Altered microstructural pattern of the cortex and basal forebrain cholinergic system in wilson’s disease: an automated fiber quantification tractography study

Basal forebrain (BF) cholinergic projection neurons form a highly extensive input to the cortex. Failure of BF cholinergic circuits is responsible for the cognitive impairment associated with Wilson’s disease (WD), but whether and how the microstructural changes in fiber projections between the BF and cerebral cortex influence prospective memory (PM) remain poorly understood. We collected diffusion tensor imaging (DTI) data from 21 neurological WD individuals and 26 healthy controls (HCs). The experiment reconstructed the probabilistic streamlined tractography of 18 white matter tracts using an automated fiber quantification (AFQ) toolkit. Tract properties (FA, MD, RD, and AD) were computed for 100 points along each tract for each participant, and the differences between the groups were examined. Subsequently, correlation analysis was performed to evaluate whether abnormal microstructural white matter integrity measures correlate with PM performance. Additional investigations used a tract-based spatial statistics (TBSS) approach to identify regions with altered white matter structure between groups and verify the reliability of the AFQ results. The highest nonoverlapping DTI-related differences were detected in the anterior thalamic radiation (ATR), corticospinal tract (CST), corpus callosum, association fibers, and limbic system fibers. Additionally, PM parameters of the patient group were highly correlated with white matter microstructure changes in the inferior longitudinal fasciculus. Our study highlights that the performance of projections between cholinergic input and output areas—the cerebral cortex and BF—may serve as neural biomarkers of PM and disease prognosis.


Introduction
Wilson's disease (WD) is a treatable monogenic, autosomal recessive hereditary disorder caused by a defect in copper metabolism that primarily affects the liver and central nervous system (Cleymaet et al., 2019). The range of clinical manifestations is wide, but the most significant and basic symptom of the disease is cognitive impairment. Memory is one of the most important aspects of cognitive function. Prospective memory (PM) is the core memory component in daily life, which refers to forming memory and the ability to stimulate memory and execute intention when cues occur. It can be divided into time-based PM (TBPM), which uses a specific time as a cue, and event-based PM (EBPM), which uses a specific external event stimulus as a cue (Graa & Ergis, 2021;Henry, 2021). Over the past few years, there Yutong Wu and Sheng Hu contributed equally to the writing of this article. Sheng Hu hushengustc@163.com Hongxing Kan 984377701@qq.com 1 has been increasing interest in the study of PM in WD, as WD with PM impairment will develop into a larger extent of cognitive impairment. It has been revealed that PM damage in WD patients is associated with volume reductions in the basal ganglia and abnormalities in white matter fibers (Dong et al., 2016;Dong et al., 2019;Hu et al., 2022).
The basal forebrain (BF) is a waystation for many ascending and descending pathways. Its cholinergic neurons play a role in eliciting cortical activation and arousal and are involved in memory and attention (Blake & Boccia, 2018). Previous studies have reported that memory impairment is due to significant degeneration of BF neurons and loss of cortical cholinergic innervation (Chiesa et al., 2019;Daulatzai, 2016;Mesulam, 2004). Furthermore, Gargouri et al. (Gargouri et al., 2019) created tracks between the BF and cortical masks using the bedpostx and probtrackx tools and found evidence of connectivity deficits between cortical regions and the BF in PD patients. However, very little is known about whether and how BF-cortical connectivity has an influence on prospective memory (PM) in WD and, in particular, whether white matter dysfunction is localized at specific regions of the fiber tracts.
Diffusion tensor imaging (DTI) is the most reliable noninvasive neuroimaging technique for identifying and quantifying white matter microstructural damage in WD patients (Dong et al., 2019;Hu et al., 2021;Lawrence et al., 2016;Zhou et al., 2018). Fractional anisotropy (FA), axial diffusivity (AD), radial diffusivity (RD) and mean diffusivity (MD) are four prominent diffusion indices used to ascertain the degree of WD pathology. A cross-sectional study using the tract-based spatial statistics (TBSS) approach reported significantly higher FA in the thalamus and lentiform nucleus brain stem (Dong et al., 2019). The results from ROI-based methods showed significantly lower FA values between subcortical nuclei in WD patients (Zhou et al., 2016). These studies did not yield consistent results, which may be caused by specific shortcomings of these methods, such as the difficulty in localization of ROI-based methods and the inability of voxel-based methods to completely overcome the problem of cross-subject coregistration.
Furthermore, previous studies have found that the fibers are exposed to different biological environments during the process of stretching, and tissue characteristics may vary considerably along the tract (Johnson et al., 2014;Keller et al., 2017), including variations in myelination, axonal density, and axonal diameter as well as the influence of tract curvature, partial volume effects and the entrance and exit of smaller axonal bundles from the larger tract. Some pathological alterations may occur in circumscribed regions of the tract rather than along the entire tract. Hence, calculating the average diffusion properties of the entire tract in conventional DTI analysis may obscure potentially important regional information. Automated fiber quantification (AFQ) is a fully automated DTI fiber tract tracking technology that can automatically extract the FA, MD, AD, and RD values at different nodes of 20 major white matter fiber tracts in an individual's brain (Yeatman et al., 2012). Recently, AFQ has been successfully applied to research in clinical and neuroimaging studies (Dou et al., 2020;Keller et al., 2017;Wang et al., 2021). Therefore, AFQ may provide a promising strategy to investigate changes in white matter microstructural integrity at specific locations in the BF-cortical network in WD patients.
In the current study, we aimed to evaluate the connectivity between cholinergic input and output areas, the cerebral cortex and BF, in WD patients by using AFQ tractography and to determine whether the diffusion properties along the tracts can serve as a biomarker for PM performance in WD patients. As an exploratory analysis, we used the TBSS approach to characterize white matter integrity across the whole brain and verify the reliability of the AFQ results.

Subjects and assessments
The total sample of 47 subjects was composed of twentyone right-handed patients with neurological WD and twenty-six right-handed healthy controls (HCs) matched for sex, age, and education level. All subjects were recruited from the First Affiliated Hospital of Anhui University of Chinese Medicine (AUCM). The WD group included 10 women (47.6%) with a mean age of 22.39 years (SD = 6.35 years). WD patients received drug treatment, including penicillamine and zinc salts. After being evaluated by an expert neurologist and a trained neuropsychologist with comprehensive clinical interviews, patients diagnosed with WD based on the clinical manifestations (extrapyramidal symptoms, pyramidal symptoms, and behavioral problems), neuroimaging findings, low serum total copper and ceruloplasmin levels, elevated urinary copper excretion, liver biopsy, and a Kayser-Fleischer ring were included. Neuropsychiatric symptoms caused by something other than WD precluded participation. The HC group included 11 women (42.3%) with a mean age of 22.78 years (SD = 7.34 years). The criteria for HC subjects were no history of neurological or mental disease, no drug abuse history, no history of psychiatric or mental illness, and no cognitive impairment caused by other diseases or drugs. Informed consent was obtained from each participant prior to enrollment. The work described has been carried out in accordance with the Code of Ethics of the World Medical Association (Declaration of Helsinki) and received ethical approval from the Human Research Committee of the First Affiliated Hospital of AUCM.
In our study, all patients with neurological WD underwent standardized neuropsychological assessments performed by an experienced neuropsychologist, including (1) the Mini-Mental State Examination (MMSE) for evaluating global cognitive functions and (2) EBPM and TBPM for quantitative assessment of memory. These tests were performed as previously described (Esposito et al., 2015;Folstein et al., 1975;Gonneaud et al., 2014;Loprinzi et al., 2018). Full details are provided in the Supplementary Information. Details of the demographic and clinical characteristics of the included subjects are reported in Table 1.

MRI data acquisition
Whole-brain diffusion tensor imaging and high-resolution T1-weighted images were obtained in the magnetic resonance room of the First Affiliated Hospital of AUCM by using a 3.0 T MR system (Discovery MR750, GE Healthcare; Chicago, IL) with an eight-channel high-resolution radio-frequency head coil. During scanning, all participants were required to remain stationary and awake with their eyes closed. Sagittal T1WI high-resolution images were collected from all individuals using a T1-3D BRAVO sequence (repetition time = 8.16 ms, echo time = 3.18 ms, flip angle = 12°, matrix = 256 mm × 256 mm, field of view = 256 mm × 256 mm, slice thickness = 1 mm without gaps, 200 axial slices). Diffusion tensor images were also acquired from all individuals using an echo-planar imaging (EPI) sequence (repetition time = 6000 ms, echo time = 81.7 ms, matrix size = 128 mm × 128 mm, field of view = 256 mm × 256 mm, slice thickness = 3 mm, diffusion sensitivity coefficient b = 0 s/mm 2 and 2,000 s/mm 2 , 64 direction).
All MRI images were visually inspected by two experienced neuroradiologists for data quality control. In the event of disagreement, consensus was arrived at by discussion. Finally, we discarded two WD patients according to the criteria that head motion was restricted to less than 2 mm in the x, y, or z direction or 2 degrees of rotation in each axis.

Image preprocessing
Diffusion images were converted from DICOM to NIFTI format (with the extraction of diffusion gradient directions) using the dcm2nii tool in the MRIcron software toolbox (https://www.nitrc.org/projects/mricron). DTI data were then preprocessed using the FMRIB Software Library tools (version 5.09, https://www.fmrib.ox.ac.uk/fsl). First, the eddy current and motion distortion correction were conducted using FMRIB's Diffusion Toolbox (FDT, http:// www.fmrib.ox.ac.uk/fsl/fdt). The nondiffusion-weighted images, that is, the first volume image used as a reference in DTI data, were skull stripped using the brain extraction tool (BET) and used to mask all diffusion-weighted images. The registered images were then automatically aligned to T1-weighted images using FMRIB's Linear Image Registration Tool (FLIRT). After coregistration, an average b0 dataset was created. Finally, a diffusion-tensor model was fitted at each voxel using the DTIFIT tool, generating diffusion measure (FA, MD, AD, and RD) maps and an S0 map.

AFQ analysis
Several anatomical landmarks, including the anterior commissure (AC), the posterior commissure (PC), and the midsagittal plane, were identified by hand in the T1 images. With these landmarks, all anatomical images were aligned to the AC-PC plane using the script mrAnatAverageAcpc-Nifti. The script dtiMakeDt6FromFSL was used for aligning the T1 image -the anatomical reference -to the S0 image, obtaining a dt6 MATLAB format file for further analysis. The full diffusion MRI dataset was imported into the AFQ software package (version 1.2, https://github.com/yeatmanlab/AFQ) running on MATLAB R2012a (The MathWorks, Natick, MA, USA). The procedure returned tensor-based measures for 100 equidistant segments along 20 tracts with

Correlation analysis
For the results of the AFQ method, WM tracts demonstrating significant diffusion property differences between the groups were subjected to Pearson's correlation analysis to investigate the relationship between mean properties on significantly different tracts and the PM scores while controlling for age, sex, and years of education (FDR correction, p < 0.05).
For the results of the TBSS method, Pearson's correlation analyses between DTI parameters and PM scores were performed in a voxelwise manner within a mask of areas with group differences while controlling for age, sex, and years of education (FDR correction, p < 0.05).
To evaluate whether head motion influenced the results, we also performed correlation analysis between framewise displacement (FD) and neuropsychological symptoms. In addition, to avoid the effect of potential brain atrophy of the BF on the results, we first defined the BF by using the Anatomy toolbox r 3.0 toolbox (https://github.com/inm7/ jubrain-anatomy-toolbox) and adopted voxel-based morphometry (VBM) analysis to evaluate the differences in BF gray volume between groups ( Supplementary Fig. 3) and then performed the correlation analysis between the volumes of the BF and PM scores.

Demographics
The demographic and clinical data are summarized in Table 1. There were no significant differences between the WD and HC groups in sex, age, education level, and handedness (P > 0.05, Table 1). All patients presented with a KF ring, and the duration of disease and MMSE scores were 5.45 ± 3.14 years and 26.42 ± 0.90, respectively.

Pointwise differences in tract profiles
Between-group differences at the pointwise level were determined by diffusion metrics (FA, MD, AD and RD) with the AFQ. Multiple comparisons were corrected across all 100 points on each tract using the false discovery rate at p < 0.05 to reduce type I error. For the majority of tracts, diffusion properties varied significantly along the length of the tract.
FA: Significantly altered locations of fiber tracts were as follows: (1) (Yeatman et al., 2012;Yeatman et al., 2018). In particular, AFQ is composed of a three-step procedure: (1) whole-brain fiber tractography, (2) waypoint ROI-based fiber tract segmentation (Wakana et al., 2007), and (3) cleaning and refinement of fiber tracts based on a probabilistic fiber tract atlas (Hua et al., 2008). More details of the process are presented in the Supplementary Information.
In our study, the selection of fiber bundles was based on the BF-cortical fiber tract map defined by Gargouri et al. (Gargouri et al., 2019), including the bilateral anterior thalamic radiation (ATR), bilateral corticospinal tract (CST), bilateral cingulum cingulate (CG), callosum forceps major, callosum forceps minor, bilateral cingulum hippocampus, bilateral inferior fronto-occipital fasciculus (IFOF), bilateral inferior longitudinal fasciculus (ILF), bilateral superior longitudinal fasciculus (SLF), bilateral uncinate fasciculus (UF), and bilateral arcuate fasciculus (AF). Tract properties for the cingulum could not be computed for three participants as a result of crossing fibers, artifacts, or head motion. These participants were excluded from analyses involving these particular fiber tracts. Fiber tracts that were missing in more than 10% of participants were also removed. These tracts included the bilateral cingulum hippocampus, which refers to the mesial temporal portion of the cingulum, also commonly referred to as the parahippocampal white matter bundle. These two fiber tracts were excluded from further analysis. All subsequent studies were based on 18 major WM tracts of the whole brain.
To compare the AFQ-identified diffusion properties between the WD and HC groups, mean values and standard deviations (SD) were plotted, and only ≥ 3 adjacent nodes were reported. The tract profiles extracted along the entire WM tract point-by-point can be compared to evaluate local alterations associated with WD. We performed group-level pointwise analyses of the diffusion indices at 100 points of each fiber tract (Yeatman et al., 2018). A one-way analysis of variance (ANOVA) was performed on each pointwise diffusion indicator for each fiber tract to determine betweengroup differences (p < 0.05, false discovery rate (FDR) corrected for all 100 points). In addition, to verify significant differences between each pair of groups, we also performed a post hoc analysis using a two-sample t test after controlling for age and sex (p < 0.05, FDR corrected).

TBSS analysis
Groupwise voxel-based statistical analysis of FA was performed using TBSS (Smith et al., 2006). A detailed description of the TBSS analysis is provided in the Supplementary Information.

TBSS group differences
The voxelwise TBSS analysis results exhibiting MD, RD, and AD differences between the WD and HC groups are shown in Supplementary Fig. 1. Significantly increased MD, RD, and AD were observed throughout almost the whole WM skeleton, including the CG, ILF, SLF and UF, which are considered to be associated with cognitive function, in the WD group compared with the HC group. There were no WM regions that showed FA changes in the WD patient group.
MD: Significant alterations in the pointwise comparison were mainly located in the following tracts: (1) (9) the temporal lobe portion of the right UF (nodes 3-20) (Fig. 3).
RD: Significant alterations in the pointwise comparison were mainly located in the following tracts: (1) the anterior Fig. 1 Plots of the FA profiles of identified fiber tracts in healthy control and patient groups (orange: HC, blue: WD). Each plot shows the mean tract FA profile ± 1 standard error of the mean confidence interval for each group. The light yellow bars indicate regions of significant differences between HC and WD patients. The x-axis represents the location between the beginning and termination waypoint regions of interest. Abbreviations: ATR, anterior thalamic radiation; HC, healthy control; WD, Wilson's disease; FA, fractional anisotropy parameters and clinical scores was found in the WD group. In addition, no correlation was found between head motion ( Supplementary Fig. 2) or volume atrophy of the BF (Supplementary Fig. 4) and PM performance. p = 0.029) and AD (r=-0.545, p = 0.015) values of the left ILF were negatively correlated with the disease duration. The corresponding scatter plot is shown in Fig. 5.
The TBSS comparisons are shown in Supplementary Tables 1, and no significant correlation between DTI Fig. 2 Plots of the MD profiles of identified fiber tracts in healthy control and patient groups (orange: HC, blue: WD). Each plot shows the mean tract MD profile ± 1 standard error of the mean confidence interval for each group. The light yellow bars indicate regions of significant differences between HC and WD patients. The x-axis represents the location between the beginning and termination waypoint regions of interest. Abbreviations: ATR, anterior thalamic radiation; ILF, inferior longitudinal fasciculus; SLF, superior longitudinal fasciculus; CST, corticospinal tract; HC, healthy control; WD, Wilson's disease; MD, mean diffusivity Fig. 3 Plots of the AD profiles of identified fiber tracts in healthy control and patient groups (orange: HC, blue: WD). Each plot shows the mean tract AD profile ± 1 standard error of the mean confidence interval for each group. The light yellow bars indicate regions of significant differences between HC and WD patients. The x-axis represents the location between the beginning and termination waypoint regions of interest. Abbreviations: ATR, anterior thalamic radiation; ILF, inferior longitudinal fasciculus; SLF, superior longitudinal fasciculus; CST, corticospinal tract; HC, healthy control; WD, Wilson's disease; AD, axial diffusivity Fig. 4 Plots of the RD profiles of identified fiber tracts in healthy control and patient groups (orange: HC, blue: WD). Each plot shows the mean tract RD profile ± 1 standard error of the mean confidence interval for each group. The light yellow bars indicate regions of significant differences between HC and WD patients. The x-axis represents the location between the beginning and termination waypoint regions of interest. Abbreviations: ATR, anterior thalamic radiation; IFOF, inferior fronto-occipital fasciculus; CST, corticospinal tract; HC, healthy control; WD, Wilson's disease; RD, radial diffusivity Drachman & Leavitt, 1974). White matter deficits associated with impairment of cholinergic transmission may affect memory function during tasks in WD. To provide a more detailed anatomical picture of WD progression, we identified the localized significance of BF-cortical network WM structures using a recently established method called AFQ in WD patients. Notably, several findings emerged from this study: (1) widespread disruption was distributed in specific locations of different tracts in WD patients, which is also consistent with the TBSS results in this paper, and (2) AFQ analysis confirmed that the cingulum cingulate and ILF may

Discussion
The basal forebrain cholinergic system spans multiple brain regions and is the major source of acetylcholine for the cerebral cortex in humans (Blake & Boccia, 2018;Gargouri et al., 2019). The lesions primarily involve the white matter tracts in the occipital, temporal, frontal, and parietal lobes. Many have tested the hypothesis that failures of the cholinergic circuitry of the basal forebrain are responsible for the cognitive impairments associated with neurodegenerative disorders (Ballinger et al., 2016;Bartus et al., 1982;  AD, axial diffusivity; RD, radial diffusivity; MMSE, Mini-Mental State Examination; TBPM, time-based expected memory; p, p value; r, Pearson's correlation coefficient any significant correlations between PM and diffusion properties that were averaged over the entire length of these fiber tracts. One possible theory for these findings is that microstructural changes in these fibers may indirectly impact PM. Another possible explanation is that the AFQ technique can provide more information about white matter damage that is not obvious from the entire fiber bundles. We believe that a portion of the variance in diffusion properties that we observed along the length of these fascicles is affected by the biological properties of the axons in a voxel as well as the regional interacting tissue characteristics. Furthermore, altered microstructural integrity (as measured by AD values) within the ILF correlated significantly with cognitive abilities measured by the MMSE and TBPM in the WD groups, which is consistent with previous data from Hu et al. (Hu et al., 2021). The ILF connects the occipital cortex with the anterior temporal lobe and amygdala, and the functional properties of these cortical areas confer on the ILF a role in memory (Tusa & Ungerleider, 1985). In Alzheimer's disease (Luo et al., 2020) and semantic dementia (Powers et al., 2013), an association between disruption of ILF integrity and cholinergic-innervated cognitive impairment has been reported. Our result is an intriguing finding, confirming that sudden disruption of the ILF by neurologic insult may constitute memory disorders. Notably, although some of the functions mediated by the ILF seem to be relatively lateralized, disruption of the tract's middle and posterior parts may be dynamically compensated for by the contralateral portion. The cingulum cingulate, as the core structure in the Papez circuit of the cholinergic system, is an important pathway to maintain communication within the limbic system, which plays a role in attention, emotion, and memory (Bubb et al., 2018;Bueno et al., 2018;Maldonado et al., 2020). The loss of cholinergic neurons in the basal forebrain leading to cingulate bundle axon and myelination degeneration, which causes memory impairment in people with neurodegenerative diseases, was found in previous studies. However, we did not find a significant correlation between PM and altered microstructural integrity in the cingulum cingulate. One possible theory for these findings is that the white matter structure of the brain may be affected by drug use and that the drugs are effective in improving cognitive functions in WD patients.

Limitations
The present study has several limitations. First, our sample size was relatively small, especially in terms of the WD group. Although the findings were significant, the small sample size in this study may not have sufficient statistical power. One of our future research plans is to expand the be neuroimaging markers of monitoring disease progression and memory status.
White matter abnormalities are common in WD patients and typically manifest asymmetrically, involving primarily the frontal lobe but also the temporal lobe and, to a lesser degree, the parietal and occipital lobes (Poujois et al., 2017;Salari et al., 2018;Sinha et al., 2006). Our TBSS and AFQ results were consistent with these previous studies. In particular, we found degeneration of the fibers of the ATR, CST, and callosum; association fibers (including IFOF, SLF, UF, and ILF); and limbic system fibers (including cingulum cingulate), which can be explained by the observation of gliosis and spongiosis in pathology studies (Vaillancourt et al., 2009). Considering the consistency of the results obtained by the classical TBSS analysis method and the AFQ method, the reliability of the AFQ technique applied in WD patients was demonstrated.
Notably, the new local quantification means to process the data could reveal more information. We find that diffusion properties vary substantially along the trajectory of a fascicle. For example, in the callosum forceps minor, MD is low as the tract terminates at various targets within the frontal lobe and rises as the tract forms a cohesive bundle while crossing the midline via the genu of the corpus callosum. Disruption of this fiber may potentially affect brain connectivity, leading to abnormal functional connectivity in the cholinergic circuitry of the basal forebrain related to cognitive function (Selden et al., 1998;Zhou et al., 2019). Similarly, RD is high in the thalamic radiation where the tract originates but drops considerably as the tract enters the dense collection of myelinated white matter in the anterior limb of the internal capsule and rises again as the tract enters the cortical gray matter. The ATR contains the white matter fibers that connect the frontal cortex with the thalamus and basal ganglia, and disruption of this fiber may cause memory impairment (Niida et al., 2018;Wakana et al., 2004). The IFOF connects the occipital lobe and frontal lobe and contains fibers that connect the frontal lobe with the posterior part of the parietal and temporal lobes (Taoka et al., 2006). The present study showed that the most severe alterations in WD occur in the anterior and posterior parts of this fiber tract, which may explain the cognitive and motor deficits in WD. The SLF has been shown to play an important role in advanced cognitive functions (Zheng et al., 2021), and WD-associated abnormalities in the SLF have also been confirmed by previous related studies (Hu et al., 2021). The UF connects the orbitofrontal cortex with the anterior part of the temporal lobe and plays an important role in memory. The UF is reportedly associated with memory (Highley et al., 2002), and this daily functional ability is significantly decreased with diffusion abnormalities in WD (Hu et al., 2021). However, we have not found sample size to confirm the results that we report here. Second, HCs did not receive the neuropsychological tests and were recruited solely based on an unremarkable medical history, which may limit the neurophysiological alterations specific to individuals at risk for WD. Third, due to the strict criterion in fiber tracking, fiber tracts such as the bilateral cingulum hippocampus failed to track in some subjects. In the future, more advanced algorithms will be incorporated into longitudinal studies to validate the feasibility and accuracy of AFQ methods. Finally, DTI detected changes in BFcortical connectivity, but we could not determine whether such changes were the result of the degeneration of cholinergic neurons in the BF or whether they were indirectly caused by neurons in other regions. As already adopted by Liu et al. (Liu et al., 2017) and Nemy et al. (Nemy et al., 2020), using an ROI-based approach, the use of the BF as a seed region to map cholinergic pathways in our study may provide more precise evidence of defective connectivity between the BF and target cortical regions.

Conclusion
In summary, we applied AFQ to identify major WM fiber bundles of the BF-cortical network in HC and WD patients. We found that microstructural integrity was vulnerable in the BF-cortical network in WD. Furthermore, the alterations in ILF properties were significantly correlated with TBPM and MMSE scores. These results support the hypothesis that impairment of the projections between cholinergic input and output areas-the cerebral cortex and BF-may be the primary cause of PM impairment in WD. Treatments targeting the microstructure of the BF-ILF tracts may contribute to restoring and improving PM performance in WD. Finally, the current study demonstrates that AFQ techniques can be used to increase our understanding of white matter changes in WD.
Author contribution SH and YW designed the experiment, analyzed experimental results, and wrote the manuscript; YW revised the manuscript; HW and TD researched the literature and decided whether the literature was included when disagreements occurred; AW and CL conducted preprocessing of MRI data; HK guided, reviewed, and revised the manuscript and provided unique insights into the direction of the discussion. All of the authors read and approved the final manuscript.
Funding This work was supported by grants from the Natural Science Foundation of Anhui Province (KJ2021A0580) and the Natural Science Foundation of Anhui Province (KJ2020A0419).

Data availability
The data that support the findings of this study are