Collect a batch of dMRI data

Hi Everyone. Recently, I need to collect a batch of dMRI data to fit DTI, DKT, HARDI, NODDI models, there are two options: 1000 (30) 2000 (30) 3000 (30), 1000 (30) 2000 (60), which one is more suitable? Or have a better suggestion?

Hi @wly,

You might find this page helpful (DSI Studio’s suggestion for DWI acquision).

In general, given a finite acquisition time, you should acquire more directions in higher weighted shells, as higher b-vals are more noisy, and you’d rather be more confident in your noisy signal than over sample clean data at lower shells. DTI (which personally I do not recommend), DKI, and NODDI can be fit on spherical shells or cartesian-grid sampling schemes.

The page linked above makes a case for using a cartesian-grid, but you can still choose a spherical shell scheme if your later post-processing calls for it. For example, if you want to use a MRTrix3-based spherical harmonics pipeline, you CANNOT use cartesian-grid schemes. But if you want to use diffusion spectrum imaging (DSI) models, you NEED a cartesian-grid scheme.

The page linked above says that if you would prefer to use spherical shells:

  1. acquire one 32-direction DTI at b=1500 and another 60-direction DTI at b=3000 (built-in ep2d_mddw protocol)
  2. In-plane resolution: 2mm, slice thickness: 2mm (isotropic resolution is essential) If scanning time is an issue, use 2.5 mm isotropic.
  3. Matrix size: 104x104
  4. Slice number: 72 (can be reduced if ignoring the cerebellum), no gap
  5. Minimum TE and TR, but the TE for the b=1500 DTI should be the same as the TE of the b=3000 DTI.
  6. Make a copy of the sequence and invert its phase encoding direction (acquire AP and PA for phase distortion correction)


Hi, Steven. Thanks for your guidance. We also want to use a cartesian-grid, grid-258 b-table is too time consuming, but the grid-101 b-table only gives the direction, not the b-value, do you know how to get it?

If you are using a Siemens scanner, I think you can just follow the same directions as listed on the website, but enter your 101-direction b-table instead:

You will need to acquire both “dMRI_dir258_1_Siemens” (full 258-direction DWI) and also its opposite phase encoding b0 acquisition “dMRI_dir258_2_Siemens” (only the b0), which is for distortion correction.

You may need licenses such as Siemens SMS EPI license (for MB imaging), Siemens DTI package license (for diffusion table), High-performance gradient (HCP) that is installed in Prisma (for high bandwidth readout).

In the diffusion tab of the sequence, switch “MDDW” to “Free” mode and load the grid-258 b-table. The table should be placed under C:\MedCom\MriCustomer\seq. You may need to rename the current DiffusionVectors.txt file to another name first and then copy the grid-258 table to DiffusionVectors.txt. Then set b-value1=0 and b-value2=4000

Unfortunately, we are using GE machines.

The grid-101 file with b-val is here:

And then you can follow these instructions:

Steps to install the 12-min q-space scheme on Other scanners

Convert the grid-258 b-table to its compatible format.

The following are acquisition parameters:

  1. In-plane resolution: 2.0 mm, slice thickness: 2.0 mm (if your SNR is not good enough, increase them to 2.4 mm.)
  2. Matrix size: 104x104
  3. Slice number: 72 (can be reduced if ignoring the cerebellum), no gap
  4. Multiband acceleration factor: 3 or 4
  5. “Bipolar” diffusion scheme (for eliminating eddy current). Some may prefer “Monopolar” and correct Eddy current using FSL’s eddy. This only works on shell acquisition and is not recommended for the grid scheme.
  6. Minimum TE and TR
  7. Pixel bandwidth: ~1700
  8. Phase encoding direction: A to P
  9. Make a copy of the sequence and invert its phase encoding direction (P to A). Only b0 is needed here for phase distortion correction.

Great! Thanks for your guidance. Does DSI Studio now support both of cartesian - grid and spherical shell scheme processing ?

It should, but I recommend QSIPrep for pre- and post-processing your data (many of its processes use DSI Studio).

Thanks for your guidance.I will try it. Is the “Bipolar” diffusion scheme just AP-PA acquisition and then use topup+eddy_openmp to process it?

No, AP-PA refers to the phase encoding direction, which is different from the polarity mentioned here. You can read more about uni vs bipolar acquisitions here:

Could you help me how to change bipolar for GE machines? Or do I have to install additional sequences? I asked the GE engineer and she didn’t know how to change it :joy:, We guess if turning on dual spin echo will do it?our machine is Premier.

I work with a GE Machine. To invert polarization direction, you need to have access to the GE research mode.
You have then several option, but all basically boil down to flip the “pepolar” user CV value.
This can be done from the user interface or directly by editing the PSD.

For future reference,

Type-in PSD(s) allows to flip phase encoding direction (pepolar) in clinical mode (No need for research mode.

Here is the full list of type-in PSDs depending on the use of HyperBand and/or interpolation (native resolution):