EPSI sequence analysis

1

Hi everyone,

I’m fairly new to MRSI and have run into a wall with a Siemens 3-D ¹H EPSI dataset. All I have are the original .IMA (DICOM) files—no accompanying TWIX .dat, RDA, or separate water-reference scan. My goal is to get the data into FSL-MRS, but first I need to reconstruct the spectroscopic volume. I tried converting the series with “spec2nii” but then “ValueError: cannot reshape array of size 3800000 into shape (32,38,100,1000)”. From digging through the headers, this is what I saw:
(0008,0016) UI [1.3.12.2.1107.5.9.1] # 20, 1 SOPClassUID
(0008,0008) CS [ORIGINAL\PRIMARY] # 16, 2 ImageType
I don’t have MIDAS installed, and downloading MIDAS no longer seems to be an option. I don’t know what the next steps would be…

Has anyone successfully reconstructed Siemens EPSI data in this situation—or patched spec2nii to handle it? Any guidance, code snippets, or alternative tools would be hugely appreciated.

Thanks in advance,
Ilyas

2

For spec2nii, @wclarke should be able to guide.

@bsoher is currently probably the best person to talk to get some sort of continued support for MIDAS.

3

@Ilyas I think I’ve come across this before. See FSL-MRS Spec2Nii reshape issue

Are you sure that this is the MIDAS package EPSI? Or is there another EPSI floating around somewhere? Can you find out the source of the sequence?

4

Hi,
I believe these DICOMs are from the Siemens WIP-814 “3-D EPSI” package—the one developed in collaboration with the Miami/MIDAS group. At the moment, I only have the raw .IMA files and need to perform the full reconstruction and pre-processing. I’ve read the FSL-MRS Spec2Nii reshape issue post, and it seems unlikely that MIDAS will be revived anytime soon…

5

Thanks for the information. Does that WIP come with any documentation and if so, how does it say to handle the data? @bsoher, do you know anything about this WIP?

6

Yes, there is documentation about the sequence, and I’m uploading here.
10303439_814_APP Guide-ASD-W01-00.pdf (364.7 KB)
The WIP can output “Processed” (spatially reconstructed & coil-combined), “Raw” (FID data), or both. Unfortunately, I was only able to output the raw data. Because Siemens’s standard spectroscopy tool doesn’t support these DICOMs, the guide recommends using MIDAS for all reconstruction and analysis.