search

spine-generic / protocols

Generic acquisition protocol for quantitative MRI of the spinal cord
MIT License
14 stars 4 forks source link

Spinal Cord MT acquisition, with or without WE #4

Open ebannier2018 opened 3 years ago

ebannier2018 commented 3 years ago

Hi there, We have a new study with Spinal Cord MRI starting soon on Siemens scanners and wanted to get nearer to the consensus protocol for the MT acquisition. In our previous acquisitions we were using Water Excitation (normal) and the consensus protocol does not use WE. Doing somes tests on phantoms, we have a more pronounced signal loss on extreme slices without rather than with WE. Same results are observed on subject data, we lose the signal of about 2 vertebraes on each end. I am attaching a screenshot of the phantom acquisition (sagittal reconstruction shown).

Screenshot_Rennes_MT1_NoWE_WE

Plus, drawing a line through the profile gives the following plots : MT0MT1_WE_noWE The MTR and MTSat ratio are only sligthly modified though. Has any of you some insight there? Do you use WE or no WE? Any feedback and opinion here would be very much appreciated so we can decide to go for WE or not WE. If we go for WE, we would be sorry to not use the consensus protocol though.
Cheers, Elise

jcohenadad commented 3 years ago

@ebannier2018 thank you for pitching the discussion. I am puzzled as to why the slice profile is dramatically different with/without WE? What would cause these discrepancies? In any case, given that the excitation profile of 3D GRE scans is never perfect, we recommend to discard the 2-3 extreme slices, so this issue about slice profile doesn't seem like a dealbreaker to me.

In my preliminary investigations, I noticed that when using WE, the MTR was reduced by about 50%. It made me worry that this reduction of MTR would be associated with a lower sensitivity to detect myelin loss (which is primarily the role of MTR as a biomarker). Here are some (old) investigations, comparing various imaging parameters: all_results.xls.zip.

While still investigating the use of WE across sites and vendors, we noticed another issue: WE is very sensitive to poor shimming. With a slight offset in the static B0 field, there would be a drastic drop in the signal. Given that the spine region is prone to poor shimming and that shimming performance is highly dependent on the experimental setup, subject and expertise of the MR tech (location of the FOV wrt. isocenter, positioning, tilting of the head, amount of fat, size of shimming box, algorithm for shimming, quality of GRE scans used for shimming, etc.), we decided to drop WE for the MT protocol. You can see in the CHANGES.md the date when we dropped it (2017-12-01) and the reason stated.