Fig. 1 Schematic of semi-automated tissue segmentation pipeline.

Fig. 2 Temporal FA and MD profiles in the four deep gray nuclei across 12 years. In each line chart, the blue solid line indicates mean DTI metric in healthy controls, whereas the red dotted line indicates mean DTI metrics in PD patients. The upper and row charts refer to FA and MD changes respectively.

Figure 1. Investigation of white matter tract changes between Parkinson’s and healthy subjects, using fixel-based analysis (FBA). The white matter tracts that are statistically significant (p<0.05) in fiber cross-section (FC), fiber density (FD) and fiber density and cross-section (FDC) are shown overlaid on a group-averaged template with a heat map. The cross-hair in each sub-figure points at the same location.

Figure 2. Investigation of white matter changes between Parkinson’s and healthy subjects, using voxel-based analysis with fractional anisotropy (FA) and mean diffusivity (MD). The regions that are statistically significant (p<0.05) are shown overlaid on a group-averaged template with a heat map. The cross-hair in each sub-figure points at the same location.

Figure 2. Mean diffusion metrics in healthy controls (HC), early-stage Parkinson’s disease (EPD) group, and moderate-late-stage Parkinson’s disease (MLPD) group. (a) Mean diffusion metrics in substantia nigra (SN) in HC, EPD, and MLPD. (b) Mean diffusion metrics in globus pallidus (GP) in HC, EPD, and MLPD. (c) Mean diffusion metrics in thalamus in HC, EPD, and MLPD. * and※ indicate p < 0.05 and p < 0.01.

Figure 4.Receiver operating characteristic curves of the diagnostic performance of diffusion metrics in substantia nigra(SN), globus pallidus(GP), thalamus and their combinations for discriminating Parkinson’s disease (PD) patients from healthy controls (HC) or between early-stage Parkinson’s disease group (EPD) (and moderate-late-stage Parkinson’s disease group (MLPD)) and HC. Plots of the FA, AD, MD, MK, AK in the SN and MK, RK in the GP and FA, MK and RK values in the thalamus and their combinations for diagnosing PD (a), EPD (b), and MLPD (c).

Figure 1. (A) Substantia nigra mask (orange region) from the CIT168 atlas. (B) Mean kurtosis in each region of interest compared between groups. Bars show the raw uncorrected data while p-values and adjusted R2 values given are for the corrected model controlling for age, sex and education. (C) Radial plots of diffusion kurtosis indices for each region and both groups. (D) Plots of the estimated marginal means from repeated-measures ANOVA. Error bars show 95%CI. * Significant for p<0.05.

Figure 3. The brain areas with significantly decreased track density values postoperatively, overlaying on the T1 template. A) Left globus pallidus and B) genu of the corpus callosum. The color bar represents the t-value of the T-test.

Figure 1. The A) fractional anisotropy (FA) and B) track density image (TDI) from the same patient. TDI has a higher spatial resolution than that of the FA image (1 × 1 × 1 mm3 vs 2.2 × 2.2 × 2.2 mm3, respectively).

Figure 2: Change over disease duration in striatal DaT, SNc NM and SNc iron in PDs relative to HCs. For all PDs relative to HCs, in each functional subregion of the striatum (A-C) and SNc (D, E) in the most affected brain hemisphere, the mono-exponential fitting functions for the percent ratios of DaT striatal binding ratio (SBR), NM SNR and QSM are plotted against disease duration

Figure 1: Group differences in striatal DaT and SNc NM and iron levels. Functional subregions in the striatum (A) and the SNc (D). The panels show the results of the statistical comparison between groups in the clinically most (B, E) and least affected brain hemispheres (C, F). Each panel shows the inter-group comparison between iRBDs and HCs (left column), PDs and HCs (centre column), or PDs and iRBDs (right column). P-values were converted into z-scores

Figure 1. Illustration of the analysis pipeline: first, an age-specific healthy control’s mean brainstem template was generated, along with their respective spatial prior probability maps with three classes: brainstem outside of the SN (BS), SN, and irrelevant background (BG and it is not shown here). The same procedure was performed to transform NM-MRIs in the same space as the prior probability maps. Finally, the Bayesian classification was applied for partitioning the SN from the brainstem in both image intensity and in space and the mean intensity of the brainstem was extracted.

Figure 2. The annualized %change of MRI metrics (a. % increase Free-water and b. % reduction of NM contrast to background ratio) of ventral and dorsal SN in PD and controls. c. The scatter plot of annualized % increase of FW values versus the annualized % reduction of NM contrast ratio in PD. Error bars show SEM.

Figure 4. Significant FC-traits of PLS component 1 (shown in blue in Figure 3) Each FC-trait is represented by its FC circular plot thresholded to 1% of the strongest connections grouped by functional networks organization, and the map of FC node degree (i.e. sum of functional connections from and to each brain region). The variance explained by each FC-trait is also indicated.

Figure 5. Significant FC-traits of PLS component 2 (shown in yellow in Figure 3). Each FC-trait is represented by its FC circular plot thresholded to 1% of the strongest connections grouped by functional networks organization, and the map of FC node degree (i.e. sum of functional connections from and to each brain region). The variance explained by each FC-trait is also indicated.

Figure 1. Significantly weaker FC with bilateral POCs in the left substantia nigra (SN), temporal pole, interior temporal gyrus, insular cortex, and posterior hippocampus and parahippocampal gyrus of PDs (2-sample t-test, uncorrected, p < 0.005, extent threshold = 6 voxels)

Values are presented in mean ± SD. *, p < 0.05, MoCA, Montreal Cognitive Assessment; UPDRS, Unified Parkinson’s Disease Rating Scale; N/A, not applicable.

Figure 2. (A) 3D-rendered view of postoperative CT images of DBS patients showing extracranial trajectories of leads and extensions. (B) Schematic of the experimental phantom showing a full DBS system as well as examples of some trajectories used during the RF heating experiments. (C) Setup for the measurement of the magnetic force on the IPG.

Figure 3. Steps of DBS implantation and artifact measurement. (A) Formalin fixed cadaveric brain placed inside a 3D printed skull. (B) 1.5T MRI of the cadaveric brain were transferred to BrainLAB iPlan server for localizing subthalamic nucleus. (C) Leksell G Frame and stereotactic arc attached to the skull for electrode implantation. (D) Burr hole cover placed to fixate the lead in position. (E) Artifact was quantified on transverse and coronal planes for different imaging sequences.

Figure 1: Cerebellum atrophy in comparison of(a) HC and SCA2 (b)HC and SCA12 (c) SCA2 and SCA12 at the significant level of p<0.001, uncorrected (t-test) displayed on SUIT Flatmapview(color map: jet and threshold 0:0.1)

Table 1. Significant difference in the white matter between SCA2 and SCA12 (analysis using SUIT, one way ANOVA, p<0.05 FWE, voxel threshold (k)≥5)

Table 1: Parameters forwhole-brain network, cerebral subnetwork and embedded cerebro-cerebellar subnetwork.Optimal value for global coupling (Gcoupl) and Pearson correlation coefficients (PCC) between structural connectivity (SC), experimental (expFC) and simulated functional connectivity (simFC) are reported.

Figure 1: A) Experimental matrices of structural connectivity (expSC) and functional connectivity (expFC) in the patient with Joubert syndrome. B) Simulated functional connectivity (simFC) matrices for whole-brain network (left), cerebral subnetwork (center) and embedded cerebro-cerebellar subnetwork (right). In each matrix, rows and columns represent a specific brain region (node), while each intersection point represents a connection, weighted by the number of streamlines, between two nodes (edge).

Figure 1 Data processing procedures. A. The labels of subcortical nuclei in processed QSM images. B. The framework of quality control steps.

Figure 2 The association between genetic variations and imaging phenotypes. A. Heat map with significant associations between SNP and QT at p<10-5 (blocks labeled with “X”). Color key on the top left coded the magnitude of -log10(p-values). B, C. Manhattan and Q-Q plot of the most significant association (p<10-6) (rs602201-R_SN). The horizontal line displayed the cutoff for p<10-6. Shown on (C) is the Q-Q plot of the distribution of the observed p-values (-log10(observed p-value)) versus the expected p-values (-log10(expected p-value)) under the null hypothesis of no association.

Tract-based statistical analysis of DKT/DTI metrics between symptomatic SCA3 patients and health controls. Red-yellow highlights show areas of decreased DKI/DTI metrics values in symptomatic SCA3. Blue highlights show areas of increased MD value in symptomatic SCA3.(P＜0.05, TFCE correction)

Tract-based statistical analysis of DKT/DTI metrics between symptomatic SCA3 patients and health controls. Red-yellow highlights show areas of decreased DKI/DTI metrics values in symptomatic SCA3. Blue highlights show areas of increased MD value in symptomatic SCA3.(P＜0.05, TFCE correction)

Figure 2: Representative CERES image of cerebellum taken from Healthy control (A) and GA patient (B).

Table 2: Volumetric estimation of Brain structures of GA patients and Healthy Controls.

Figure 4: ICC score boxplots for cerebellum pacellations for A) intra-scanner, B) Pre-harmonization inter-scanner and C) Post-harmonized inter-scanner.

Figure 2: Boxplot of ICC scores for GM/WM hemispheric masks of Cerebrum & Cerebellum for inter- and intra-scanner. The ICC score is interpreted on a scale of 0 to 1, with 1 indicating absolute repeatability and anything below 0.5 considered poor for reproducibility.

Figure 1. Flow of chart illustrating the voxel-wised analysis. PD- Parkinson’s disease, HC-healthy control, QSM- quantitative susceptibility mapping, NM_CNR- contrast to noise of neuromelanin.

Figure 4. Negative Pearson’s correlation map of substantia nigra pars compacta (QSM-NM_CNR overlap) PD- Parkinson’s disease, HC-healthy control, QSM- quantitative susceptibility mapping, NM_CNR- contrast to noise of neuromelanin.

Figure 1. The stages of mapping the boundaries from the template space to the original space for neuromelanin. Each column represents a different slice. A) NM template; B) the transformed NM template; C) the same boundaries superimposed on the original midbrain images; and D) final boundaries after DPA was used to refine the boundaries.

Figure 2. The stages of mapping the boundaries from the template space to the original space for the QSM data. Each column represents a different slice. A) the QSM template; B) the transformed QSM template; C) the same boundaries superimposed on the original midbrain images; and D) final boundaries after the DPA was used to refine the boundaries. The fourth and fifth columns show the SN boundary in red and the STN boundary in orange. The second and third rows show the RN boundary in light green.

SN ROI derived from (1) semi-automated segmentation on SMWI images (left) by thresholding for voxels with high iron deposition containing signal intensity 7 standard deviations less than background in and (2) NMS (right) by thresholding for voxels with signal intensity 4 standard deviations higher than background in PD patient LRRK2 risk-variant carriers (top row) and non-carriers (bottom row).

ROC analysis of QSM susceptibility, high-iron SN size derived from SMWI, and combination model of QSM susceptibility and high-iron SN size to classify LRRK2 risk-variant carriers and non-carriers in PD patients.

Figure 1. Illustration of (A) independent component analysis for calculation of cortico-cerebellum during semantic word recognition task in congenital blind relative to sighted control. (B) Difference in cerebellar areas effect size and (C) estimation of gray matter volume.

Table 2. Significant functional connectivity of cerebellum and language integration estimated by using independent component analysis for congenital blind children group as compared to sighted control group.

Figure 1: Substantia Nigra pars compacta regions of interest (left in yellow and right in pink) based on neuromelanin-sensitive imaging of a representative healthy volunteer.

Figure 2: Box plot of Substantia Nigra pars compacta volume in Atypical Parkinsonian along with healthy volunteers. HV is Healthy Volunteers, PD is Parkinson's Disease, PSP is Progressive supranuclear palsy, DCB is CorticoBasal Degeneration, MSA is Multiple System Atrophy, with its cerebellar (MSAc) and Parkinsonian (MSAp) subtypes, and DCL is dementia with Lewy body.

Figure 2. Fixels with significant (p < 0.05, FWE-corrected) decrease in fixel-based metrics. (a) MSA-P versus controls. Subtle changes were observed in the cohorts with disease duration≦3 years; on the contrary, FC and FDC were significantly decreased in the (b) MSA-C versus controls. Streamlines were colored by direction (anterior-posterior: green; superior-inferior: blue; left-right: red).

Figure 1. Significant changes in fixel-based metrics in patients with MSA compared to healthy control subjects. Regions of significant changes in FD, FC and FDC were displayed stereoscopically in the (a) sagittal, (b) superior left frontal and (c) inferior right occipital view. Streamlines corresponding to significant fixels (family-wise error corrected p < 0.05) were illustrated and colored according to p values. Cold color represented reduction, whereas warm color represented increase.

Fig. 2. Graph shows receiver operating characteristic curve to assess the utility of two different models with the clinical variables included or not (with and w/o clinical information) for predicting global motor outcome.

Fig. 1. Illustration of the processing pipeline of the radiomics model with machine learning (RA-ML). RF=radiomics feature; RFE= recursive feature elimination

A, B, and C: Relationships between LC integrity, the change rate of UPDRS III score, and the change rate of somatomotor network synchronization in PD group. D: The difference of somatomotor network synchronization among HC, PD during OFF and ON.

Comparison of CNR_LC between the HC and PD groups. (A) Signal intensity measurements of the LC (two small red circles) and pontine (a big red circles) from a HC; (B) and (C) The location of LC in continuous layers (red arrow); (D) Significantly decreased CNR_LC was found in PD group when compared with HC group.

Figure 3. The architecture of CNN-LSTM for motion assessment

Figure 4. Confusion Matrix and ROC analysis (validation set) (a), (c) CNN , (b), (d) CNN-LSTM.

Figure 2. Representative QSM images from control, RBD, PD Stage 1 and PD Stage 2 subjects. Note increase in susceptibility in substantia nigra (indicated by red arrows) in PD Stage 1 and PD Stage 2 compared to healthy controls or RBD. Scale bar in ppm (parts per million); RBD, REM-sleep behavior disorder

Figure 1. Average susceptibility in the substantia nigra (SN; left) and red nucleus (RN; right) across all cohorts. Left: there is significant increase in susceptibility PD as compared to controls. Right: there are no statistically significant differences in susceptibility in the RN across all cohorts. RBD, REM-sleep behavior disorder; ppm, parts per million; *, p < 0.05 (multiple comparison t-test, Bonferroni correction)

Figure 1. NAA/Cr was significantly decreased in the contralateral SN of a patient with tremor-dominant Parkinson’s disease (PD) (A) compared with that of a patient with essential tremors (ET).

Figure 2. Sample dentate nucleus ROIs and contrast of mpMRI data at 3T. T1w: blue indicates the location of the participants’ left dentate nucleus; QSM: heatmap indicates estimated iron content in the left dentate nucleus mask as measured via R2* mapping (color scale shows rendering of values 5.0-40.0 s-1); DWI: dominant fiber orientation coded as follows: red=left-right, green=anterior-posterior, blue=superior-inferior. DD=disease duration. SARA=Scale for the Assessment and Rating of Ataxia.

Figure 3. Sample spectroscopy data. Green circles indicate the location of the Myo-inositol peak (3.5ppm) for FA patient #4 (left panel) and a healthy volunteer (HV; center panel). The white square shows the location of the MRS volume of interest placed over the right dentate nucleus (right panel).

Combining in vivo (A) and postmortem (B) MRI with 3D immunohistochemistry (D, E) to study the anatomical underpinning of the swallow tail sign (ST). The ST was segmented (C) as a bright stripe in SN on in vivo MRI (A). N1 was segmented (F) as dark-pigmented areas on grayscale BF images (D) verified by calbindin immunohistochemistry (E). Co-registration (G) of 3D immunohistochemistry, postmortem MRI and in vivo MRI revealed contrast mechanisms of N1 and its relation to the swallow tail sign.

Masks of the ST and N1 for three randomly assigned pairs of in vivo and postmortem datasets (A, B, C) overlaid over BF. In all cases, ST covered a large part of N1. While N1 consistently showed a narrow width, ST was approximately twice as wide as N1. The ST only covered the superior-posterior-lateral portion of N1, while it did not match the ventromedial part of the rostral extent of N1. The anatomical medial (M), lateral (L), superior (S), and inferior (I) directions are illustrated in A.

Figure 2: Statistical parametric maps showing positive and negative correlation between grey matter volume and UPDRS-III scores at 48 months using voxel-based morphometry analyses.

Results showed a positive association between grey matter volume and UPDRS-III in right pallidum at 48 months (p<0.05 TFCE, FWE corrected, age, sex and total intracranial volume adjusted). Conversely, a negative association was identified in bilateral thalamus at 48 months. No findings were significant at baseline.

Figure 4: Correlation between grey matter volume and UPDRS-III scores at baseline and 48 months.

At baseline, one positive correlation was identified between grey matter volume and UPDRS-III scores in the right pallidum. At 48 months, a positive correlation was observed between grey matter volume and UPDRS-III scores in bilateral pallidum whereas a negative correlation was identified in bilateral thalamus. Pearson correlations are reported with a statistical threshold of p<0.05.

Figure 1. Regions of significant CBF decreases of LP compared to HP in all of the participants according to general cognitive and executive domain scores. The color bars represent t values. R: Right, L: Left.

Table 1. Regions of CBF differences between two groups defined based on composite z-scores in general cognitive domain and executive domain within the entire subject group.

* The composite z-scores are the average of the z-converted raw scores of ACE-R, MMSE and MOCA. Median of the composite z-scores: - 1.22.

** The composite z-scores are the average of the z-converted raw scores of WCST percentage of perseverative responses score and Stroop interference duration score. Median of the composite z-scores: 0.56.

Figure 1. Functional Connectivity (FC) between the basal nucleus of Meynert (BNM) and whole brain. (A–C): BNM-FC was evaluated in healthy control (HC) (A), Parkinson’s disease with normal cognition (PD-NC) (B), and Parkinson’s disease with mild cognitive impairment (PD-MCI) (C) groups. (D): BNM-FC was decreased in the right superior parietal lobe (SPL) of the PD-MCI group compared to the PD-NC and HC groups.

Figure 2. Classification accuracy of functional connectivity (FC) in the basal nucleus of Meynert (BNM) (BNM-FC) in the right superior parietal lobe (SPL) determined by the leave-one-out cross-validation (LOOCV) method in distinguishing Parkinson’s disease with mild cognitive impairment (PD-MCI) from Parkinson’s disease with normal cognition (PD-NC) patient (A–C). Box plots with whiskers (min–max) show BNM-FC in the right SPL of the three groups (D).

Fig. 1 Identical paired standard T1-weighted images of a healthy 84 year-old male control illustrating automated anatomical placement of color-coded landmarks for midbrain (yellow) and pons (blue) areas (first row), followed by left and right SCP and MCP widths respectively. Abbreviation: MCP = Middle Cerebellar Peduncle, SCP = Superior Cerebellar Peduncle.

Fig. 3 Bar charts of coefficients of variance (CV, y-axes) of values and subcomponents of MRPI (x-axes) from (A) manual and automated MRPI derived from longitudinal SGH-PD cohorts, and (B) automated MRPI derived from single TP ADNI1 (1.5T) and ADNI 2/GO (3T) cohorts. Abbreviation: MCP = Middle Cerebellar Peduncle, SCP = Superior Cerebellar Peduncle, TP = Time Point.

Fig 1. The volume and contrast ratio (CR) of neuromelanin, and susceptibility in the substantia nigra pars compacta in HC and EPD

Fig 2. Correlations between the contrast ratio (CR) of neuromelanin and susceptibility in the substantia nigra pars compacta in HC and EPD groups (adjusted for age and gender)

Figure 1. (A) T1 map and (B) T2 map from a patient with tremor-dominant Parkinson’s disease (PD). (C) T1 map and (D) T2 map from a patient with essential tremors (ET). The T1 value in the substantia nigra is significantly higher in the patient with tremor-dominant PD than in the patient with ET.

Figure 3: (A) 3D reconstruction of a DBS electrode placement with respect to the manually delineated GPe (green)/GPi (yellow) for a specific PD patient. (B) 3D reconstruction of the same patient’s data showing the DBS electrode with respect to GP-net’s segmentation of the GPe (orange)/GPi (blue). Green arrow – anterior (A), blue arrow – superior (S) and red arrow – right (R).

Figure 4: (A) 3D reconstruction of a DBS electrode placement with respect to the manually delineated STN for a specific PD patient. (B) 3D reconstruction of the same patient’s data showing the DBS electrode with respect to STN-net’s segmentation of the STN. Green arrow – anterior (A), blue arrow – superior (S) and red arrow – right (R).

Figure 1. The results of the Segmentation and Skeletonization for the PD patients. a) Male PD patient, 63 years old. The venous tree is visually incomplete, the superficial cortical veins are missing (red arrowheads); b) Male PD patient, 74 years old. Increased tortuosity and fractal dimension (red arrowheads) of the venous system.

Figure 2. The results of the Segmentation and Skeletonization of the AMC, Female, 48 years old.

CBF comparison between PD-MCI and normal controls (Yellow: PD-MCI>NC, corrected p<0.05)

CBF comparison between PD-MCI and PD-NC (Yellow: PD-MCI>PD-NC, corrected p<0.05)

Figure 2. Images showing brain regions with significant differences between PD and healthy controls (HC). (A) The left temporal fusiform cortex, posterior division (TFCp) is shown in red, right TFCp is shown in green. (B) The right parabrachial pigmented (PBP) nucleus is shown in blue. R2* of right TFCp and right PBP was significantly higher in PD than in HC; T2* of left TFCp, right TFCp and right PBP was significantly lower in PD than in HC; T1 mapping ofright PBP was significantly lower in PD than in HC.

Figure 1. Images from T1-weighted enhanced imaging (T1WE), T1 mapping, T2*, R2*, and susceptibility-weighted imaging (SWI) mapping acquired using strategically acquired gradient echo imaging. T1WE shows improved image quality over the other methods.

Figure 1. Results of MNI PD25 atlas segmentation using Lead-DBS SPM and ANTs co-registration, normalization. Left shows the 22 structures in 3D models overlaid on top of an example subject’s b0 FA image.

Figure 2. Connectivity matrices of significant differences in qa, fa, md, ad, rd of MNI PD25 basal ganglia structure interconnectivity. Left column represents connectivity matrix of mean differences in qa, fa, md, ad and rd diffusion measures. Middle column represents p-values of student t-tests done for every connectivity matrix pair with significance at p < 0.05. Right column represents the adjusted p-values of t-tests using Benjamin-Hochberg procedure with p < 0.2.

Figure 1. Experimental design for investigating the effects of PM. According to the living condition, three experimental groups were established, including a control group, HEPA group and PM group. The tractography analysis and the rotarod performance test were conducted to respectively examine the effects of PM on the structural changes within rat brains and the motor-behavioral performances.

Figure 2. Whole-brain in-vivo DTI tractography. Based on the function of the neural circuits, DTI tractography analysis was conducted with the ROIs divided into the following three groups, including (A) the olfactory circuit, (B) motor control circuit and (C) dopaminergic pathway.

Table 1 The clinical data of all groups

Figure 1 ROI selection for gray matter nuclei

Figure 3. Box plots of FW for the HC, GBA-PD, and iPD groups in Braak stages III and IV. The bottom and top of the box are the first and third quartiles, and the thick band inside the box is the median. Whiskers represent maximum and minimum values of all data.

Abbreviations: FW, free water, GBA-PD, Parkinson’s disease with GBA1 mutations; HC, healthy controls; and iPD, idiopathic Parkinson’s disease.

Figure 2. Box plots and tract profiles of FA_T, MD_T, and FW for HC, GBA-PD, and iPD groups in the white matter pathways.

Abbreviations: ATR, anterior thalamic radiation; CCG, cingulum cingulate gyrus; FA_T, free water-corrected fractional anisotropy; Fmajor, forceps major; Fminor, forceps minor; FW, free water; GBA-PD, Parkinson’s disease with GBA1 mutations; HC, healthy controls; ILF, inferior longitudinal fasciculus; iPD, idiopathic Parkinson’s disease; MD_T, free water-corrected mean diffusivity; UF, uncinate fasciculus.

Figure 5: Betweenness Centrality and nodal Clustering Coefficient positive correlations (red nodes) and negative correlations (blue nodes) with EFs scores in nodes enrolled in the analysis. SFG: superior-frontal-gyrus; IPG: inferior-supramarginal-parietal-gyrus.

Figure 3: Topological distribution of node degree (k); the size of each node represents k. SFG: superior-frontal-gyrus; OG: orbital-gyrus; SPG: superior-parietal-gyrus; IPG: inferior-supramarginal-parietal-gyrus

3D illustration of DBS leads and subcortical structures. Subthalamic nucleus in orange, volume of tissue activated in red and streamlines in dark blue.

Hyperdirect pathway activation. In blue, points corresponding to not clinical effect; in orange, points corresponding to clinical effect. Correct predictions are marked as circles and incorrect predictions as crosses.

Table 2. Small-worldness mediation analysis results

Table 3. Rich-club coefficients mediation analysis results

Figure2:Cluster centroids of brain states

Figure1: Independent functional components and corresponding static functional connectivity network

Table 3 The track count number of nigrostriatal pathway of both left and right sides in DBS ON/OFF images and their comparisons.

Figure 2 The deterministic tracking result of the same representative subject in two states: a) DBS ON, and b) DBS OFF.