the pointsToLeftshift and read 4128 points. However, it still doesn’t match the set value. Is this mismatch caused by zero-filling and oversampling? Do you currently have any scripts for processing raw 7T Siemens data and generating basis sets?
Hi @Zhangren,
Thanks for reaching out.
From what I can gather from your post you are looking at the TWIX .dat files from Siemens. The TWIX files are oversampled by a factor of 2, therefore, if you set 2048 datapoints in the exam card you will end up with 4192 points in the TWIX file. If you, for example, look at the RDA files (not recommended for the analysis), you will see that those have 2048 points because they are not oversampled.
One critical point during the loading is to identify the correct field in the TWIX header for the points before echo or pointsToLeftShift. Do you know how many points were acquired before the echo? The 64 seems like a lot. If you are able to share the data, I could take a look and see if I can correctly identify the field.
Also, what do the spectra look like?
Cheers,
Helge
Thank you for your prompt response. Yes, the data format is Siemens TWIX, and the data acquisition information is shown in the screenshot above. On my machine, I have set the acquisition points to 2048 with a bandwidth of 2000Hz. The attached data is from my sequence, Semi-Laser (eja_svs_slaser), version MR_VE12U. I didn’t save it in .RDA format, but I have saved it in .IMA format, where the number of points matches, i.e., 2048 points. I understand that this is the result after machine processing, but I’m not clear about the entire process. Besides oversampling (2x) and pointsToLeftshift, if it involves zero-filling, it seems that the number of zeros filled each time is not consistent. Do you have an understanding of the entire process? Additionally, how can Ospery be used to generate the corresponding basis set for 7T?
Furthermore, it seems that I cannot upload large files here.
Correct the IMA file also has the same number of points as set in the exam card. I don’t know the detailed steps that are performed, but for RDA/IMA, the points before the echo are removed, and no oversampling is applied. Therefore, the number matches the exam card settings.
For TWIX, the points before echo are not removed and the data is oversampled (2x). I don’t think it involves any zero-filling and the number of points should be in your case consistently 4192.
Osprey does not generate basis sets itself. You have to generate them yourself using, for example, MRSCloud or FID-A. Or maybe the programmer of the sequence has supplied you with a .BASIS file, which you can use (if you run LCModel) or transform to an Osprey compatible basis set.
You have to use an external storage place to share your data, e.g., OneDrive, GoogleDrive, etc.
The attached link contains the specific data. Microsoft OneDrive