We have demonstrated the feasibility of DTI with tractography as a new method to identify the complex morphology and architectural organization of the anal canal sphincters, also providing relevant quantitative information.
To our knowledge, this is the first evaluation of DTI of the anal canal for the follow up of treated perianal fistulas and the 3D color-coded fiber-tracking resulted in a satisfied viewing of the complex muscles fibers shape and orientation.
Our study is in concordance with initial anatomical studies that demonstrated the potential of in vivo DTI with 3D representation, to illustrate the levator ani musculature , the anal canal  or to explore the female pelvic floor .
With this noninvasive method, which does not require any contrast agent, we reached an excellent image quality, made possible by the overwhelming recent improvements of MR technology.
The circular configuration of the internal sphincter and puborectalis were clearly demonstrated as well as the combination of craniocaudal and anteroposterior orientation of the external sphincter.
Our inter-observer concordance was substantial with a reasonable acquisition and processing time with 64 diffusion-weighted directions, meaning that this evaluation could be applicable in clinical routine. Furthermore, some studies on the urethral sphincter advocated that encoding fewer diffusion directions even allowed good fiber tracking with a shorter acquisition time [13, 14].
The DTI quantitative parameters provide with valuable information regarding the microstructure of fibers by assessing the direction and extent of random movement of water molecules. Additionally, the ADC value is an indicator of the extent of irregular Brownian motion of water molecules in tissue. Thus, higher ADC value indicates higher range of irregular movement by unit of time.
The FA value (i.e. ranging 0 to 1) is an indicator of the direction of this movement. For instance, a FA value around 0 indicates a quasi isotopic diffusion, meaning a uniform distribution of all directions of movement, while higher FA values indicate more ordered and directional movements inside the microstructure.
In our study we demonstrated anisotropy for all segments of the anal canal, in agreement with previous studies, meaning that the water molecules move preferentially in a certain direction within the muscular fibers.
Regarding the normal DTI values of the anal canal Goh et al , described mean FA values (with standard deviation) of 0.337 ± 0.049; 0.415 ± 0.072 and 0.361 ± 0.089 to 0.407 ± 0.062, for the internal, external sphincters and puborectalis respectively. These values were in the same range as those we recorded in our normal group, except for the external sphincter which was less directional in our study. Additionally, their mean ADC values (in 10− 3 mm²/s) for the internal, external sphincters and puborectalis were 1.59 ± 0.19, 1.51 ± 0.28 and 1.54 ± 0.29, respectively, which are also in agreement with what we observed in our normal group, thus meaning similar intensity of the molecular water movement.
When considering the unaffected (normal-appearing) puborectalis muscle as reference, Wang et al  described DTI parameters (in active perianal fistula) close to what we observed in terms of mean FA (0.382 ± 0.084 vs 0.343 ± 0.070), which meant that there is a similar direction of the water movement within fibers. If the mean ADC value (in 10− 3 mm²/s) was significantly higher (1.703 ± 0.432 vs 1.864 ± 0.336), the SD was quite large, so that we can consider the obtained values as overlapping.
Zijta et al [15–16] reported lower DTI values of normal pelvic floor muscles in women, with lower mean FA (0.25 ± 0.04 and 0.30 ± 0.04) and ADC (ranging from 1.30 x 10− 3 mm²/s ± 0.08 to 1.42 x 10− 3 mm²/s ± 0.28) values. However, they evaluated the anal canal sphincters globally, with no distinction between the internal and the external sphincters.
If Goh et al.  described higher value for FA in the striated anal canal musculature, in our study, there was no significant difference in terms of of FA between the internal anal sphincter (i.e., smooth muscle) and the other (i.e., straited) muscular components. In our work for both groups, the muscular type (i.e., smooth or striated) did not seem to interfere with the value of anisotropy. The explanation cannot be attributed to the small size of our cohort, as it is similar to Goh et al. (n = 25), but could be related the type of patients. Goh et al. patients were older men with prostatic carcinoma,  while our patients were younger and from both genders. However, our results were concordent regarding the ADC value in the normal group, thus indicating a similar overall diffusivity. To note, Goh et al. recorded similar FA value of the striated gluteus maximus muscle to what we recorded for the internal sphincter. Additionally, the FA value of the striated puborectalis muscle observed by Wang et al.  was in the same range as what we observed for the smooth internal sphincter. In parallel, two recent papers on the male urethral sphincter [13, 17], described significantly higher FA values in the proximal sphincter, but this anatomical part is mainly composed of smooth muscle fibers.
As a result, we cannot confirm the proposed hypothesis that the FA value varies according to the muscular type (i.e., smooth vs striated muscle), with higher values for striated muscle.
Measurements of anisotropy may provide additional information of microstructural muscular disruption secondary to any types of injury such as trauma, pelvic radiotherapy, fistulous disease, or tumor. It is anticipated that such injury may decrease anisotropy values because of loss of the normal fiber orientation. Surprisingly, there was no statistically significant difference in terms of quantitative parameters of DTI whether the morphology on T2-weighted images was normal or abnormal, in our cohort of patients with healed anal fistula, most of which were asymptomatic. A possible explanation for this finding is a complete recovery of muscular fibers in healed anal fistula, with normal water movement inside the muscle fibers. The persistent pain in 13 patients remains unexplained on both DTI and T2-weighted images, especially as 2/7 patients had an abnormal sphincter on T2-weighted images. Therefore, DTI appears more pertinent than the routine T2-based morphological evaluation of the muscular components of anal canal to assess healing.
However, some limitations can be noted.
Firstly, we are aware that our cohort is quite small (n = 21) but to our knowledge there is no larger cohort in the literature. Additionally, the DTI values in our normal group are in line with recently published data.
Secondly, this work is observational with no correlation to histology or confrontation with other non-invasive imaging techniques (e.g., high-resolution ultrasound or manometry). However, even if the morphological analysis on T2-weighted imaging is a subjective parameter, this is the standard of care to detect anal canal abnormalities on imaging and we also took the clinical data into account.
Thirdly, DTI should probably be considered only as a biomechanical tool to delineate muscle fibers, rather than a quantitative tool for the functional evaluation of the anal sphincters.
Lastly, we used a 3T MRI, to ensure a higher signal than with a lower field magnet of 1.5T, as in other studies. The values of the quantitative parameters should not be vendor specific, but could differ with the magnetic field (1.5Tvs 3T) . Thus, this technique should be evaluated on lower field MRI (1.5T) which is more commonly found worldwide, despite growing access to 3T MRI.