Thursday, November 8, 2012

Maximum fluid intensity -- bSSFP MRA

An MRI sequence that's not always recognized as an angiogram is the balanced steady-state free precession (bSSFP sequence... a.k.a. "trueFISP" or "b-TFE" or "FIESTA").

Example of an axial b-TFE image

But what is a balanced steady-state free precession sequence?

Well... first of all, what is a steady-state free precession sequence?

1) Steady-state sequence and "unspoiled" GRE

The steady-state sequence is inherently a gradient echo technique... but it's considered "T2/T1 weighted" (which results in it having the highest signal intensity of all sequences)... so what exactly does this mean?

First... how is it T2-weighted?

Unspoiled gradient echo relies on a short TR (time to repetition). If the TR is shorter than the T2 relaxation time, residual coherent transverse magnetization builds with each RF pulse (T2-weighted imaging relies on transverse magnetization for its effects). Remember that for a time of flight sequence (see posts from 10/14 and 10/17), repetitive short TR radiofrequency pulses do not allow much recovery of longitudinal magnetization, suppressing the the T1 signal and allowing for flow-related enhancement.  The situation is similar here, with quick, repetitive RF pulses, building up coherent transverse magnetization

With a short enough TR, there will still be some coherent transverse magnetization at the time of the next pulse (arrows are not very widely dispersed and there is a net transverse magnetization in the xy direction). There is also limited T1 longitudinal relaxation with a short TR (arrows not much recovered in the z axis). For a long enough TR, however, the spins have more time to dephase (transverse magnetization decays completely), and there is no net transverse magnetization vector (arrows going all whichaway).  There is also more time for longitudinal relaxation along the z axis.

The residual transverse magnetization is then added back to the longitudinal magnetization.

The amount of residual transverse magnetization is based on the T2 time (substances with long T2 times will have more residual magnetization to add back to the longitudinal magnetization, which can be used in subsequent excitations.  This is called an unspoiled GRE sequence when there is residual transverse magnetization to be added back to the longitudinal magnetization.

Because both effects are used to generate signal, steady-state sequences are T2/T1 weighted.

One important caveat... since the signal generated from a steady-state image is a summation of the T1 and T2 effects, it is possible to encounter a situation in which you get the same amount of signal from a shorter T1/shorter T2 and longer T1/longer T2 lesion... therefore, reliance on the signal intensity for diagnosis in a steady-state sequence is misleading.  The sequence is much more appropriate for delineation of anatomy or, in our case, angiography, rather than discrimination of pathology based on relative signal intensity.  The highest signal would occur from a lesion with a T2 almost as long as the T1 (a tissue's T2 cannot be longer than its T1)... and this causes fluid and fat to be the highest intensity tissues.

It seems clear that the shorter the TR, the more the T1 effect is diminished and the T2 effect is augmented...but the short TR GRE sequence seems similar to how a TOF sequence is formed, so why do the steady-state and TOF sequences look so different?

The answer is that the TOF sequence is a "spoiled" GRE. The residual transverse magnetization is "spoiled" by the application of a "spoiling gradient" or "radiofrequency spoiling" that dephases the transverse magnetization, and effectively leaves only the longitudinal (T1) component.

2) Steady-state free precession sequence

As seen above, not only does an RF pulse affect the longitudinal magnetization, it also affects the transverse magnetization that has not decayed over the TR.  The unspoiled gradient echo technique rotating transverse magnetization into the longitudinal axis is sometime called coherent steady-state free precession.

Each RF pulse (regardless of flip angle) excites longitudinal magnetization, as expected. But it also changes the rotational phase of transverse magnetization (as above).  So each RF pulse in GRE creates transverse magnetization and refocuses the transverse magnetization from the preceding RF pulse.  Refocusing the transverse magnetization causes a spin echo.

The example above is the most simple example of a spin echo, in which the flip angle is 90 degrees and the takes place at TE/2, but any RF pulse -- regardless of flip angle -- changes the rotational phase of transverse magnetization to some extent, and causes some refocusing of the transverse magnetization, although it may not be as complete as the 180 degree refocusing pulse.

So each RF pulse in a GRE sequence creates transverse magnetization and refocuses existing transverse magnetization. 

The free-induction decay of the transverse magnetization is sampled as a gradient echo.  The refocused transverse magnetization is sampled as a spin echo. If the TR becomes short enough, the two signals begin to merge into a "steady-state" signal.

3) Balanced steady-state free precession sequence

As the TR is brought lower and lower, the FID and spin echo signals can merge into a constant magentization throughout the pulse sequence. This requires centering the peak of the gradient echo exactly between each RF pulse so that the TE is exactly one half the TR.

This T2/T1 sequence maximizes signal amplitude, and amplifies tissues with long T2 decay.  Its signal to noise ratio is very high.

Stationary as well as moving fluids demonstrate extremely high signal, causing them to have relatively higher contrast, and functioning as a type of angiogram.

Two images from a FIESTA sequence (a balance steady-state free precession sequence).  The fluid (blood and CSF) and fat is extremely bright due to the T1/T2 effect.

The balanced steady-state free precession sequence is also extremely fast, with image acquisition times on the order of less than a second, allowing its use in dynamic cardiac imaging.

1. "MRI Principles" Mitchell DM and Cohen MS. 2nd ed. (2004)
2. "The Physics of Clinical MR Taught Through Images" Runge VM, Nitz WR, Shmeets AH, et al. (2005)