Keywords
3 Tesla - 3D sequence - abscess - perianal fistula - ramification - VIBE
Introduction
The perianal fistula is an abnormal tract between the perianal skin and the anal canal.
Multiple secondary tracts may also exist and extend from the same primary opening.
It is characterized by chronic purulent discharge or cyclic pain associated with reaccumulation
of the abscess followed by spontaneous intermittent decompression.[1] Because of their high propensity to recur, perianal fistulas significantly contribute
to morbidity,[2] particularly when it comes to incontinence and poor quality of life that necessitates
recurrent surgical procedures. Of all patients with an initial perianal abscess, up
to 65% will develop a chronic or recurrent perianal fistula.[3]
Recurrence usually occurs because of secondary tracts that cannot be detected during
surgery and, therefore, cannot be fully treated.[4] Therefore, accurate preoperative assessment is essential for surgical success.
Until recently, imaging methods played a limited role in evaluating perianal fistulas.
Magnetic resonance imaging (MRI), on the other hand, now offers more accurate details
on the anatomical makeup of the anal canal, the anal sphincter complex, the relationship
between the fistula and the pelvic floor organs, and the levator ani muscle. MRI can
also accurately delineate the fistulous tract and detect additional tracts or abscesses.
This reduces the risk of recurrence and prevents side effects such as fecal incontinence
by providing precise information for appropriate surgical treatment.
Currently, conventional sequences for MRI scans of anal fistulas include two-dimensional
(2D) T2-weighted imaging (T2WI), 2D T1-weighted imaging (T1WI), inversion recovery
(IR), diffusion-weighted imaging, and postcontrast 2D T1WI. Perianal anatomy is best
evaluated using the 2D T1WI sequence, whereas a fistula or abscess is best evaluated
using the 2D T2WI and 2D T1WI postcontrast sequence.[5] There are still many things that need to be improved in conventional MRI sequences.
Because of their secretive-filled nature, perianal fistulas and perianal structures
demonstrate strong signals in non-fat T2WI, resulting in poor image contrast and risking
a poor diagnosis. Suppression of fat signals can significantly improve the contrast
of the fistula in fat-suppressed (FS) T2WI but leads to poor visualization of the
perianal anatomy.[2]
[6]
In contrast, three-dimensional (3D) sequences provide high-quality multiplanar reconstruction
images. It also has the advantage of reducing the total imaging time and eliminating
the need for additional 2D images in other planes.[7] A 3D model can help surgeons understand the complex relationship between sphincter
anatomy and fistula, guiding surgical decisions, and improving outcomes. In a study
by Cerit et al, they observed that contrast-enhanced (CE) 3D sequences (T1 SPACE and
CE 3D T1 VIBE) outperform conventional 2D sequences in the preoperative evaluation
of perianal fistulas with a shorter scanning time.[8] In the study done at our tertiary care center, we assessed the role of volumetric
MRI sequences (CE 3D T1 SPACE and CE 3D T1 VIBE) over conventional (CE 2D T1WI) MRI
sequences in the presurgical assessment of perianal fistula.
Materials and Methods
This prospective observational study was conducted in the department of radiology
in a tertiary care hospital after obtaining approval from the hospital's ethics and
scientific committee. The patients presented to our hospital with perianal pain, fever,
tenderness, redness, and discharge, whose physical examination findings were consistent
with perianal fistula, and those who underwent CE MRI of anorectum followed by surgery
within 2 weeks, from December 2020 to July 2022 were included in the study. Patients
who had previously undergone perianal fistula surgery and those unwilling to undergo
surgery were excluded. A total of 44 patients were enrolled in this study. Among these
44 patients, 12% were associated with Crohn's disease, and 2 patients had anal fissures.
MRI Technique
MRI was performed in a 3T (Siemens 3T MAGNETOM Skyra). 2D and 3D MRI protocols were
performed on all the patients. Conventional (2D) sequences followed by postcontrast
2D sequences (in coronal and axial planes) orthogonal to the anal canal and 3D T1W
sequences were taken ([Table 1]). To prevent bias, the postcontrast sequences were performed in random order. This
was done by implementing a randomized block design for MRI sequences. The sample population
was divided into distinct blocks based on variables and within each block, a randomization
method (computer-generated random numbers) was used to assign the postcontrast sequences
(2D T1, 3D VIBE, and 3D SPACE) in a random order. The random sequence assignment process
for each block was repeated independently. MRI scans were done according to the randomized
sequence order for each participant, and the results and relevant data were recorded.
Table 1
MRI sequence parameters for CE T1WI, CE 3D T1 VIBE, and T1 SPACE
|
FOV (cm)
|
TR (ms)
|
TE (ms)
|
Voxel size
|
TA
|
|
T1WI (axial)
|
25
|
636.0
|
12.0
|
0.8 × 0.8 × 3.0 mm
|
3 min 56 s
|
|
T1WI (coronal)
|
25
|
654
|
12
|
1.0 × 1.0 × 3.0 mm
|
3 min 55 s
|
|
VIBE
|
17
|
4.26
|
1.58
|
0.7 × 0.7 × 0.9 mm
|
2 min 56 s
|
|
SPACE
|
20
|
504
|
24
|
0.8 × 0.8 × 1.0 mm
|
3 min 56 s
|
Abbreviations: CE, contrast-enhanced; FOV, field of view; MRI, magnetic resonance
imaging; TA, time of acquisition; TE, echo time; TR, repetition time; T1WI, T1-weighted
imaging; 3D, three-dimensional.
Evaluation of MRI Findings
The MRI was evaluated by a radiologist with more than 15 years' experience in abdominal
MR reporting. Following the assessment of conventional MRI, the results of postcontrast
MRI were evaluated in series. In all the patients, CE 2D T1WI, CE 3D T1 SPACE, and
CE 3D T1 VIBE sequences were analyzed in random orders to avoid bias. All findings
were assessed and scored. The confidence scores of all three datasets were recorded
in terms of artifacts, overall image quality, the type of the fistula, visibility
of internal orifices, presence of ramifications, the number of detectable internal
orifices, and the presence of abscesses. The types of fistula-in-ano are intersphincteric,
transsphincteric, suprasphincteric, and extrasphincteric fistulas.[9] Ramifications or secondary tracts are the tracts from the primary tract that cross
from one side to the other (horseshoe type) or may be visualized as a side track within
the ipsilateral intersphincteric plane in a patient with an intersphincteric fistula.
They can also be seen as a tract from the primary tract extending into the ischioanal/ischiorectal
fossae in trans/suprasphincteric fistulas. Scan time was recorded for every sequence.
The performance of each sequence in terms of image quality, presence of artifacts,
fistula type, presence of abscess, and visibility of internal orifice were compared.
The surgery was done immediately after the MRI scan in simple fistulas and after 1
to 2 weeks of antibiotic therapy in complex cases by a surgeon with more than 20 years
of experience. After the surgery, surgical reports were recorded. The postsurgical
findings were defined as the reference standard. Finally, an agreement between MRI
findings and intraoperative findings was evaluated.
Qualitative Analysis
The visibility of the origin point in the anal channel (visibility of the internal
orifice) is defined, and the rating system was as follows: 0, not visible; 1, visible.
The number of internal orifices was also counted. A five-point scale was used for
the confidence level of image artifacts: 1, unreadable artifacts, images of nondiagnostic
quality; 2, severe artifacts, images degraded but interpretable; 3, moderate artifacts
with a few but not serious effects on diagnostic quality; 4, minimal artifacts, no
effect on diagnostic quality; and 5, no artifacts.[10] A five-point scale was also used for the confidence level of overall image quality:
1, unacceptable; 2, faint; 3, acceptable; 4, fine; and 5, perfect or ideal.
Statistical Analysis
Statistical analyses were performed using SPSS Statistical Software, version 20.0.
Categorical and quantitative variables were expressed as frequency (percentage) and
mean ± standard deviation, respectively. Sensitivity and specificity were calculated
to determine the diagnostic accuracy of MRI sequences. The chi-square test was used
to determine the association between categorical variables. For all statistical interpretations,
p < 0.05 was considered the statistically significant threshold.
Results
A total of 44 patients were included in the study. The sample group's mean age was
38.8 years, with a range between 13 and 59 years. Among 44 patients, males were more
in number accounting for 72% of the total and females were 28% of capacity.
Fistula Type
Among 44 patients, intraoperatively the majority showed transsphincteric (44%) and
intersphincteric (44%) fistula types (i.e., 19 patients each). Suprasphincteric kind
of fistula is seen in two patients. Four patients showed both transsphincteric and
extrasphincteric types of fistula. There was 100% agreement of above-mentioned intraoperative
findings with preoperative MRI findings in all three sequences ([Fig. 1]).
Fig. 1 Postcontrast images in coronal plane: transsphincteric fistula (yellow arrow) on
right side seen on 3D T1 space (A), 3D T1 VIBE (B), and 2D T1 (C) sequences. 2D, two-dimensional;
3D, three-dimensional.
Abscess
Abscesses were identified in 28 patients (64%) intraoperatively, and the remaining
16 patients (36%) did not have any abscesses.
Among the 28 patients with abscesses, 16 patients were detected to have abscesses
in all three sequences in preoperative MRI ([Fig. 2]). The presence or absence of abscess was identified correctly in 41 patients out
of 44 in preoperative MRI, with a sensitivity of 89% and specificity of 100% in all
three sequences ([Table 2]). In the rest of the three complex patients, since abscess was not picked up in
any of the conventional/postcontrast sequences, abscess was thought to have developed
between the MRI acquisition and surgery.
Fig. 2 Extrasphincteric abscess (yellow arrow) demonstrated as nonenhancing focus with peripheral
enhancement on 3D T1 space (A, sagittal plane), 3D T1 VIBE (B, sagittal plane), and
2D T1 (C, axial plane) sequences. 2D, two-dimensional; 3D, three-dimensional.
Table 2
Agreement of abscess and ramifications in MRI with intraoperative findings
|
MRI sequence
|
Finding on MRI
|
Intraoperative findings
|
Sensitivity
|
Specificity
|
|
Present
|
Absent
|
|
Presence of abscess
|
|
T1 VIBE
|
Present
|
25
|
0
|
89.29%
|
100%
|
|
Absent
|
3
|
16
|
|
|
|
T1 SPACE
|
Present
|
25
|
0
|
89.29%
|
100%
|
|
Absent
|
3
|
16
|
|
|
|
2D T1WI
|
Present
|
25
|
0
|
89.29%
|
100%
|
|
Absent
|
3
|
16
|
|
|
|
Presence of ramifications
|
|
T1 VIBE
|
Present
|
14
|
0
|
100%
|
100%
|
|
Absent
|
0
|
30
|
|
|
|
T1 SPACE
|
Present
|
12
|
0
|
85.71%
|
100%
|
|
Absent
|
2
|
30
|
|
|
|
2D T1WI
|
Present
|
5
|
0
|
35.71%
|
100%
|
|
Absent
|
9
|
30
|
|
|
Abbreviations: MRI, magnetic resonance imaging; T1WI, T1-weighted imaging; 2D, two-dimensional.
Ramifications
Ramifications were seen in 14 patients (32%) intraoperatively, and the remaining 30
patients (68%) did not show any ramifications. Among the 14 patients with ramifications
in preoperative MRI, 14 patients showed ramifications in CE T1 VIBE and 11 patients
in CE T1 SPACE ([Figs. 3] and [4]). However, in the CE 2D TIWI sequence, only five patients showed ramifications.
CE 3D VIBE, CE 3D SPACE, and CE 2D T1WI had a sensitivity of 100, 86, and 36%, respectively,
for identifying ramifications. All three sequences had a specificity of about 100%
([Table 2]).
Fig. 3 Transsphincteric fistula with two ramifications (yellow and red arrows) demonstrated
on 3D T1 space (A), 3D T1 VIBE (B), and 2D T1 (C) sequences in coronal plane. 2D,
two-dimensional; 3D, three-dimensional.
Fig. 4 Postcontrast images in axial plane: Secondary ramification with adjacent inflammation
(yellow arrow) in the left anterolateral aspect of anal canal clearly visualized on
3D T1 VIBE sequence. 3D T1 space (B) and 2D T1 (C) sequences demonstrate inflammatory
changes at the site of secondary ramification. The tract was not clearly demonstrated
in these sequences. 2D, two-dimensional; 3D, three-dimensional.
Number and Visibility of Internal Orifices
In this present study population of 44 patients, surgery revealed that all 44 patients
had internal orifices. In surgery, 33 patients had a single internal orifice, and
the remaining 11 patients had more than one internal orifice (seven patients—two internal
orifice; two patients—three internal orifice, and two patients—four internal orifice).
In preoperative MRI, the number of internal orifices was identified correctly for
all 44 patients (100%) in the CE TI VIBE sequence and 40 patients (92%) in the CE
T1 SPACE sequence. However, the number of internal orifices was identified correctly
in only 35 patients (80%) in the CE 2D T1WI sequence ([Fig. 5]).
Fig. 5 Postcontrast oblique axial images: Internal opening at 7'o clock position (yellow
arrow) demonstrated in on 3D T1 VIBE (A), 3D T1 space (B), and 2D T1 (C) sequences.
The internal opening at 9'o clock position (red arrow) was demonstrated on 3D T1 VIBE
(D) and 3D T1 space (E) sequences and not in 2D T1 image (F). 2D, two-dimensional;
3D, three-dimensional.
The surgery revealed more than one internal openings in 11 patients. On preoperative
MRI evaluation, all internal orifices were visible in all 11 patients in CE T1 VIBE
sequence cases (sensitivity—100%). However, the number of internal orifices was same
as in surgery in seven patients in CE T1 SPACE sequence and only in four patients
in the CE 2D TIWI sequence. The sensitivity of 3D SPACE and 2D T1 images in identifying
more than one internal orifice were 63 and 36%, respectively.
Image Artifacts
Overall artifact scores are higher in CE 2D T1WI sequence than CE T1 SPACE and CE
T1 VIBE sequences, with no significant difference in artifact scores between CE T1
SPACE and CE T1 VIBE sequences, as detailed in [Table 3]. The higher degree of CE 2D T1WI sequences artifacts could be due to increased scanning
time and patient movements.
Table 3
Distribution of the sample according to image quality and artifacts
|
MRI
|
Number of patients
|
|
T1 VIBE
|
T1 SPACE
|
2D T1WI
|
|
Image quality
|
|
Acceptable
|
0
|
2 (4%)
|
0
|
|
Fine
|
14 (32%)
|
7 (16%)
|
14 (32%)
|
|
Perfect
|
30 (68%)
|
25 (80%)
|
30 (68%)
|
|
Artifacts
|
|
Moderate
|
2 (4%)
|
2 (4%)
|
0
|
|
Minimal
|
9 (20%)
|
11 (24%)
|
19 (44%)
|
|
None
|
33 (76%)
|
31 (72%)
|
25 (56%)
|
Abbreviations: MRI, magnetic resonance imaging; T1WI, T1-weighted imaging; 2D, two-dimensional.
Acquisition Time
In this present study, the acquisition time for each CE T1 SPACE and CE T1 VIBE sequences
was shorter compared with combined CE 2D T1WI sequences with no compromise in image
quality.
Mean acquisition time for CE T1 SPACE and CE T1 VIBE sequences were found to be 3 minutes
56 seconds and 2 minutes 56 seconds, respectively, whereas the mean acquisition time
for CE 2D T1 sequences (axial and coronal planes) was found to be 7 minutes and 54 seconds,
which is more than the time required for both CE T1 SPACE and T1 VIBE combined.
Image Quality
There is no significant difference in overall image quality between sequences ([Table 3]).
Discussion
MRI has a significant role in presurgical assessment of perianal fistula because it
can demonstrate the tract of a primary fistula, associated pelvic structures, hidden
areas of sepsis, and secondary extensions, thereby increasing the success rate of
surgical treatment.
The introduction of 3T MRI has been a game changer, as it introduced high spatial
resolution isotropic 3D data acquisition, which is invaluable for perianal fistula
mapping.[11] The main advantages of 3D compared with 2D imaging techniques are free slab positioning,
operator independence, coverage of a larger area with thinner sections without intersection
gaps, and increased signal-to-noise ratio (SNR) achieved with reduced imaging time.
Every manufacturer has their 3D imaging technology: Siemens (VIBE, SPACE), GE (LAVA),
and Philips (THRIVE, VISTA).
VIBE and SPACE were used in the present study. VIBE is a spoiled gradient-echo sequence
that can produce T1WI with high contrast and spatial resolution in a relatively short
acquisition time.[12] 3D SPACE is another volumetric MRI sequence that has recently been used in 1.5 and
3 Tesla scanners to produce different imaging weighting, including T1, T2, proton
density, T1W fluid-attenuated inversion recovery (FLAIR), T2-FLAIR, double IR, and
phase-corrected IR techniques.[13]
[14] SPACE sequence has a low specific absorption rate value because of low flip angles.
A low flip angle results in higher spatial and contrast resolution.[15]
CE 2D T1WI requires multiple planes for precise perianal fistula mapping. Oblique
coronal and oblique axial images must be acquired to assess the fistula tract's anatomical
details and show supralevator and skin extension. Retrospective postprocessing of
CE 2D T1WI is impossible in case of errors in the field of view or imaging planes.
However, CE 3D SPACE and 3D VIBE imaging overcome those limits. They are used to perform
isotropic data acquisition on a single slab, allowing multiplanar reformatting in
different imaging planes to improve examination efficiency. Free slab positioning,
operator independence, and increased SNR are other advantages of 3D imaging. The 3D
acquisition has the ability to cover a larger area with thinner sections without intersection
gaps in a shorter imaging time. In the present study, the overall acquisition time
for each 3D sequence was shorter compared with combined 2D lines (oblique axial and
oblique coronal), with no significant difference in image quality. The results mentioned
earlier were similar to the study done by Cerit et al.[8]
Image artifacts are less in 3D sequences than 2D sequences, possibly due to increased
table time in 2D sequences leading to patient movement. In this study, the type of
fistula was correctly categorized in all patients (100%) on the CE 3D T1 SPACE, CE
3D T1 VIBE, and CE 2D T1WI sequences. Thus, in assessing the type of fistula, this
study did not show any statistically significant differences between 2D and 3D sequences.
Similarly, in a study conducted by Yildirim et al,[16] the contribution of various MRI sequences in determining the type of perianal fistula
and for obtaining critical information for surgical decisions, as well as defining
the optimal combination of sequences for readers with varying levels of experience
were assessed. In their study, for all three readers, a statistically significant
concordance was obtained between fistula classifications and surgical findings with
the postcontrast FS T1WI. The p-values were 0.022, 0.011, and 0.043, respectively. The authors concluded that a high-resolution
MRI of the perianal fistula with FS CE T1WI sequence showed statistically significant
agreement with surgical findings in perianal fistula classification and had a low
interobserver variance.
The precise identification and localization of internal orifices have an important
role in reducing the recurrence rate and guiding surgical treatment in the perianal
fistula. In this study, all internal orifices could be viewed on CE 3D T1 VIBE. On
CE 3D T1 SPACE sequences, internal orifices were correctly depicted in 92% of the
cases, and on CE 2D T1WI sequences, internal orifices were defined only in 80% of
the cases. In support of our study, various researchers have shown that 3D MRI sequences
are superior to 2D sequences in representations of the internal orifice of the perianal
fistula. A study done by Torkzad et al[17] stated that postcontrast 3D T1-weighted gradient echo sequence with fat saturation
(THRIVE) is superior to 2D T1 fat saturation (spectral presaturation with inversion
recovery) in the depiction of the internal orifice. In their study, postcontrast 3D
T1-weighted sequence showed better agreement compared with postcontrast FS 2D T1 sequences
in the description of fistulae (Cohen's kappa = 0.94 or strong agreement), identifying
internal openings (Cohen's kappa = 0.97 or perfect agreement), and evaluating inflammation
(Cohen's kappa = 0.94 or strong agreement). The corresponding counts for the postcontrast
2D FS T1 sequences were 0.71 (good concordance), 0.66 (good concordance), and 0.87
(strong concordance), respectively. In another study by Zhao et al, they compared
the CE FS 3D T1-weighted VIBE and FS T1-weighted turbo spin echo (TSE) sequence in
characteristics of anal fistula.[18] They concluded that 3D T1 VIBE was more accurate and more valuable than 2D TSE T1
images in the assessment of anal fistula on evaluating the position of internal opening,
secondary tracts, and classification of the complex fistula.
In our study, the ramifications were correctly identified in all patients in the 3D
T1 VIBE sequence and 11 out of 14 patients in the 3D T1 SPACE sequences. Ramification
was identified in only five patients in postcontrast 2D T1 sequences. Consequently,
3D sequences played a major role in treatment planning in our study. Similarly, Schaefer
et al[19] demonstrated that high-resolution subtraction MRI fistulography consisting of high-resolution
3D T1-weighted gradient echo sequences (3D FLASH) and the image subtraction technique
were important complements to surgical exploration, especially in the case of complex
perianal fistulas. The study found that 90% of cases had complete agreement between
MR fistulography and surgery.
Beyond the different study designs characterizing all the studies mentioned earlier,
a common result is represented, that is, in the evaluation of perianal fistula, especially
in complex fistulous tracts with ramifications and multiple internal orifices, the
higher accuracy is for CE 3D T1 sequences over CE 2D T1WI sequence. The main limitation
of our study was the smaller sample size. Further research with a large number of
participants is necessary. The cases were examined by only one radiologist. It would
have been beneficial if two experienced radiologists had seen it and analyzed the
interreader agreement. In addition, there is a possibility of bias because all the
contrast sequences were read during the same session.
Conclusion
In conclusion, CE 3D T1 sequences outperformed CE 2D T1 sequences, particularly in
the evaluation of complex perianal fistulas in terms of visibility of internal orifices,
number of internal openings, and identification of ramifications with a shorter scanning
time in our study. Of the 3D sequences, CE 3D T1 VIBE was slightly better than the
CE 3D T1 SPACE sequence in our study. However, the relatively small number of patients
in this study requires this statement be validated by further research involving a
larger number of patients.
