IR works by using a 180 degree inversion radiofrequency pulse to flip all the spins by 180 degrees. A second 90 degree excitation pulse is used prior to data aquisition. At the time of the second excitation pulse (TI, or the inversion time), some tissues will have zero longitudinal magnetization and will produce no signal. The remaining tissues will be in varying stages of recovery. It can be very difficult to mentally visualize which tissues will have more or less signal without a graph. The time between the 180 degree inversion pulses is the TR of the sequence. The time between the 180 degree excitation pulse and the 90 degree excitation pulse is called the inversion time (TI). Note that TR must be larger than TI or the sequence does not make any sense.
The equation that describes T1 longidudinal recovery after an inverison pulse, and thus the signal produced by tissues with specific T1 values is
It should be noted that this is an oversimplification. The tissue filter should actually be because the 180 degree spin echoes repeatedly invert the nulled, recovering, longitudinal magnetisation. Because the time between echoes is much less than the tissue T1, it is estimated that the net longitudinal magnetisation is about zero at each spin echo. The actual recovery time is the time between the last spin echo and the next inversion pulse, which is . This is going to be ignored in subsequent discussions and we will assume the T1 tissue filter is simply .
This means that the IR times in this discussion may be off by 100 ms or so from the ones used in acutal scan protocols, but the shapes of the tissue filters are accurate enough for explaining concepts.
The T1 FSEIR filter is plotted in blue. Most inversion recovery sequences use a long TR > 5000 so near full recovery of longitudinal magnetization is accomplished for most tissues prior to each inversion pulse (i.e. the FSE T1 signal filter has reached S = 1 for most values of T1).
You can overlay the traditional FSE T1 filter by clicking the "FSE T1 filter" box. This plots the T1 filter previously discussed (red curve). Re-convince yourself that for high TRs the red curve is flat and near S = 1 for most soft tissues. There is little contrast for most normal soft tissues since most signal has recovered. This is why most inversion recovery sequences use a high TR - to get full longitudinal recovery of most tissues. IR sequences rely on the inversion time for generating T1 contrast, not the TR.
Notice that for most values of TI there is a lot of contrast (different values on the y-axis) between most soft tissues. Inversion recovery can provide increased T1 weighting provided you choose the TI wisely.
The steepest part of the inversion recovery filter is to the left of the point where the relative signal is zero (i.e. where the line contacts the x-axis, or the x-intercept). The steepest part of the curve can be determined by setting the the second derivative of the tissue filter to zero and solving for TI. When you use the ln(T1) axes, the steepest part of the filter is at T1 = TI. It can easily be shown that the maximum slope of the IR tissue filter is twice as steep as the maximum slope for the FSE T1 filter discussed previously.
Choose two tissues. If you choose a TI such that the x-intercept of the filter is larger than the T1s of both tissues then there is strong negative T1 contrast; the graph has a large negative slope between the tissues. This is what we expect from a traditional "T1 weighted" sequence. If the x-intercept is less than the T1s of both tissues then there is positive T1 contrast between the tissues. If the TI is chosen so that the x-intercept is between the T1s of both tissues then there could be negative, positive, or no contrast. This is why inversion recovery sequences can be confusing. Is there an inversion time so that grey and white matter have the same relative signal?
There are typically three types of IR sequences. Short IR sequences like STIR, with a TI of 240ms, results in positive T1 contrast for most tissues. You can demonstrate this on the graph. For most tissues, an increase in T1 results in increased signal. This is opposite what we expect for T1 weighted images. Note that STIR is generally considered a T2 weighted image. This point will become important in the next section.
Long IR sequences like FLAIR, with TI of 2500, result in weakly negative T1 for most tissues. For most tissues, a decrease in T1 results in decreased signal - which is what you would expect from a T1 weighted image. You should demonstrate this on the graph above. Note though that FLAIR is generally considered a T2 weighted image. This point will be discussed in the next section.
Intermediate IR sequences, like mp-RAGE, use an effective TI of approximately 1100. (Sequence TI is usually 800-900 but as the acquisition time is long and there is a sequential, gradient echo filling of k-space, the actual time between the 180 degree inversion pulse and the center k-space filling is closer to 1100ms). Enter a TI of 1100ms. This results in negative T1 contrast for tissues like grey and white matter. Because the slope of the IR filter is twice as steep as the slope of the FSE T1 filter these images have excellent T1 weighting. (remember that the sensitivity to changes in T1, or the relative T1 weighting, is the slope of the filter. For this reason, IR techniques such as mp-RAGE can be used to create nice anatomic images of the brain as the contrast between grey and white matter is very high. It is also sensitive to changes in white and grey matter T1 because the filter is steep at the T1s of white and grey matter.
Besides generating T1 contrast, inversion recovery is used to remove signal from tissues with a specific T1. For TRs much larger than tissue T1, the inversion time that results in zero relative signal for a specific tissue is 0.69 times the T1 of that tissue. The STIR sequence nulls signal from short T1 tissues such as fat. STIR stands for Short T1 Inversion Recovery. The FLAIR sequence nulls signal from CSF. FLAIR stands for FLuid Attenuating Inversion Recovery. But you can also use inversion recovery to null signal from white or grey matter too. We will talk about why this is important later.
Dont forget that the inversion recovery T1 filter is just one of three filters we've discussed. The standard STIR sequence uses a TI of 240. Based on this filter alone tendons have much more relative signal than muscle, which is not what you see on a standard STIR image. Remember there are also proton density and T2 filters that affect the final image. We will put the inversion recovery T1 filter together with the proton density and T2 filters when we discuss STIR and FLAIR in the next sections.