Example of shim adjustment steps

(Button action is indicated by {})

Human Brain

  1. Determine frequency of water.
  2. Acquire localizer MRI (e.g. axial and sag FLASH).
    Define an area of interest, making sure enough sensitivity is present  and signal is contiguous in this area and that the dimensions are at least 5mm for rat brain and e.g. 30 mm for human brain. The position and size for shimming need not coincide 100% with spectrum localization and typically the shimmed area is a bit larger.
    Transfer the location of the center-of-mass to a exp that has a FASTMAP parfile (see above) loaded and the parameters adjusted input X0,Y0,Z0 for the center of mass of the ROI and tof for water frequency.
    Then use {Setup/Setup ROI} to define the dimensions of  a non-cubic ROI, or use {Setup/diam} for a spherical ROI.
  3. On 4 Tesla human scanners for human brain, you can use {FASTMAP/APS} which will start a series of 4 iterations in a predefined manner. For 4T this usually converges. First, a non-epi linear adjustment with tau=0.02 is executed. After EPI adjustment of linears in APS, the second orders are adjusted twice, once with short TR (0.3s) (the output can look like this.) and once with longer TR (1 s), which then can look like this.
  4. Occasionally you may want to rerun the last adjustment step, especially if you suspect that the 2nd orders are not close to well adjusted and are not pegged to the max DAC current, by {FASTMAP/Auto all}.
  5. In cases where you measure far away from isocenter, or the last second order changes were large, you might want to trim the linears, by using {FASTMAP/Autolin-6 proj}, especially if the last change had large changes on the 2nd orders.
  6. In cases where eddy current effects are not very well compensated, you may at the end want to adjust the linears in non-epi mode. To do so, go to {FASTMAP/Setup/nf_def=N} and enter 0 ("zero"). Then set tau to something between 0.01 and 0.02 for in vivo brain and run {FASTMAP/Autolin-6 proj}.

Convergence of shimming: This is usually achieved in human brain, when the linears (kx, ky, kz) are very low, e.g., on the order of 0.1-0.3 Hz/cm and the second orders are below 0.5 Hz/cm^2.

Animal Brain

For animal experiments, pretty much the same rules apply as for the human brain, with a few exceptions, that mostly are a result of the much smaller ROI:

Convergence of shimming: Because the ROI to be shimmed is almost an order of magnitude smaller, e.g., in rodents, the residual shim terms can be larger. For example, second-order shims (kxz, kyz, kxy, kx2y2, kz2) can be a few Hz/cm^2 and the linears can be 1 Hz/cm.

A second point to consider is that due to the smaller organ size, the initial inhomogeneities are much stronger and hence APS may not converge as easily. For Magical-savvy users it is easy to adjust the macro called by APS to their individual needs. Generally it is recommended to first get a good shim of the linears, before proceeding to second-orders. This may require non-epi mode initially and very small tau (even 0.001 s). Because even the second-orders can have huge correction terms, it is advisable to run second order shim several times until the second-orders fall reasonably close to zero (see above). Then it is recommended to adjust the linears only {FASTMAP/Autolin-6proj.}. 

Manual adjustment

  1. First MANUAL adjustment of linear shims:
    From main menu {FASTMAP}
    Use optional tpwr adjustment (RF power) is unsure: {Adjustments} and enter range for tpwr.
    Use setup to define tau: {Setup} {tau=} {Setup} {nt=}
    Define exp. for 3 main axes: {3 proj.} {return}
    Now run the expt: {manual adjust} {go}
    Determining the shim corrections: {Spherical ROI} {linear fit} and proposed values are displayed, and conversely for a rectangular ROI use {General ROI} {linear fit}. Check "error shell" and screen for good convergence, as indicated by RMSD being  typical for the sample used or below 1 Hz. If fit looks noisy and RMSD is high (>1Hz), check if signal is present (.e.g vf=100 dfsh), if present check voxel position, which is possible using voxflash in {FASTMAP}{Setup}. If fit looks ok, calculate new shims: {Calculate ...} which generates appropriate shim files and shim history (up to 3 oldn files) and displays calculated values. Although shim values in current expt are updated, DAC's are not, and must be updated by using {Update DAC's}.
  2. Readjust water frequency if necessary (in different expt).
  3. If linear changes determined in 4.) were large, run automatic linear adjustment: {FASTMAP} {Auto linear}, which runs step 4 automatically.
  4. If linewidths appear to be large, adjusting Z2c and x2y2 before all 2nd order shims are adjusted is recommended: {FASTMAP} {Auto 3 proj}. Always be sure to check water frequency and to check linear shims when
        incurring huge changes in 2nd order shim coils.
  5. Else proceed, to full adjustment using {FASTMAP} {Auto all}. be sure to check water frequency and to check linear shims when incurring huge changes in 2nd order shim coils.

back to main