Functional Imaging Scanner Protocols?

I’m curious about which scanner sequences different labs are using for their BOLD fMRI experiments on the Prisma. What TR, TE, voxel size, multiband, or GRAPPA accelerations are you using for your experiments?

We’ve just collected a round of multi-echo data with the following parameters, and it has looked good (task, multivariate analyses, and resting-state)

TR: 2000 ms, TEs: {14.6 31.86 49.12}, Voxel size: 2.7 mm isotropic, MB: 2, GRAPPA: 2

2 Likes

In my last experiment on Prisma at NIH, we were doing TR = 1s, voxel size = 3x3x3, multiecho 3 (echo times 13.6, 31.86, 50.12), multiband 4. Could have gotten away with MB3 and fewer slices but I refuse to cut off the cerebellum :slight_smile: Trying to track down the full protocol PDF so can share exactly what various acceleration factors were once I get that. Was a little worried about MB4, but did some (admittedly brief and unsystematic) piloting of an MB3 sequence with a larger slice gap and fewer slices, and the tSNR was pretty comparable in most regions, so went with the MB4.

2 Likes

Cool! I haven’t played with multiecho for awhile. I recently saw an interesting paper on a new multiecho sequence, curious what @Kodiweera thinks about it. We’ve previously have not had great luck combining Multiband with Grappa, but maybe it’s not an issue with the multi echo? @adamsteel @emily.s.finn do you mind sharing a picture of temporal SNR to see the coverage through susceptibility artifact areas? My previous experiences with multiecho on GE scanners we had much better signal coverage in the OFC.

We are currently using CMMR multiband-multiecho sequence, which is the same one Peter Bandettini’s group uses for their TEDANA analysis ( https://tedana.readthedocs.io/en/latest/) . Researches acuqire 3 to 5 echoes ( contrasts), typical is 3. It does not affect MB and does not add much time to the total scan time. As a matter of fact, we can think the first echo is free! Data quality is appreciable. We have built several protocols at DBIC so that researches can customize for their flavour. As for the TEDANA, some researches mentioned they don’t see much improvement in results with it. However, I would say, acquiring three echoes does not hurt, as they can always use the middle echo for the traditional analysis if they choose not to do TEDANA.
We don’t have a WIP implemented for EPTI. They say EPTI is good for fast aquisition, but it is highly undersampled. I would say it needs extensive comparable results if we want to acquire data for a project using it. We should find out if other groups are using this sequence and how the results look compared to traditional/standard EPI. If someone wants to play with it, I can look for an EPTI WIP.

1 Like

@ljchang here’s a quick screen shot of the OFC in our data (single echo (middle TE), Optimal combination of the three echoes, and optimal combination + ME-ICA denoising). Data is 12 minutes of resting state with the parameters listed above. Happy to share the raw data if it’d be helpful! Note that there is automated masking of low signal voxels in the OC and OC+ME-ICA data where dropout is most severe.

Screen Shot 2020-09-23 at 5.15.30 PM|690x389

1 Like