Scan Optimization for UCLA BMC Trio

NOTE: This page is a work in progress. If you have additional data that are relevant to these tests, please contact Russ Poldrack (poldrack@ucla.edu).

We have collected two sets of test data on the new BMC Trio using a range of acquisition parameters. Here are some of our results and suggestions for protocol setup on the Trio. Once we have finalized our parameters, we will place them into a protocol called Poldracklab_Trio_Rev1 under the Poldrack group in the user protocols.

Note that all of the following results were obtained with the 12-channel head coil; we have not yet tested the 32-channel coil.

Structural imaging

We have not extensively tested the structural protocols.

Functional Imaging

Here are activation maps from a set of different parameters, using a 24 second on/off visual/motor task.

No IPAT, TE = 30 ms, TR=2340 ms

IPAT (Grappa factor=2), TE = 30 ms, TR=2000 ms

IPAT (Grappa factor = 2), TE=24 ms, TR=2000 ms

IPAT (Grappa factor = 4), TE = 24 ms, TR=2000 ms

These data suggest that IPAT with a TE = 30 ms is probably the best bet. However,as you will see below, if signal from dropout areas is critical, then it might be a good idea to lower it to 24 ms, though this will reduce BOLD signal from the rest of the brain.

One thing that is not appreciable here is the substantial increase in shot-to-shot noise when IPAT level 4 is used in comparison to IPAT level 2, but this is almost certainly the cause of the lower statistical effects in the IPAT 4 scan.

We also compared these different scans in order to get a better feel for the differences in distortion and dropouts across them. Here are four representative slices

There are two things to see here. First, as expected, shorter TE's provide better signal in dropout areas. Second, and perhaps more important, there is an appreciable decrease in susceptibility-based distortion when using IPAT in comparison to standard imaging. To see this, compare the two images at the right, which both have 30 ms TEs. It's clear that the structures are in different places in the no-IPAT image.

Based on these results, we are adopting the following parameters with the EP2D_BOLD sequence:

IPAT Level 2
TE = 30 ms
Automatic motion correction off (we have not tested with it on, so for now we keep it off)
For best signal from ventral prefrontal regions, use an oblique axial slice prescription that is about 20-30 degrees back from the AC-PC line.
4 mm slice thickness, zero skip

Please note that your mileage may vary, and it would be best if your group tested these parameters on paradigms that you understand well.

Diffusion Weighted Imaging

The first thing to note is that the DTI sequences on the Trio are loud and shake the bed quite violently. It is probably a good idea to warn your subjects beforehand.

We tested the 64-direction DTI sequence (ep2d_diff_orth 64 DIR) on two subjects, with IPAT Level 2 turned on for one subject and IPAT turned off for the other. Using IPAT reduces the scan time from about 9 minutes to less than 8. However, after processing the data using FSL's dtifit tool, we noticed appreciably more noise in the images for the IPAT series compared to the no-IPAT series.

No IPAT	IPAT2

Based on these results, we will use the diffusion sequences with IPAT turned off.

There are some outstanding issues with the EP2D_DIFF sequence:

Currently it only collects a single image at b=0. We are looking into how to increase this.
We do not have the official gradient table. We have used the table that is extracted from the DICOM header by dcm2nii, but we have been warned that these may not be correct.

We will post additional information here as we receive it.

Document Actions

Print this

Sections

Personal tools

Scan Optimization for UCLA BMC Trio

Structural imaging

Functional Imaging

Diffusion Weighted Imaging

Document Actions