Abstract
Anorectal malformations (ARMs) are congenital anomalies of the distal part of the hindgut often associated with sacral and/or spinal anomalies. We investigated anatomical and microstructural properties of the lumbosacral plexus of ARM patients from imaging data. Twenty-five patients (16 males), median age 4 months (2–49), 13 high and 12 low ARM, underwent 3 Tesla magnetic resonance imaging with diffusion tensor sequences (dMRI) before repair. A 3D model was built from manual segmentation and used to guide novel AI algorithms for the segmentation of the nervous pelvic network. Volume and diffusion parameters were obtained for each root (L5 to S4) and compared among patients with high and low ARMs using a nonparametric Wilcoxon test. Comparison was also made between the groups with (n = 9) or without (n = 16) sacral and/or spinal cord anomalies. When compared with low ARMS, high ARMs exhibited the following: a smaller volume of S1, S2, and S3 roots and of S1 and S3 for patients without sacral and/or spinal cord abnormalities; an overall significant alteration of the roots micro-architecture reflected by a diminution of the fractional anisotropy and an increase of the axial diffusivity and radial diffusivity measures. This first analysis of the lumbosacral plexus from dMRI in children with ARMs shows differences in the development and microarchitecture of the lumbosacral nerve roots between high and low ARMs. This observation supports the hypothesis that high ARMs may result from a more regional developmental abnormality than low ARMs and open new ways to visualize and assess the lumbosacral plexus in children and adults.
Keywords: Anorectal malformation, Magnetic resonance imaging, Diffusion tensor imaging, Tractography, Anatomical knowledge representation, Nerve segmentation and recognition
Introduction
Anorectal malformations (ARMs) are congenital malformations of the distal part of the hindgut with a wide spectrum of severities which vary from the simple mislocated anterior anus or fistula, included in the term low ARMs, to the complex high ARMs with recto-urinary fistula in male or cloacal malformation in female. In these later cases, the continence may be of very poor quality, with more than 60% of patients requiring bowel management [1] and/or intermittent urethral catheterization for neurogenic bladder [2]. The incidence of spinal cord anomalies, increased by the presence of sacral anomalies [3], may be up to 34% and reach 65% in cloacal malformations [4, 5]. They are known to worsen the functional prognosis in patients with ARMs [6, 7].
Magnetic resonance imaging (MRI) is increasingly used in this context to better explore associated malformations such as spinal cord [4, 8] genital or urinary tract anomalies, musculature trophicity [5, 9] or to assess the anatomy of the anorectal tract [10]. Diffusion MRI (dMRI) is usually used in adults in this region to analyze muscular [11], sphincter [12, 13], and more rarely nervous [14, 15] structures; however, it has been rarely applied in the context of ARM. It has been shown to be useful to evaluate the anatomy of the malformation [16, 17] but also for comparing surgical techniques on musculature injury after ARMs repair [18–20]. To the best of our knowledge, only one study reports evaluation of the lumbosacral plexus based on conventional diffusion tensor imaging (DTI) imaging [21].
In this study, we make use of an AI-based segmentation and recognition method exploiting 3D tractography in order to study the lumbosacral plexus on routine MRI scans of children with ARM. Several methods for diffusion MRI reconstruction exist; among them, DTI [22] is the most popular one since it requires a low scanning time, but its results are highly dependent on the manual placement of Regions of Interest (ROI) for the starting and ending locations of the tracts [23]. In order to avoid the use of manual ROI placement and at the same time reduce the amount of false positives encountered in whole-body tractograms, we use an AI algorithm developed and already validated by our team [24–27] that exploits spatial relations between nerve bundles and segmented anatomical structures, in order to filter the tractography results and segment and recognize the nerves of interest. This novel approach made it possible to show differences between high and low ARMs and therefore contributing to the phenotypic description of the two types of malformation and supporting the hypothesis of a very different embryological origin. Above this application, this tool may be used in other pediatric pathologies such as tumors and adult’s pathologies requiring a better assessment of the lumbosacral plexus. However, this symbolic AI algorithm is not the focus of this article and will only be described in its main steps in the next section. We refer the reader to other more technical articles, such as [26, 27] for further technical details.
Materials and Methods
Population and Subgroups
From December 2015 to April 2020, patients with ARMs were included in a prospective study following the General Data Protection Regulation compliance requirements with ethics committee approval (N°2015-AO1705-44). Inclusion criteria were the presence of any type of ARMs, no contraindications for MRI examination, and family consent. Conventional work-up included renal, urinary tract, cardiac and medullary ultrasound (US), and MRI imaging on a 3 Tesla MRI (General Electric). Imaging was performed in all patients before any pelvic reconstructive surgery and any neurosurgical intervention. Sedation was obtained using a feeding bottle without medication nor general anesthesia.
ARMs were definitively classified after surgery according to the Wingspread International Classification [28], updated at the Krickenbeck Conference of 2005 [29]. This latter classification distinguishes between only two types of ARMs, low and high (Table 1).
Table 1.
Comparison of Wingspread and Krickenbeck classifications
| ARM | Wingspread classification |
| Low | Anal stenosis, anocutaneous fistula, anovestibular fistula |
| Intermediate | Recto vestibular fistula, rectovaginal fistula, anal agenesis without fistula, recto bulbar uretral fistula |
| High | Rectal atresia, anorectal agenesis with recto vaginal fistula, anorectal agenesis with recto uretral prostatic fistula, anorectal agenesis without fistula, cloacal malformation |
| ARM | Krickenbeck classification |
| Low | Anal stenosis, imperforate anus without fistula, anal agenesis with rectoperineal fistula or recto vestibular fistula |
| High | Anal or anorectal agenesis without fistula, rectal atresia, anorectal agenesis with recto-uretral bulbar fistula or recto-uretral prostatic fistula or recto vesical fistula or rectovaginal fistula, cloacal malformation |
Sacral and spinal cord anomalies were obtained from MRI analysis. Images were separately reviewed by one neurosurgeon (MZ) and one radiologist (LB) and anomalies were retained only if both reviewers agreed on the image analysis. Minor sacral abnormalities, originating beyond the 5th sacral vertebra and beyond the emergence of the sacral root S4 (four patients), were allocated to the group without anomalies. Medullary anomalies were classified according to the recognized dysraphism classification [30]. Patients with sacral and/or spinal cord anomalies were pooled in the same group due to their low number and classified as present or absent. The wide variety of anomalies observed, and the small numbers of patients in the series precluded a more refined stratification.
Data Acquisition
The acquisition protocol was developed for the exploration of the pelvic structural and nervous anatomy. A volumetric T2-weighted image was used for organ 3D segmentation and modeling and a diffusion MRI (dMRI) for the nervous system. dMRI sequence parameters were optimized for the processing and the visualization of the sacral plexus in young children. T2 images were acquired with a voxel size of 0.88 × 0.85x0.88 mm3, TE = 0.0676 s, and TR = 2.196 s. Diffusion images were acquired with 25 directions, b = 600, a voxel size of 1.25 × 1.25x3.5 mm3, TE = 0.0529 s, and TR = 4 s. In total, the two sequences amounted to 10 additional minutes to the patient’s examination time. The images were acquired with a 3 Tesla MRI (General Electric).
AI Segmentation Method
MRI acquisitions were pre-processed as follows. The T2-weighted and the dMRI images were denoised and corrected for artifacts [31]. To represent the nervous fibers, a tractogram (namely a fibers reconstruction) of the whole pelvic area was constructed from the dMRI images using the deterministic tractography DTI method [22] implemented in mrtrix3 [32], resulting in millions of fibers.
To distinguish the nerves of interest (roots L5 to S4) from the multitude of streamlines produced by the tractography DTI method, a symbolic artificial intelligence segmentation and recognition method was used integrating prior anatomical knowledge on the spatial and geometrical properties of the roots and their known relationships to other pelvic structures [24–27]. The approach belongs to the domain of symbolic spatial reasoning, a branch of AI that does not involve machine learning methods. The method is structured as follows:
In a first step, the spatial references were defined manually by a clinician from the T2-weighted images. The modeled structures were bones (hips, sacrum and L5 vertebra), iliac vessels, muscles (piriformis, obturator and levator ani), bladder, rectum, sacral canal, and sacral holes (S1 to S4). These are given as inputs together with the tractogram.
The nerve definitions are encoded in rule form in the algorithm and include directional, path, connectivity, and orientation information of the nerve bundles with respect to the anatomical structures. These definitions, expressed in natural language, are translated into mathematical formulas in the fuzzy set theory and then integrated into operational algorithms. Fuzzy definitions of spatial relations [33], similarly to [34], are used to cope with the intrinsic imprecision of nerve descriptions in natural language, as well as with the complexity of the spatial organization of organs and nerves.
Each nerve bundle is then recognized as a set of fibers of the tractogram satisfying the nerve definition with respect to the spatial references. The extraction of the recognized set of fibers is identified as the segmentation of the corresponding nerve bundle.
The developed method (reconstruction plus recognition) allows us to segment the fiber tracts without any reference manual segmentation of the nerves (very difficult to produce). However, for this reason, the algorithm is semi-automatic; in fact, in order to deal with false positives that are not properly filtered, the resulting nerve fibers were supervised by clinicians and corrected if necessary (Fig. 1). We would like to underline that nerves are highly difficult structures to segment; in addition, diffusion IRMs are very difficult images to analyze visually, and on the T2-w IRMs, only the big and/or proximal roots are clearly visible and not the distal ones; all these problems are amplified in pediatric patients. To the best of our knowledge, there are to date no methods to manually segment nerve fibers from L5 to S4. Given the difficulty in generating manual segmentations for nerves, a quantitative comparison cannot be made, we refer the reader to [26, 27] for qualitative comparisons.
Fig. 1.
A 3D imaging of L5, S1, and S2 roots showing lesser volume of roots in a 8 months old boy with high ARM (A) when compared with a 3-month-old boy with an low type of ARM (B). From top to bottom, the represented fibers are L5 nerve root (yellow), S1 nerve root (green), and S2 nerve root (yellow), while the represented bones are vertebra L5 (ivory-white), sacrum (sand-brown), and hips (ivory-white). The resulting nerve fibers were supervised and corrected by clinicians
Due to the variation in MRI acquisition quality and to differences in children’s sizes, the sacral plexus analysis was limited to the portion of the image below the highest point of the L5 vertebra.
Measures Analysis
Volumes of lumbo-sacral nerves (L5 to S4) were computed based on the voxels that were crossed by the fibers and thus expressed in mm3. The volume of each sacral root analyzed (from L5 to S4) resulted from the addition of the volumes of the right and left roots for this root.
The evaluation of the microstructural properties of nerve fibers was based on the study of four diffusivity parameters [35, 36]:
Fractional anisotropy (FA) measures the degree of anisotropic diffusion in all the voxels of the image and may vary between 0 (fully isotropic diffusion) and 1 (diffusion constrained to one single direction).
Axial diffusivity (AD) is the most prominent diffusion direction and represents more specifically the structure of the fiber.
Radial diffusivity (RD) is the axis perpendicular to the most important diffusion direction and seems to be correlated to the quality of the myelin of the tracked fibers.
Mean diffusivity (MD) is the mean of the diffusion intensities for every acquisition direction.
The distributions of all the diffusivity parameters along the points of every fiber composing the sacral roots were analyzed. Parameter distributions were found reproducible between all streamlines, allowing a root-based analysis.
The median of volumes and of each diffusion parameter was given for each root.
Statistical Analysis
The median nerve volume and the median value of the diffusivity parameters were compared using a nonparametric Wilcoxon test [37] in three different scenarios: (i) for the whole population between high and low ARMs, (ii) for patients without sacral and/or spinal cord anomalies between high and low ARMs, (iii) and for the whole population between patients with and without sacral and/or spinal cord anomalies. The results were considered statistically significant for a p value inferior to 0.05. The analysis was performed using R (v. 4.1.2) in the RStudio environment (v. 2021.09.1).
Results
Characteristics of the Population
Twenty-five patients were prospectively included. The population was composed of 16 males (64%) and 9 females, with a median age of 4 months (2–49 months). Thirteen patients presented high ARMs and twelve low ARMs. The median age was 4 months in both cases, 2–17 months for high ARM and 2–49 months for low ARM. Sixteen patients with a median age of 3 months (2–49 months) did not exhibit any sacral and/or spinal cord anomaly. Nine patients with a median age of 6 months (3–17 months) had sacral (n = 4), spinal cord (n = 4), or sacral and spinal cord anomalies (n = 1). Among them, six were high ARMs. Patients’ characteristics are given in Table 2 and age distribution in high (group A) and low (group B) ARMs in Fig. 2.
Table 2.
ARMs patients characteristics. Age in months. M male, F female
| Patients | ARM | Age (months) | Sex | Sacral anomalies | Medullary anomalies |
|---|---|---|---|---|---|
| 1 | Low | 2 | M | ||
| 2 | Low | 2 | F | ||
| 3 | Low | 2 | M | ||
| 4 | Low | 3 | M |
Filum lipoma Terminal syringomyelia |
|
| 5 | Low | 3 | F | ||
| 6 | Low | 3 | F | ||
| 7 | Low | 5 | M | ||
| 8 | Low | 11 | M |
Filum lipoma Syringomyelia from T5 to L1 |
|
| 9 | Low | 12 | F | ||
| 10 | Low | 14 | F |
S3-S4 fusion S5-coccyx agenesis |
|
| 11 | Low | 45 | M | ||
| 12 | Low | 49 | F | ||
| 13 | High | 2 | M | ||
| 14 | High | 2 | M | ||
| 15 | High | 3 | M | ||
| 16 | High | 3 | F |
S4 hypoplasia S5-coccyx agenesis |
|
| 17 | High | 3 | M | ||
| 18 | High | 3 | M | S4-S5-coccyx agenesis |
Filum lipoma Low conus medullaris |
| 19 | High | 4 | M |
S4-S5 right hemi agenesis Coccyx agenesis |
|
| 20 | High | 4 | F | ||
| 21 | High | 5 | M | ||
| 22 | High | 6 | F | ||
| 23 | High | 6 | M | Filum lipoma | |
| 24 | High | 7 | M | S4-S5-coccyx partial agenesis | |
| 25 | High | 17 | M |
Filum lipoma Syringomyelia from C6 to L1 Limited dorsal myeloschisis |
Fig. 2.
Age distribution in high ARMs (A) and low ARMs (B). Each point represents an individual. The bold lines represent the median values. The Y axis is in logarithmic scale
Volume Comparisons
Since the median of the volumes is a parameter that in itself is not informative (e.g., variation in age or subjects size), only the difference between the medians is analyzed here.
When considering the whole population (n = 25), the volume of the S1 (p = 0.008), S2 (p = 0.016), and S3 (p = 0.042) roots were significantly smaller in patients with high ARMs (n = 13) when compared with patients with low ARMs (n = 12), with a difference in median ranging from about 1000 mm3 for S1 to about 3000 mm3 for S3 (Figs. 1 and 3a).
Fig. 3.
Boxplot of sacral roots volumes from L5 to S4. a In the whole population, comparison between high ARMs (blue) and low ARMs (red). b In the group of patients without medullary or sacral anomaly (n = 16), comparison between high ARMs (blue, n = 7) and low ARMs (red, n = 9). c in the whole population, comparison according to the presence (blue, n = 9) or not (red, n = 16) of sacral, and/or medullary anomaly. Volumes are expressed in mm3. Results are reported for individual roots from L5 to S4. Individual boxplots report median, 25th and 75th percentiles. For each root, significative difference (p < 0.05) between high and low ARMs (or with and without anomalies) volume roots are indicated with a*
In the group of patients without any sacral and/or spinal cord anomaly (n = 16), the volume of S1 (p = 0.029) and S3 (p = 0.024) roots was significantly smaller in high ARMs (n = 7) when compared with patients with low ARMs (n = 9), with a difference in median of about 2000 mm3 for both roots (Fig. 3).
In the group of patients with sacral and/or spinal cord anomaly (n = 9), the volume of L5 (p = 0.034) root was significantly smaller when compared to the group of patients without these associated anomalies (n = 16), with a difference in median of about 2000 mm3 (Fig. 3).
Diffusion Characteristics Analysis
For diffusion measurements, these are informative in themselves, indicating fiber properties. For this reason, we believe that the analysis of medians combined with the Wilcoxon test is sufficient to understand these measures.
The FA, MD, AD, and RD of the lumbosacral roots were compared among patients with high (group A) and low (group B) ARMs.
For the whole population (n = 25), the FA of the L5 root was significantly lower (p = 0.006) and the RD higher (p = 0.024) in patients with high ARMs, when compared with patients with low ARMs (Table 3 and Fig. 4).
Table 3.
Median of diffusivity parameters in patients with high (n = 13) versus low (n = 12) ARMs. MD, AD, and RD in mm2/s × 103
| Roots | FA | MD | AD | RD | ||||
|---|---|---|---|---|---|---|---|---|
| High | Low | High | Low | High | Low | High | Low | |
| L5 | 0.33* | 0.38* | 1.57 | 1.45 | 2.31 | 2.16 | 1.33* | 1.10* |
| S1 | 0.41 | 0.39 | 1.57 | 1.55 | 2.37 | 2.49 | 1.14 | 1.10 |
| S2 | 0.38 | 0.43 | 1.49 | 1.43 | 2.33 | 2.18 | 1.27 | 1.03 |
| S3 | 0.37 | 0.32 | 1.62 | 1.51 | 2.34 | 2.16 | 1.20 | 1.34 |
| S4 | 0.52 | 0.41 | 1.57 | 1.50 | 2.50 | 2.33 | 1.11 | 1.13 |
*p < 0.05
Fig. 4.
Boxplot of sacral roots diffusivity parameters in patients with high (n = 13) versus low (n = 12) ARMs. a FA, b MD in mm2/s × 103, c AD in mm2/s × 103, and d RD in mm2/s × 103. Results are reported for individual roots from L5 to S4. Individual boxplots report median, 25th and 75th percentiles. For each root, significative difference (p < 0.05) between high and low ARMs diffusivity parameter for roots are indicated with a *
When considering only patients without any sacral or spinal cord anomaly (n = 16), the FA of the L5 root was significantly lower (p = 0.005) and the AD (p = 0.030) and RD higher (p = 0.003) in patients with high ARMs versus low ARMs. A higher RD was also found for the S1 root of patients with high ARMs (p = 0.021) (Table 4 and Fig. 5).
Table 4.
Median of diffusivity parameters in patients with high (n = 7) versus low (n = 9) ARMs, without sacral and/or medullary anomaly. MD, AD, and RD in mm2/s × 103
| Roots | FA | MD | AD | RD | ||||
|---|---|---|---|---|---|---|---|---|
| High | Low | High | Low | High | Low | High | Low | |
| L5 | 0.34* | 0.38* | 1.72 | 1.39 | 2.37* | 2.12* | 1.37* | 1.05* |
| S1 | 0.40 | 0.42 | 1.76 | 1.47 | 2.68 | 2.40 | 1.28* | 1.06* |
| S2 | 0.47 | 0.40 | 1.49 | 1.44 | 2.34 | 2.15 | 1.15 | 1.05 |
| S3 | 0.44 | 0.31 | 1.64 | 1.51 | 2.46 | 2.08 | 1.21 | 1.34 |
| S4 | 0.48 | 0.46 | 2.08 | 1.68 | 3.25 | 2.64 | 1.46 | 1.32 |
*p < 0.05
Fig. 5.
Boxplot of sacral roots diffusivity parameters in patients with high (n = 7) versus low (n = 9) ARMs, without sacral and/or medullary anomaly. a FA, b MD in mm2/s × 103, c AD in mm2/s × 103, and d RD in mm2/s × 103. Results are reported for individual roots from L5 to S4. Individual boxplots report median, 25th and 75th percentiles. For each root, significative difference (p < 0.05) between high and low ARMs diffusivity parameter for are indicated with a *
The FA, MD, AD, and RD of the lumbosacral roots were then compared between patients without and with sacral and/or spinal cord anomalies. The FA of S2 root was lower (p = 0.02) in patients cumulating these anomalies when compared with patients with only ARMs (Table 5 and Fig. 6).
Table 5.
Median of diffusivity parameters in patients with (n = 9) and without (n = 16) sacral and/or medullary anomaly. MD, AD, and RD in mm2/s × 103
| Roots | FA | MD | AD | RD | ||||
|---|---|---|---|---|---|---|---|---|
| High | Low | High | Low | High | Low | High | Low | |
| L5 | 0.35 | 0.33 | 1.44 | 1.64 | 2.16 | 2.60 | 1.13 | 1.33 |
| S1 | 0.40 | 0.39 | 1.57 | 1.55 | 2.48 | 2.33 | 1.10 | 1.22 |
| S2 | 0.43* | 0.33* | 1.45 | 1.47 | 2.20 | 2.20 | 1.11 | 1.27 |
| S3 | 0.40 | 0.32 | 1.62 | 1.46 | 2.34 | 2.18 | 1.21 | 1.15 |
| S4 | 0.48 | 0.44 | 1.86 | 1.34 | 2.95 | 2.31 | 1.46 | 1.02 |
*p < 0.05
Fig. 6.
Boxplot of sacral roots diffusivity parameters in patients with (n = 9) and without (n = 16) sacral and/or medullary anomaly. a FA, b MD in mm2/s × 103, c AD in mm2/s × 103, and d RD in mm2/s × 103. Results are reported for individual roots from L5 to S4. Individual boxplots report median, 25th and 75th percentiles. For each root, significative difference (p < 0.05) between (or with and without anomalies) diffusivity parameter for are indicated with a *
Discussion
Although the embryogenesis of anorectal malformations (ARM) is not clearly outlined, the frequent association of sacral and medullary anomalies suggests that they are part of a larger developmental anomaly involving the entire caudal part of the embryo [38]. We hypothesized that the organization of the pelvic peripheral nervous network could be altered in this context. We take advantage of a 3D modeling of the peripheral nerves, obtained using a novel AI method, to better characterize the lumbosacral plexus of patients with ARM, before any intervention. High forms of ARM had smaller root volumes and diffusion parameters suggesting poorer fiber microarchitecture than low forms of ARM. This observation was never done before and may reflect a difference in the development of the lumbosacral nervous plexus according to the different types of ARM.
Several methods for diffusion MRI reconstruction exist; among them, diffusion tensor imaging (DTI) [22] is the most popular one since it requires a low scanning time, but its results are highly dependent on the manual placement of regions of interest (ROI) for the starting and ending locations of the tracts [23]. Tiryaki et al. applied such DTI with ROI placement in patients older than 6 years already operated for ARM comparing their FA and apparent diffusion coefficient (ADC) to healthy controls of various ages [21]. They did not find any differences between the two groups which may be due either to the methodology of DTI signals processing (e.g., ROI placements, denoising, artifacts corrections) or to the choice of the groups.
In order to avoid the use of manual ROI placement and at the same time reduce the amount of false positives encountered in whole-body tractograms, we use an AI algorithm that exploits spatial relations between nerve bundles and segmented anatomical structures, in order to filter the tractography results and segment and recognize the nerves of interest [24–27]. Despite the small age of the population, with a mean age of four months, fiber tractography of the lumbosacral plexus was obtained for all 25 patients. The use of a 3 T MRI and anatomy-based algorithms achieved an optimized representation of the nerve roots compared with the tools available on MRI stations.
Due to the impossibility of obtaining values from controls of the same age for ethical reasons, the results were compared between two different types of ARMs, the high and low types, with distinct prognosis but the same median age (4 months). It is important to note that MRI exams were performed before any surgical intervention, which excludes any bias related to possible lesions induced by surgery and its complications.
It was found that the volume of the sacral roots S1, S2, and S3 was lower in patients with high ARMs when compared with patients with low type of ARMs. In order to know if this result was related to the patients with spinal and/or sacral anomalies, an analysis was performed in the subgroup of patients without sacral and/or spinal cord abnormalities, confirming that volume roots of S1 and S3 were smaller in high ARM when compared to patients with low types of ARM. Moreover, in the small group of nine patients with sacral and/or spinal cord anomalies (where 66% had high ARMs), the volume of the L5 root was lower when compared with patients (both high and low ARM) without these associated anomalies.
These results could help to explain the difference in functional outcomes inherent to the two types of malformation since high forms have a reduced development of the pelvic muscles and a more severe prognosis in terms of continence. In this area, the somatic and autonomic systems are indeed highly entangled, with the pelvic splanchnic nerves coming directly from the ventral branches of the second, third, and fourth sacral spinal nerves.
In addition, an overall significant alteration of the roots micro-architecture was observed for the most severe ARMs, reflected by a diminution of the FA and an increase of the AD and RD measures [1]. For the central nervous system, FA has been shown to be a marker of white matter integrity [39]. MD has been associated with axonal loss and RD with myelin quality. The increase in RD is considered to reflect the destruction or alteration of myelin, modifying the overall fibrous structure and facilitating diffusion along the axis perpendicular to the axon. Like for other neurodegenerative disorders [39], we observed a significant decrease in FA and an increase in MD measures. A single study has investigated lumbosacral plexus diffusion parameters in a pathological pediatric population comparing cases of spina bifida to healthy adult subjects; nevertheless, the study is difficult to interpret due to the difference in age of both populations [23, 40, 41].
We acknowledge that this study has some limitations, particularly the number of patients which precludes any more refined classification, for instance, considering the level of the fistula with the urinary tract. On a technical level, it is possible that the volume measurements have been affected by image artifacts from dMRI sequences, which may generate reconstruction inaccuracies. This may have been the case for the S3 and S4 sacral roots, which were difficult to reconstruct for some patients.
Insofar as this is a pilot study, we have remained cautious in concluding on potential clinical applications. Nevertheless, if a correlation could be established between the volume and structure of the lumbosacral plexus and the functional prognosis in MARs, this would make it possible to refine the classification of ARMs and, above all, to have an additional prognostic element for informing families and managing patients.
Conclusion
To the best of our knowledge, this study is the first to report and analyze the 3D original anatomy of the lumbosacral plexus, before any surgery, in children with ARMs thanks to the use of an original processing of MRI images. By combining the segmentation of anatomical structures from MRI imaging and the processing of diffusion MRI images via the spatial properties of the nerve fibers, we were able to reconstruct them. We found that, beyond the malformation of the anus and rectum, there are also differences in the volume and microarchitecture of the lumbosacral plexus between the two major types of ARMs, i.e., high and low. This observation supports the hypothesis that high ARMs result from a more regional developmental abnormality than low ARMs, potentially involving the entire caudal pole, and opens new ways to visualize and assess the lumbosacral plexus in children and adults.
Author Contribution
JG: conceptualization, data curation, methodology, writing-original draft and writing-review and editing. SS, IB, GLB, EB, PG, AD: conceptualization, data curation, methodology, writing- review and editing. PM, QP, EM, JBM: data curation, segmentation and sequences tuning, approval of final version. CL, LB, NB: data curation and interpretation, approval of final version. TB, SB, CC, MZ, COM: data curation, surgery, approval of final version.
Funding
J. Goulin received financial support for the research from University Hospital of Angers and Medical University of Angers. All others received no financial support for the research, authorship, and/or publication of this article.
Data Availability
Not applicable.
Declarations
Ethics Approval
Ethical approval for this study was obtained from “Comite de Protection des Personnes Ile de France III” (2015-AO1705-44).
Consent to Participate
Written informed consent for participation was obtained from legally authorized representatives before the study.
Consent for Publication
Not applicable.
Conflicts of Interest
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Nash O, Zobell S, Worst K, Rollins M D: Bowel management for the treatment of fecal incontinence and constipation in patients with anorectal malformations. Semin Pediatr Surg 29(6):150993, 2020. [DOI] [PubMed] [Google Scholar]
- 2.Guys JM, Hery G, Haddad M, Borrione C: Neurogenic bladder in children: basic principles, new therapeutic trends. Scand J Surg 100(4):256–63, 2011. [DOI] [PubMed] [Google Scholar]
- 3.Esposito G, Totonelli G, Morini F, Contini G, Palma P, Mosiello G, Longo D, Schingo PM, Marras CE, Bagolan P, Iacobelli BD: Predictive value of spinal bone anomalies for spinal cord abnormalities in patients with anorectal malformations. J Pediatr Surg 56(10):1803–1810, 2021. [DOI] [PubMed] [Google Scholar]
- 4.Kim SM, Chang HK, Lee MJ, Shim K W, Oh J T, Kim DS, Kim MJ, Han SJ: Spinal dysraphism with anorectal malformation: lumbosacral magnetic resonance imaging evaluation of 120 patients. J Pediatr Surg 45(4):769–776, 2010. [DOI] [PubMed] [Google Scholar]
- 5.Nah SA, Ong CC, Lakshmi NK, Yap TL, Jacobsen AS, Low Y: Anomalies associated with anorectal malformations according to the Krickenbeck anatomic classification. J Pediatr Surg. 47(12):2273–2278, 2012. [DOI] [PubMed] [Google Scholar]
- 6.Borg H, Holmdahl G, Olsson I, Wilklund LM, Sillén U: Impact of spinal cord malformation on bladder function in children with anorectal malformations. J Pediatr Surg 44(9):1778–85, 2009. [DOI] [PubMed] [Google Scholar]
- 7.Minneci PC, Kabre RS, Mak GZ, Halleran DR, Cooper JN, Afrazi A, Calkins CM, Corkum K, Downard CD, Ehrlich P, Fraser JD, Gadepalli SK, Helmrath MA, Kohler JE, Landisch R, Landman MP, Lee C, Leys CM, Lodwick DL, McLeod J, Mon R, McClure B, Rymeski B, Saito JM, Sato TT, St Peter SD, Wood R, Levitt MA, Deans KJ; Midwest Pediatric Surgery Consortium: Can fecal continence be predicted in patients born with anorectal malformations? J Pediatr Surg 54(6):1159–1163, 2019. [DOI] [PubMed] [Google Scholar]
- 8.Stafrace S, Lobo L, Augdal TA, Avni FE, Bruno C, Damasio MB, Darge K, Franchi-Abella S, Herrmann J, Ibe D, Kljiucevsek D, Mentzel HJ, Napolitano M, Ntoulia A, Ording-Muller LS, Perucca G, Petit P, Smets AM, Toso S, Wozniak MM, Riccabona M: Imaging of anorectal malformations: where are we now? Abdominal imaging task force of the European Society of Paediatric Radiology. Pediatr Radiol 52(9):1802–1809, 2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Madhusmita, Ghasi RG, Mittal MK, Bagga D: Anorectal malformations: Role of MRI in preoperative evaluation. Indian J Radiol Imaging 28(2):187–194, 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Máslová D, Holubová Z, Poš L, Newland N, Pavlikova M, Skaba R, Kyncl M: The comparison of magnetic resonance and fluoroscopic imaging options in the preoperative assessment of boys with anorectal malformations and a colostomy. Pediatr Radiol 54: 68–81, 2024. [DOI] [PubMed] [Google Scholar]
- 11.Zijta FM, Lakeman MM, Froeling M, van der Paardt MP, Borstlap CSV, Bipat S, Montauban van Swijndregt AD, Strijkers GJ, Roovers JP, Nederveen AJ, Stoker J: Evaluation of the female pelvic floor in pelvic organ prolapse using 3.0-Tesla diffusion tensor imaging and fibre tractography. Eur Radiol 22(12):2806–13, 2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Roy C, Ohana M, Labani A, Tricard T, Saussine C, Leyendecker P : Evaluation of the female bladder neck and urethra using MRI with fiber tractography: Prospective study on a large cohort of continent women. Neurourol Urodyn 40(6):1441–1449, 2021. [DOI] [PubMed] [Google Scholar]
- 13.Verde ASC, Santinha J, Carrasquinha E, Loucao N, Gaivao A, Fonseca J, Matos C, Papanikolaou N : Diffusion tensor-based fiber tracking of the male urethral sphincter complex in patients undergoing radical prostatectomy: a feasibility study. Insights Imaging 27;11(1):126, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Lemos N, Melo HJF, Sermer C, Fernandes G, Ribeiro GN, Luo ZC, Girao MJBC, Goldman SM : Lumbosacral plexus MR tractography: A novel diagnostic tool for extraspinal sciatica and pudendal neuralgia? Magn Reson Imaging 83:107–113, 2021. [DOI] [PubMed] [Google Scholar]
- 15.Ibrahim I, Škoch A, Herynek V, Jiru F, Tintera J: Magnetic resonance tractography of the lumbosacral plexus: Step-by-step. Medicine (Baltimore) 100(6):e24646, 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Zhan Y, Wang J, Guo WL: Comparative effectiveness of imaging modalities for preoperative assessment of anorectal malformation in the pediatric population. J Pediatr Surg. 54(12):2550–2553, 2019. [DOI] [PubMed] [Google Scholar]
- 17.Wong KK, Khong PL, Lin SC, Lam WWM, Lan LCL, Tam PKH: Post-operative magnetic resonance evaluation of children after laparoscopic anorectoplasty for imperforate anus. Int J Colorectal Dis 20:33–3, 2005. [DOI] [PubMed] [Google Scholar]
- 18.Smith CA, Nicassio LN, Rice-Townsend SE, Avansino JR, Hauptman J: Detection and Management of Tethered Cord in Anorectal Malformation: A Survey of Pediatric Neurosurgeons in the United States. Pediatr Neurosurg 57(6):443–440, 2022. [DOI] [PubMed] [Google Scholar]
- 19.Gangopadhyay AN, Pandey V, Gupta DK, Sharma SP, Kumar V, Verma A: Assessment and comparison of fecal continence in children following primary posterior sagittal anorectoplasty and abdominoperineal pull through for anorectal anomaly using clinical scoring and MRI. J Pediatr Surg 51(3):430–434, 2016. [DOI] [PubMed] [Google Scholar]
- 20.Onishi S, Kedoin C, Murakami M, Higa N, Yoshida A, Onitsuka K, Moriyama T, Yoshimoto K, Ieiri S: Image-guided confirmation of a precision pull-through procedure during laparoscopically assisted anorectoplasty in an open MRI operating theater: first application in an infantile case with anorectal malformation. Surg Case Rep 7(1):211, 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Tiryaki S, Celtik U, Eraslan C, Divarci E, Tekin A, Ozok G, Ergun O, Celik A, Calli C, Avanoglu A, Ulman I: Diffusion tensor imaging in anorectal malformations: a pilot study for the evaluation of lumbosacral plexus. Pediatr Surg Int. 38(4):609–616, 2022. [DOI] [PubMed] [Google Scholar]
- 22.Mori S, Crain BJ, Chacko VP, van Zijl PC: Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Annals of Neurology 45:265–269, 1999. [DOI] [PubMed] [Google Scholar]
- 23.Haakma W, Dik P, Ten Haken B, Froeling M, Nievelstein RAJ, Cuppen I, de Jong TPV, Leemans A: Diffusion tensor magnetic resonance imaging and fiber tractography of the sacral plexus in children with spina bifida. J. Urol 192(3):927–933, 2014. [DOI] [PubMed] [Google Scholar]
- 24.Muller CO, Delmonte A, Meignan P, Peyrot Q, Virzì A, Berteloot L, Grevent D, Blanc T, Gori P, Boddaert N, Bloch I, Sarnacki S: Nervous System Exploration Using Tractography to Enhance Pelvic Surgery. Surgetica 2019: Proceedings of Surgetica 2019 - Computer-Assisted Medical Interventions: scientific problems, tools and clinical applications: Rennes, 2019. p. 90-92
- 25.Muller CO, Mille E, Virzi A, Marret JP, Peyrot Q, Delmonte A, Berteloot L, Gori P, Blanc T, Grevent D, Boddaert N, Bloch I, Sarnacki S: Integrating tractography in pelvic surgery: a proof of concept. Journal of Pediatric Surgery Case Reports 48:101268, 2019. [Google Scholar]
- 26.Muller CO: Nervous system exploration using tractography to enhance pelvic surgery in children with tumors and malformations. Neuroscience. Université Paris Cité, 2020.
- 27.Vinit N, Blanc T, Bloch I, Pio L, Kassir R, La Barbera G, Bonnot E, Gori P, Goulin J, Pire A, Boddaert, Lozach C, Sarnacki S: Robotics and 3D modeling for precision surgery in pediatric. EJC Paediatric Oncology 4:100181, 2024. [Google Scholar]
- 28.Stephens FD, Smith ED and Paoul NW: Anorectal Malformations in Children, March of Dimes Birth Defect Foundation Original Series 24(4), New York: Alan R. Liss Inc., 1988. [Google Scholar]
- 29.Holschneider A, Hutson J, Peña A, Beket E, Chatterjee S, Coran A, Davies M, Georgeson K, Grosfeld J, Gupta D, Iwai N, Kluth D, Martucciello G, Moore S, Rintala R, Smith E D, Sripathi D V, Stephens D, Sen S, Ure B, Grasshoff S, Boemers T, Murphy F, Söylet Y, Dübbers M, Kunst M: Preliminary report on the international conference for the development of standards for the treatment of anorectal malformations. J Pediatr Surg 40:1521–6, 2005. [DOI] [PubMed] [Google Scholar]
- 30.Muller CO, Crétolle C, Blanc T, Alova I, Jais JP, Lortat-Jacob S, Aigrain Y, Zérah M, Sarnacki S: Impact of spinal dysraphism on urinary and faecal prognosis in 25 cases of cloacal malformation. J Pediatr Urol. 10(6):1199–205, 2014. [DOI] [PubMed] [Google Scholar]
- 31.Cordero-Grande L, Christiaens D, Hutter J, Price AN, Hajnal JV: Complex diffusion-weighted image estimation via matrix recovery under general noise models. NeuroImage 200:391–404, 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Tournier JD, Smith RE, Raffelt D, Tabbara R, Dhollander T, Pietsch M, Christiaens D, Jeurissen B, Yeh CH, Connelly A: MRtrix3: A fast, flexible and open software framework for medical image processing and visualisation. NeuroImage 202:116–37, 2019. [DOI] [PubMed] [Google Scholar]
- 33.Bloch I: Fuzzy Spatial Relationships for Image Processing and Interpretation: A Review. Image and Vision Computing 23(2):89–110, 2005. [Google Scholar]
- 34.Delmonte A, Mercier C, Pallud J, Bloch I, Gori P: White matter multi-resolution segmentation using fuzzy set theory. IEEE International Symposium on Biomedical Imaging (ISBI). IEEE, 2019. p. 459-462.
- 35.Alexander AL, Lee JE, Lazar M, Field AS: Diffusion tensor imaging of the brain. Neurotherapeutics 4(3):316–29, 2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Van der Jagt PK, Dik P, Froeling M, Kwee TC, Nievelstein RAJ, ten Haken B, Leemans A: Architectural configuration and microstructural properties of the sacral plexus: a diffusion tensor MRI and fiber tractography study. Neuroimage 62(3):1792–1799, 2012. [DOI] [PubMed] [Google Scholar]
- 37.Wilcoxon F: Individual Comparisons by Ranking Methods. Biometrics Bulletin 1(6): 80–83,1945. [Google Scholar]
- 38.Bai Y, Chen H, Yuan ZW, Wang W: Normal and abnormal embryonic development of the anorectum in rats. J Pediatr Surg 39(4):587–90, 2004. [DOI] [PubMed] [Google Scholar]
- 39.Andica C, Kamagata K, Hatano T, Saito Y, Ogaki K, Hattori N, Aoki S : MR Biomarkers of Degenerative Brain Disorders Derived From Diffusion Imaging. J Magn Reson Imaging 52(6):1620–1636, 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Chen DQ, Strauss I, Hayes DJ, Davis KD, Hodaie M: Age-related changes in diffusion tensor imaging metrics of fornix subregions in healthy humans. Stereotact Funct Neurosurg 93(3):151–159, 2015. [DOI] [PubMed] [Google Scholar]
- 41.Kronlage M, Schwehr V, Schwarz D, Godel T, Uhlmann L, Heiland S, Bendszus M, Baumer P: Peripheral nerve diffusion tensor imaging (DTI): normal values and demographic determinants in a cohort of 60 healthy individuals. Eur Radiol. 28(5):1801–1808, 2018. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Not applicable.