STIR It Well


Edematous os peroneum. Notice the lack of conspicuity of the peroneus longus tendon	Peroneus longus tenosynovitis. Notice the wide range of gray differences between the peroneus longus tendon, its surrounding tendon sheath fluid and the adjacent edema in the marrow.

These two sagittal STIR sequences show peroneal tendinopathy in two different patients. However, the quality of the two images is radically different; the image to the left is unacceptable. How do they differ?

The image to the left has very low signal-to-noise ratio (SNR).
As a consequence, it shows fluid as bright signal, and the rest of the structures are wiped out. In fact, it is even impossible to note that the brightest area corresponds to an edematous ossicle (os peroneum), or to make out the tendon of the peroneus longus as distinct from the adjacent fat-suppressed adipose tissue.
On the contrary, observe the nice range of tones on the right: notice how the muscles are still clearly brighter than the marrow, but darker than the areas of edema. Further, the edema signal in the bone is different from the fluid in the tendon sheath.
The reason is that the TE on the left is 90 msec., whereas the TE on the right is around 50 msec.
Conclusion: in a sequence with a very low intrinsic SNR, such as a STIR, prolonging the TE unnecessarily renders the sequence useless.

So, What's the "right" TE?

There is no such thing as the "right TE." 90 msec. is plain wrong, and this should be apparent just looking at the images: we waited too long to collect the information from the tissue, and there is no magnetization left. However, below 50-60 msec., choosing a TE for a STIR sequence is a matter of debate.

The shorter the TE, the more SNR in the images, and the more pleasing to the eye. Yet, there is always a trade-off in MR imaging: STIR sequences with short TE's become mere photographic negatives of a regular T1 SE with the caveat that the SNR is always lower, because of the inversion pulse intrinsic to a STIR. Here's an example:

If we mentally invert the sagittal T1 weighted image on the left, we'll come up with an image similar to the image in the center. Now, this image is remarkably similar to the actual sagittal IR, shown on the right. Therefore, the sagittal IR would be a "photographic negative" of the sagittal T1.

The rationale for fat-supressing a sequence is twofold: either to bring out the signal created by the presence of contrast material, or to make areas of edema / fluid more conspicuous. STIR is not used to underline areas of enhancement on post-contrast images for different reasons. Its main use is to fat-suppress wide areas of anatomy, which render chemical shift saturation unreliable, in order to make pathology - water - more conspicuous.


Longer TE	Shorter TE

Note the differences between the two coronal STIR images above: Both of them are technically acceptable. The image to the left has a substantially longer TE than the one on the right (TEeff 30 msec.). As a result, the signal from hematopoietic elements has dropped substantially on the left, making the focus of fibrous dysplasia more conspicuous. The image on the right, with its shorter TE, still shows a significant degree of brightness in areas of hematopoietic marrow, and therefore, the areas of AVN are relatively less conspicuous.

Similarly, notice how the muscles on the left are darker than on the right.

TI = 170 msec.

TI = 130 msec.

INVERSION TIME:

Finally, don't forget that the TI is dependent on the strength of the magnet. At 1.5 T, the TI = 150-160 msec.. However, at 1.0 T, such as in the example above, the TI should be modified accordingly: the image on the left misses completely the null point of fat using a TI of 170 msec., whereas the image on the right provides homogeneous fat-suppression by using a TI = 130 msec.