MR Protocols

From CNI Wiki

Jump to: navigation, search

This page contains an overview of several types of MRI modalities (structural, functional, and diffusion). The basic measurement protocols are described and there are links to development plans and more detailed processing strategies.

There are several typical protocols users run. These protocols involve a combination of scans. We document the most widely used protocols in this list. Click on specific links to see protocol details.


Saving your protocol parameters

Save screen-shots

At the GE console, you can save screen shots of the GE interface to show the main parameters that you have set in a protocol. Just get to the screen that you want to save, then press the 'Prnt Scrn' button on the keyboard. A little dialog will show up. You can choose to print, which will print on paper to the Laser printer in the control room. However, we strongly suggest that you save some trees and the toxic ink chemicals by saving a digital copy instead. To do this, type ina reasonable name in the filename field (default is 'screen') and hit the 'PNG" button. A PNG image will then magically appear in the 'screensaves' folder on the linux machine next to the console (cnirt). From there, you can email the images to yourself. Or, even better, create your own personal wiki page here that describes your protocol (just log in with your SUNet ID) and put the images in there. Then, you will always have them available when needed! THis is also a great way to share protocol information with your colleagues.

Get a PDF of all protocol parameters

You can get a complete PDF of all your protocol info with a few clicks of the mouse. It's not quite as easy as a screensave, so we outline the procedure here.

Simultaneous Multi-Slice EPI

The CNI, in collaboration with GE, is implementing simultaneous multi-slice EPI (also known as multiband EPI, multiplexed EPI, or, as we like to call it, "mux EPI"). Our mux-epi efforts are described on the MUX EPI page.

Structural imaging

General Options


Using a higher pixel bandwidth can help reduce chemical shift effects that push the fat signal from the scalp into the brain. [MORE HERE]

3D Geometry Correction

The 3D Geometry Correction option uses a 3D correction for gradient non-linearity, over the 2D correction that is performed when the option is not checked. By including the slice direction in the correction, the resulting images are closer to geometric truth. The model used to represent gradient nonlinearity is the same as the 2D correction ("gradwarp") and it uses the same cubic interpolation function as the 2D correction.

Suggested sequences

The CNI has stored protocols for T1-weighted, T2-weighted, and diffusion scans. Depending on the user's needs, there are several ways to run a scan session. We include all necessary scans in one stored protocol called "CNI standard anatomical", which is stored under "CNI / Head". This is menat to be used as a 'menu' from which you select the sequence that you want, based on your needs. While there are many variations stored there, here we just highlight a couple of the most common versions. A detailed list of all parameters for all scans is included in this PDF file. Some suggested ways of selecting from and these scans for your own scan session are described below.

Anatomical sequences list.jpg

3planeloc SSFSE

(0:21) 'Single Shot Fast Spin Echo' This is a 3 plane localizer or 'scout' scan meant to find the subject's head. It is also be used for slice Rx for the subsequent scans. Doing some sort of localizer is essential, and this one is the standard work-horse used by most CNI users.

3planeloc details.png
3planeloc adv.png

ASSET calibration

(0:06) ASSET calibration measures and calibrates the signals from the multiple coils. It should be run BEFORE any scans that will use ASSET, such as fMRI or Diffusion scans.

Asset calibration details.png


(0:09) High Order Shim. This scan measures the magnetic field inhomogeneity and corrects it with polynomial gradients up to 2nd order. It should be run AFTER ASSET and BEFORE T1-weighted or T2-weighted scans.

Hos wb hrbrain details.png
Hos wb hrbrain adv.png

T1 weighted scans

All the suggested T1-weighted scans use GE's "BRAVO" sequence. BRAVO is just GE's way of saying "IR-prep, fast SPGR with parameters tuned to optimize brain tissue contrast". Unless you have good reason to do so, you probably don't want to play with any parameters other than slice orientation, voxel size, and bandwidth. And for those, most users just pick one of the suggested configurations:

  • T1w 1mm ax (3:22): T1-weighted, 1mm^3 voxel size, 3D Bravo, axial slices. A single scan gives good signal-to-noise quality. If you just want a basic, fast, axial T1 weigthed scan, go with this.
T1w 1mm ax details.png
T1w 1mm ax adv.png
  • T1w 1mm sag (3:43): T1-weighted, 1mm^3 voxel size, 3D Bravo, sagittal slices. A single scan gives good signal-to-noise quality. This is similar to the 1mm axial, but with sagittal slice orientation. Compared to axial, this orientation is slightly less efficient because you need a full phase FOV, but sagittal slices usually do better than axial with artifacts from large blood vessels (e.g., carotid artifacts land in non-brain regions rather than the temporal lobes) and with fat-shift artifacts, because the shifted scalp signal usually misses the brain while with axial it can sometimes overlap the occipital lobe gray matter, causing tissue segmentation problems.
T1w 1mm sag details.png
T1w 1mm sag adv.png
  • T1w 0.9mm sag (4:49) T1-weighted, 0.9mm^3 voxel size, 3D Bravo, sagittal slices. A single scan gives good signal-to-noise quality. As with the above scan, but a little higher spatial resolution. If you can afford to take 5 minutes for a T1 scan, this one is a great choice. This is our work-horse. Note: to get true .9 isotropic voxels, enter '23.04' for the FOV. The scanner GUI will display this as '23.0', but will store and use the full-precision that you type!
T1w 09mm sag details.png
T1w 09mm sag adv.png
  • T1w 0.8mm sag (4:57 X 2) T1-weighted, 0.8mm^3 voxels, 3D Bravo, sagittal slices. Two scans (averaged in post-processing) are advised for good signal-to-noise quality. If you want to get better resolution, do two of these.
T1w 08mm sag details.png
T1w 08mm sag adv.png
  • T1w 0.7mm sag (5:41 X 3) T1-weighted, 0.7mm^3 voxels, 3D Bravo, sagittal slices. 3-4 scans (averaged in post-processing) are advised for good signal-to-noise quality. If you can afford the time, and make use of high-quality anatomical images, this is the sequence to use.
T1w 07mm sag details.png
T1w 07mm sag adv.png

T2 weighted scans

  • 3D T2 (5:03): T2-weighted, 0.8mm^3 voxel size, 3D Cube T2, sagittal slices. A single scan gives good signal-to-noise quality.
T2w cube 08mm sag details.png
T2w cube 08mm sag adv.png
  • 2D T2w/PDw FSE (4:25): A standard 2D T2-weighted scan. You also get a bonus proton-density scan. Note that the two datasets will be interleaved; you'll want to separate them in post-processing.
2d t2w pdw fse details.png
2d t2w pdw fse adv.png

Quantitative T1

Ask Aviv!

Functional imaging

Setting the TE for BOLD sequences

The optimal echo time (TE) for BOLD fMRI at 3T is 30ms. [MORE HERE- show T2* signal decay curves for oxy/deoxy hemoglobin]

Setting the optimal flip angle

When doing BOLD fMRI, we generally like to acquire data very quickly. When the TR (the repetition time) is shorter than the longitudinal relaxation time (T1) of the tissue of interest, we want to adjust the flip angle to optimize the SNR by maximizing the magnetization recovery along the z-axis (T1) during successive excitations of the same tissue. The optimal flip-angle is found by the Ernst equation:

flip-angle = acos(exp(-TR/T1)) 

[Note: this formula will return values in radians, which then need to be converted to degrees. Alternatively, if using Matlab, use the acosd function which will return degrees.]

  • A typical T1 value for gray matter is (3T): 1.33 seconds (Kruger, et al, 2001). (At 1.5T, it is closer to 0.9 seconds.)
  • Or use the following values for typical TRs at 3T:
TR (s): 1 1.5 2 2.5 3 3.5 4 5 6 7
flip (deg): 61.9 71.1 77.2 81.2 84.0 85.9 87.2 88.7 89.4 89.7

EPI Gradient Echo 32-channel, Whole brain (3sec TR)

Run these scans first:

  • Localizer
  • 3D T1-weighted anatomical, .9mm isotropic

While the T1w is running (about 5 minutes), you can prescribe the following sequences (in this order):

  • ASSET calibration
  • BOLD EPI (TR=2, ~3mm isotropic)
  • Higher-order shim
  • Field map (copy the BOLD EPI Rx)

Then, run the ASSET calibration (20 seconds), the higher-order shim (~1 minute), and the field map (~1 minute). While these scans are running, get your stimulus ready. Once the field map finishes, you can run the BOLD EPI. Be sure to run Prep Scan first to ensure that triggering works as expected! To repeat the BOLD EPI, just copy and paste the series and run it again. You might consider doing another field map at the end to assess any drift. For long functional runs, you may even want to do additional field map measurements between BOLD EPI runs.

See the Improving EPI page for information on fixing some common image problems with EPI images.

Spiral Gradient Echo for 32-channel

Please add info about the spiral sequences available. Atsushi?

Diffusion weighted imaging

Diffusion imaging at the CNI uses a modified version of GE's DW-EPI sequence. The sequence was modified so that for dual-spin-echo scans, the polarity of the second 180 degree pulse is inverted relative to the first 180. This causes off-resonance signal from fat to get defocused and thus help reduce fat-shift artifacts (See Sarlls et. al. Robust fat suppression at 3T in high-resolution diffusion-weighted single-shot echo-planar imaging of human brain. MRM 2011, PubMed PMID: 21604298 and Reese et. al. Reduction of eddy-current-induced distortion in diffusion MRI using a twice-refocused spin echo. MRM 2003, PubMed PMID: 12509835).

Diffusion Tensor Imaging (DTI) =

We recommend doing a 10-minute scan that includes 60 directions with a b-value of 1000 and a voxel size of 2mm^3. Most users choose to acquire 60-70 axial slices.

Dti 2mm b1000 60dir details.png
Dti 2mm b1000 60dir adv.png
Dti 2mm b1000 60dir diffusion.png

High Angular Resolution Diffusion Imaging(HARDI)

For HARDI analysis, we suggest acquiring a 17-minute 96-direction DW scan with 2mm^3 voxels and a b-value of 2500, 60-70 axial slices. If you are pressed for time, you can drop the b-value to 2000 and/or reduce the number of directions to 80. Then, reduce the TR until the maximum number of slices is one slice higher than the number of slices that you want.

Dti 2mm b2500 96dir details.png
Dti 2mm b2500 96dir adv.png
Dti 2mm b2500 96dir diffusion.png

HARDI protocol optimization

To decide on an optimal HARDI acquisition protocol, see:

  • White and Dale (2009) Optimal diffusion MRI acquisition for fiber orientation density estimation: an analytic approach. HBM.
    • we calculate optimal b-values for maximum FOD estimation efficiency with SH expansion orders of L = 2, 4, 6, and 8 to be approximately b = 1,500, 3,000, 4,600, and 6,200 s/mm^2
    • We further demonstrate how scanner-specific hardware limitations generally lead to optimal b-values that are slightly lower than the ideal b-values.
  • Tournier et al. (2008) Resolving crossing fibres using constrained spherical deconvolution: validation using diffusion-weighted imaging phantom data. NeuroImage
    • for a 45 degrees crossing, the minimum b-value required to resolve the fibre orientations was ... 2000 s/mm^2 for CSD, and 1000 s/mm^2 for super-CSD
  • Tournier et al. (2007) Robust determination of the fibre orientation distribution in diffusion MRI: non-negativity constrained super-resolved spherical deconvolution. NeuroImage.

HARDI data analysis

  • Camino
  • dipy
  • mrTrix

Diffusion spectrum imaging (DSI)

More here soon.

Technical Details

We use a modified version of the stock GE DWI-EPI pulse sequence. The resulting dicoms contain the diffusion parameters in these fields:

  • b-value (in sec/mm^2): 0043 1039 (GEMS_PARMS_01 block, item 1039)
  • gradient direction: [0019 10bb, 0019 10bc, 0019 10bd] (GEMS_ACQU_01 block, items 10bb - 10bd)

In mrTrix (mapper.cpp), the following code is used to convert the dicom gradient values to the saved gradient directions:

// M is the image transform
M(0,0) = -image.orientation_x[0];
M(1,0) = -image.orientation_x[1];
M(2,0) =  image.orientation_x[2];
M(0,1) = -image.orientation_y[0];
M(1,1) = -image.orientation_y[1];
M(2,1) =  image.orientation_y[2];
M(0,2) = -image.orientation_z[0];
M(1,2) = -image.orientation_z[1];
M(2,2) =  image.orientation_z[2];
M(0,3) = -image.position_vector[0];
M(1,3) = -image.position_vector[1];
M(2,3) =  image.position_vector[2];
M(3,0) = 0.0; M(3,1) = 0.0; M(3,2) = 0.0; M(3,3) = 1.0;
H.DW_scheme(s, 0) = M(0,0)*d[0] + M(0,1)*d[1] - M(0,2)*d[2];
H.DW_scheme(s, 1) = M(1,0)*d[0] + M(1,1)*d[1] - M(1,2)*d[2];
H.DW_scheme(s, 2) = M(2,0)*d[0] + M(2,1)*d[1] - M(2,2)*d[2];

If you get the data from the CNI Neurobiological Image Management System (NIMS), then the b-values and b-vectors have already been extracted for you and are provided along with the NIFTI file containing your data. These three files (the NIFTI, bvals, and bvecs files) can be send directly into most diffusion data analysis packages, such as the Stanford Vita Lab [mrDiffusion] or FSL's [FDT]. The b-values file contains a set of numbers (one for each acquired volume) that describe the b-value of the corresponding volume. The b-vecs file contains a triplet of numbers for each acquired volume, describing the diffusion-weighting direction for the corresponding volume. E.g., if you run our 60-direction scan, you will get 6 non-DW volumes followed by 60-DW volumes. Thus, you nifti file will contain 66 volumes. The b-vals file will contain 66 numbers (six 0's, fllowed by 60 1000's) and the b-vecs files will contain 66 triplets describing the DW directions for each volume (the triplets for the first 6 non-DW volumes are meaningless and can be ignored).


We are in the process of setting up a spectroscopy protocol for GABA GABA spectro.

Device specific processing

The General Electric processing includes various steps that can influence the signal-to-noise of your data. We explain what we have learned about this and how to control it in the GE Processing page.

Basic Scanning on the GE Scanner - How to Guide

For those new to MRI scanning here is a "How to Guide" for basic scanning. The following set of slides shows how to start a scanning session with entering subject information, entering information correctly into the "Patient ID" field for automatic data transfer to NIMS, continuing on by running a protocol with both anatomical and fMRI sequences, and ending the scanning session.

Subject Data Entry

Subject Data Entry When you first arrive at the scanner, you will see the interface shown in this slide. This is the subject information data entry form. In order to enter information the cells must be white. If they are not, clicking the icon in the upper left hand corner (circled in red) will turn them into enterable fields.

Subject Data Entry

Subject Data Entry Key data fields that must be entered include: Subject Name, Subject Weight, Subject Date of Birth, and Operator Name The subject’s weigh must be included in the weight field for SAR calculations otherwise the scanner will not run. For data retrieval and archiving please make sure that your data is stored in the your correct lab folder. The format requirements for the "Patient ID" field are as follows (and are also posted by the GE console):[SUNet ID of faculty PI] [slash] [description] (all lower case, no spaces, alphanumeric only) Examples are: gross/study1, wandell/abc, kalanit/xyz754. Click on Start Exam (circled in red) to proceed to protocol list.

Protocol Retrieval

Protocol Retrieval Click on the protocol you wish to run (highlighted protocol will appear in yellow). To move it into the “protocol basket” click on the arrow to the right of the protocol list (circled in red). By double clicking on the protocol you can view all the sequences in the protocol., You can also move individual sequences to the ‘protocol basket” from other protocols.

Protocol Retrieval

Protocol Retrieval Once you have collected a protocol, click accept (circled in red) to move to the next display.

Choosing dB/dt and SAR Limits

Choosing dB/dt and SAR Limits You will need to choose either Normal Mode or First Level before proceeding further. To accept First Level click accept (circled in red) or you can choose Normal Mode by clicking that box.

The Localizer

The Localizer Every protocol will have a localizer as its first sequence. The purpose of the localizer is to verify the land marking of the subject and to provide coronal, sagittal, and axial images onto which prescription geometries can be determined for subsequent sequences in the protocol. The designation ”InRx” (circled in red) to the left of the sequence name indicates that either the parameters and/or geometry must be checked by the user for the sequence to be executable. In the case of the localizer there are no parameter adjustments needed. In order to run the sequence the “Save Rx” (circled in red) must be clicked. This will turn the “InRx” to “ACT”. The localizer can now be run by clicking on the “Scan” button (circled in red, above the “Save Rx” button), which now will be highlighted.

Asset Calibration

Asset Calibration All imaging sequences include parallel imaging techniques to speed up data acquisition. GE sequences use either ARC or ASSET. All sequences in your protocol that use ASSET need to use data from the ASSET calibration sequence. It a good practice to run this sequence after the localizer. Proceed as follows: Highlight the sequence with a single mouse click. By clicking Setup (circled in red) The designation “InRx” will appear to the left of the sequence name. Clicking in the left upper window (circled in red). will result in a grid appearing. Position the grid as shown in the screen shot to the right, making sure that the grid covers the image and also covers space outside the image as shown in the screen shot. As with the localizer first click “SaveRx” and click “Scan” to run the sequence.

3D Anatomical Sequences

3D Anatomical Sequences Depending on your method of data analysis, you will have at least one anatomical sequence in addition to the fMRI sequences as part of your protocol. These sequences can be either 2D or 3D. 3D Anatomical sequences are prescribed with slabs, shown in this slide rather than slice grids as seen with the localizer and Asset calibration, and are manipulated the same way. In the “Setup” mode parameters can be adjusted. The full parameter tab can be viewed by using the toggle button (circled in red) to display the full parameter page or the axial image views. Depending on the study it is sometimes preferable to run a 3D T1 weighted sequence first, which will then allow prescribing of the functional MRI sequences. After the geometry has been prescribed in this 3D anatomical sequence clicking “SaveRx” and “Scan” allow the sequence to run.

fMRI Sequences

fMRI Sequences fMRI sequences can be set up and run similarly to other sequences previously described by using the “Setup”, “SaveRx”, and “Scan”(circled in red) The screen shot on the right shows where a geometry has been prescribed by clicking first on the upper left image. You can then move the grid by using the mouse and can rotate it by placing the mouse on either of the two dots on the grid. Parameters can be adjusted also and they can be viewed by clicking on the pointer (circled in red) to hide the third image window.

fMRI Sequences

fMRI Sequences Parameters may be adjusted on each of the available tabs (circled in red): Details, Multi-Phase, Acceleration, Advanced.

fMRI Sequences

fMRI Sequences The parameter in the Multi-Phase tab “Phases per Location”( circled in red) can be used to change the length of time fMIR sequence is run to match the length of the paradigm.

fMRI Sequences

fMRI Sequences The parameter “ Start Scan Trigger” (circled in red) determines how the fMRI sequence is triggered relative to the paradigm. Prior to running the fMRI sequence it is important to adjust the high order shims. There is a dedicated sequence to do this and shown In the next slide. Is is also important to remember that the shimming program will use the last sequence prescribed for high order shimming.

High Order Shims (HOS)

High Order Shims (HOS) There are no parameters that need to be adjusted in the HOS sequence. The sequence is run by clicking on “SaveRx” and “Scan” and following the pop-up window choices. Clicking “Calculate Shims” and “Done” in the pop-up shim window allows the optimized shim values to be entered into the system.

After the completing the shim program. The fMRI sequence is run. Multiple copies of the sequence can be obtained by using the cutting and pasting options via the right mouse button when used over the sequence list.

Ending the Exam

Ending the Exam Exams are ended following the drop down window choices after clicking on “End” (circled in red)

Quality Assurance

Daily QA scans include:

  1. BOLD EPI sequence (analyze mean and variance over time)
  2. DW EPI sequence (analyze eddy current distortion stability)
  3. Spiral field map (analyze long-term B0 stability)

All QA scans are done on the fBIRN agar phantom. The phantom is positioned in the same orientation with the same padding (TO DO: build a [phantom holder) each day. The landmark must be set to the same. The Rx should be not touched (use the same stored Rx every day). We should do HO shim and set the shim VOI to exactly cover the sphere.

EPI Notes

To enable ARC, set the cv oparc to 1.

Spatial Resolution Notes

Slides from CNI tutorial

Session Running Script

We advise you to put together a session running script that outlines set up of the scanner and peripherals and positioning of and communications with the participant. You can find an example here (courtesy of Nanna Notthoff, Carstensen Lab).

User-specific protocols

Personal tools