Eric G. Stinson1, Joshua D. Trzasko1, Eric A. Borisch1, Phillip M. Young1, Joel G. Fletcher1, and Stephen J. Riederer1
1Radiology, Mayo Clinic, Rochester, MN, United States
Synopsis
Perianal fistula images with high spatial resolution, high SNR, and excellent fat suppression were achieved with a multi-TE interleaved acquisition and a constrained-phase graph cuts-based Dixon fat/water separation.Purpose
A perianal fistula is an abnormal connection between the perineum and
the anal canal (1). The current clinical protocol for imaging
perianal fistulas at our institution consists of multi-planar T2-weighted
imaging with and without fat saturation, multi-planar T1-weighted
imaging with a gadolinium-based contrast agent (2) with and without fat saturation, and a 3D
post-contrast LAVA Flex (2-pt Dixon) sequence. While these techniques have
proven satisfactory for identifying tracks not seen surgically, it is also of
interest to acquire 3D images with greater spatial resolution for improved
treatment of this debilitating disease. However, the product sequence places limits on
the spatial resolution and field of view (FOV) due to gradient limitations. The
purpose of this work is to demonstrate T1-weighted, contrast-enhanced perianal
fistula imaging with an interleaved multi-TE 3D spoiled gradient echo technique
that can provide high spatial resolution, excellent fat suppression, and high
SNR.
Methods
An interleaved multi-TE
3D spoiled gradient echo technique was developed to avoid the gradient
limitations of the current (multi-TE in a single TR) clinical sequence. Due to
the nature of this research study, post-contrast imaging was performed after
completion of the clinical imaging protocol – about 20 minutes after contrast
administration. Imaging parameters are shown in Table 1. The study shown here
was performed at 3.0T, however, similar studies have been performed at 1.5T
with only minor modifications to the pulse sequence and reconstruction to
account for the different rate of fat dephasing at the two field strengths. The dual-echo acquisition (with
corner cutting in k-space, but no other undersampling) was reconstructed with a Dixon-based fat/water separation with a phase constrained signal model (3) and graph-cuts optimization scheme similar to that in Reference (4). The scheme used here, however, differs from that in (4) in that only 2 echoes are required, and the initial phase of the water and fat signals are constrained to be equal.
Results
Figure 1 shows the acquired
axial images and sagittal and coronal reformatted images from a 34 year-old
female with a healing perianal fistula. The fistula is well depicted in all
images (red arrows), despite the delay between contrast injection and imaging. Good
fat suppression was achieved with the phase constrained Dixon technique, and
the nearly isotropic resolution allows viewing in all three planes.
Compared to the current
clinical protocol performed on this patient, (resolution=1.33×1.33×4mm), the
high spatial resolution images benefit from improved depiction in reformatted
planes and provide the opportunity to better model the fistulas in three
dimensions. While previous work has performed high spatial resolution perianal
fistula imaging with intermittent fat suppression (5), the Dixon technique provides additional
SNR improvement due to the averaging effect of using two appropriately timed echoes
(6). As clinical translation continues, the reduced delay between injection of contrast and imaging should further improve the depiction of the perianal fistulas. Future work will aim to further reduce
imaging times, particularly through the use of parallel imaging and iterative reconstruction techniques.
Conclusion
Dixon-based CE-MRI has achieved successful imaging
of perianal fistulas with high spatial resolution, high SNR, and excellent fat
suppression.
Acknowledgements
Funded by NIH. Grant Numbers: EB000212, RR018898References
1. De Miguel Criado J et al. RadioGraphics
2012;32:175–194.
2. Spencer JA et al. Am. J. Roentgenol. 1996;167:735–741.
3. Bydder M et al. Magn. Reson. Imaging 2011;29:216–221.
4. Hernando D et al. Magn. Reson. Med. 2010;63:79–90.
5. Loening AM et al. ISMRM. Milan, Italy; 2014. p. 2126.
6. Stinson EG et al. Magn. Reson. Med. 2015;74:81–92.