Gridder

The International Society for Magnetic Resonance in Medicine is next week in Toronto. In our genuine (sarcastic?) excitement of scientific discovery we submitted an entire abstract on a feature depicted on p. 246 of maybe the best known MRI textbook - dark, regular bands, all artifacts, and not at all useful - until recently.

The balanced gradient echo sequence is exquisitely and annoyingly sensitive to magnetic field inhomogeneities. This unfortunate feature, together with a forgotten prephaser gradient, perversely kept us in the dark about a series of banding artifacts along the direction of slice excitation.

The bands are predictable and appear whenever the accumulated spin phase per TR is 180 degrees. For example, the balanced steady-state free precession (bSSFP) frequency response looks like this:

The large signal voids occur at +- N*180 degrees and are spectacularly troublesome when images are acquired at high field or in the presence of large magnetic field inhomogeneities.

Only a stubborn MR physicist would try to take advantage of the bands. Two applications that come to mind are using ferromagnetic materials to label cells using the balanced GRE as a probe. The other is fMRI. We're perverse too, so why not attempt our own application? We added an 'unbalancer' gradient to the pulse sequence.


The unbalancer creates uniform grids on the tissue. The grids can be modified predictably by changing the gradient direction or amplitude:


But how are the bands useful? Could they persist through an acquisition to track tissues like SPAMM or DANTE? If so, the persistence isn't obvious and likely won't last longer than 30-100 TRs prior to entering the next steady-state. But both conventional tagging techniques persist a similar duration (< T1). Oh - the dream of indefinitely persisting tags.