To explore the pathogenesis of Bell's palsy using diffusion tensor image

To explore the pathogenesis of Bell's palsy using the diffusion tensor image on 3.0 T MR. The healthy people and the patients with Bell's palsy underwent intraparotid facial nerve scanning by using the DTI and T1 structural sequence at 3.0 T MR. The raw DTI data were performed affine transformation and nonlinear registration in the common MNI152_T1 space and resampled to the 0.4 mm3 voxel size. A group of 4 spherical seed regions were placed on the intratemporal facial nerves in the common space, bilaterally and symmetrically. The DTI data in the common space were used to track the intratemporal facial nerve fibers by using TrackVis and its Diffusion Toolkit. Each tractography was used to construct the maximum probability map (MPM) according to the majority rule. The fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD) and radial diffusivity (RD) were calculated and extracted on the basis of MPM. For healthy people, there was no significant difference in FA, MD, RD and AD of bilateral facial nerves. For patients with Bell's palsy, there was no significant difference in AD, there was significant difference in FA, MD and RD between the affected nerve and the healthy nerve (P < 0.02). This study showed that the myelin sheath injury of the intratemporal facial nerve is the main cause of Bell's palsy. Most neural axons are not damaged. The results may explain the pathogenesis of the Bell's palsy, which is self-limited for most cases.

adequately addressed.The study was approved by the IRB of The First Teaching Hospital of Tianjin University (reference number: TYLL2022[Z]No 010).
All volunteers included 18 healthy people and 19 patients.The 18 healthy people with normal facial nerves, including 17 right-handed and 1 left-handed (10 females and 8 males; mean age: 40.9 ± 6.6 years; range: 23-52 years).The 19 patients with unilateral acute-onset peripheral facial paresis/paralysis excluding central nervous system diseases (all right-handed; 13 males and 6 females; mean age: 45.4 ± 15.3 years; range: 23-63 years).There were 13 cases of right facial paresis/paralysis and 6 cases of left facial paresis/paralysis.The onset time of 19 patients is from 1 day to 1 month, and 1 case is 2 years.The house House-Brackmann (H-B) facial nerve grading scale facial nerve function was classified from grade III to VI, which is moderate to severe.The main symptoms were loss of frontal strip on one side and incomplete closure of ipsilateral eyelids.10 patients had a history of periauricular pain or pharyngeal pain before and after the onset, and 9 patients did not find any discomfort before the onset.

Data post-processing T1WI Data post-processing.
To construct group-level common template space of 0.4 mm 3 voxel size, the 0.5 mm 3 voxel of MNI152_T1_0.5 mm brain template (International Consortium for Brain Mapping) was normalized and resampled to 0.4 mm 3 template using the "old norm" method of SPM12, which named MNI152_ T1_0.4 mm template in the present study (Fig. 1).By using FSL-maths, a group of 4 common spherical seed regions (5 mm in diameter) was calculated and placed on the mastoid segment of intratemporal facial nerves, bilaterally and symmetrically (Fig. 2).DTI Data post-processing.After eddy, current correction of the DTI data, affine transformation, and nonlinear match ing were performed on the diffusion images matching with the MNI152_T1_0.4mm template without skull-stripping by using the FSL (Oxford, UK).Then the voxel size of the DTI was resampled to 0.4 mm 3 in the common space of the MNI152_T1_0.4mm template.DTI data were used to track each intratemporal facial nerve fiber in 0.4 mm 3 common space by using TrackVis and its Diffusion Toolkit (DTK).The intratemporal facial nerve fibers were traced with a thresh old value of FA > 0.15 and an angle threshold value of 60°.To reduce errors, the step length of fibers less than 2 mm was discarded (Fig. 3).
All results then were used to construct the maximum probability map (MPM), which assigned the most probable labels to each voxel according to the majority rule across subjects (Fig. 4).The indexes of DTI, named fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD), were calculated and extracted based on MPM.
In the present study, the MPM of 18 healthy people was named 18-Health-MPM, the MPM of 19 patients people was named 19-Patient-MPM, and the MPM of all 18 healthy people and 19 patients were named 37-All-MPM, respectively.Statistical analysis.SPSS18 (Chicago, Illinois, USA) was used for statistical analysis.To conduct a comprehensive analysis of facial nerve, this study collected DTI data of the left/right and the affected/healthy side facial nerve and then conducted descriptive statistics and normal tests on the data.For quantitative measurement, paired and unpaired t-test was used to evaluate the FA, MD, AD, and RD values.All statistical significance tests were two-tailed hypotheses, with P < 0.05 as statistically significant.

Results
All DTI indexes of left/right and affected/healthy side intratemporal facial nerve accorded with a normal distribution (K-S test, P > 0.05).

Base on 18-Health-MPM.
There was no significant difference in FA, MD, RD, and AD of the left and right intratemporal facial nerve (Table 1).

Base on 19-Patient-MPM.
There was a significant difference in FA, MD, and RD between the affected nerve and the healthy nerve (P < 0.02); there was no significant difference in AD.The FA of the affected nerve was lower than that of the healthy nerve, and the MD and RD were higher than that of the healthy nerve (Table 2).
Base on 37-All-MPM.In 18 healthy people, there was no significant difference in all DTI indexes between the left and right intratemporal facial nerves (Table 1).In 19 patients, there were significant differences in FA, MD, and RD between the affected nerve and the healthy nerve; there was no significant difference in AD (Table 2).Compared with the 19 patients and the 18 healthy people: there was no significant difference in all DTI indexes between the healthy nerves of patients (19 nerves) and the nerve of healthy people (36 nerves)

Discussion
The intratemporal facial nerve courses through the bony facial nerve canal in the temporal bone.The labyrinthine segment, which has been measured on average below 0.7 mm, is the narrowest portion of the canal.That the bony canal confines the expansion of the facial nerve caused by edema or inflammation has been speculated as the etiology of Bell's palsy, especially in the labyrinthine segment 13 .However, there is no direct evidence for the pathogenesis of Bell's palsy.www.nature.com/scientificreports/DTI can non-invasively describe the course of nerve fibers by the degree of restriction of free movement of water molecules.Specifically, the integrity of the axon and the myelin sheath of nerve fibers limits the free diffusion of water molecules.The higher the FA the higher integrity of nerve fibers, nevertheless, the damage of the myelin sheath of nerve fibers will lead to the decrease of FA.MD indicates the ability of water molecules to diffuse in all directions of motion.The nerve damage will lead to faster diffusion of water molecules and higher MD 14 .AD and RD can accurately determine the specific cause of diffusion differences.It is generally believed that the increase in AD represents axonal injury of nerve fibers, and the increase in RD indicates the myelin damage of nerve fibers 15 .
The RESOLVE sequence is a sort of multi-shot echo planar imaging (ms-EPI) technique based on the readout segmentation with a navigator echo correction approach.The RESOLVE technique can provide high signalto-noise ratio images and significantly reduce susceptibility-based distortions compared with the conventional In this study, the raw RESOLVE data were resampled to 0.4 mm 3 voxel size by performing affine transformation and nonlinear registration in the common MNI152_T1_0.4mm space (Fig. 1).The resampled RESOLVE data were used to track the intratemporal facial nerve fibers of each subject in the 0.4 mm 3 group-level common space, and the MPMs of the different group were obtained (Fig. 4).To observe the microscopic differences of intratemporal facial nerve fibers, the DTI indexes based on the different MPMs nerve were extracted.
The results showed that there was no significant difference in DTI indexes between the bilateral nerves of healthy people.There was a significant difference in FA, MD, and RD between the affected nerve and healthy nerve; there was no significant difference in AD.The decrease in FA and the increase in MD suggest that the integrity of the affected facial nerve is damaged.There was no significant difference in AD between affected and healthy facial nerves, suggesting that the axon of the facial nerve was not damaged.The increase in RD suggests that the cause of facial nerve damage is myelin injury.

Conclusion
This study showed that the myelin sheath injury of the intratemporal facial nerve is the main cause of Bell's palsy.As far as we know, the myelin sheath of the nerve is relatively easy to recover while the axon of the nerve is difficult to regenerate.This study may explain why most of Bell's palsy is self-limited.Table 3. Base on 37-All-MPM, DTI index analysis of the healthy/affected facial nerves of 19 patients and the healthy nerves (mean ± SD).MD-the unit is 10 −3 mm 2 /s.N1 refers to the number of nerves on the healthy side of the patient.N2 refers to the number of nerves on the affected side of the patient.N3 refers to the number of nerves on both sides of healthy people.N4 refers to the number of nerves on both sides of healthy people plus the number of nerves on the healthy side of patients.

Figure 1 .
Figure 1.A patient resource FA map with a voxel size of 1.2 mm × 1.2 mm × 3.0 mm was consistent in the common space of MNI152_T1_0.4mm template with a voxel size of 0.4 mm 3 by nonlinear registration.

Figure 2 .
Figure 2. The symmetrically spherical ROIs (right: red and blue; left: yellow and green) with the same diameter of 5 mm, were set on intratemporal facial nerves (right: hollow arrow; left: solid arrow) in the 0.4 mm 3 voxel MNI common space (named MNI152_T1_0.4mm).

Figure 3 .
Figure 3. DTI fiber tracing of bilateral intratemporal facial nerves was performed.A patient (A, B) is a female, 46 years old, with right facial paralysis for 3 days and H-B facial nerve function grade IV.A patient (C, D) is a male, 23 years old, with right facial paralysis for 7 days and H-B facial nerve function grade III.A patient (E, F) is a male, 59 years old, with left facial paralysis for 1 day and H-B facial nerve function grade III.The left (solid arrow) and right (hollow arrow) intratemporal facial nerves were indicated by the arrows respectively.

Figure 4 .
Figure 4.According to the majority rule across subjects, the MPM of bilateral intratemporal facial nerves was constructed.The MPM of the healthy people (a, b), the patients (c, d), all people including the healthy people and patients (e, f).The left (solid arrow) and right (hollow arrow) intratemporal facial nerves were indicated by the arrows respectively.

Table 1 .
DTI indexes of bilateral facial nerve in healthy people (mean ± SD).MD-the unit is 10 −3 mm 2 /s.

Table 2 .
DTI index analysis of affected facial nerve and the healthy nerve in patients (mean ± SD).MD-the unit is 10 −3 mm 2 /s.