Skip to main content
NCBI home page
NeuroImage: Clinical logoLink to NeuroImage: Clinical
. 2023 May 9;38:103430. doi: 10.1016/j.nicl.2023.103430

The altered spontaneous neural activity in patients with Parkinson's disease and its predictive value for the motor improvement of deep brain stimulation

Bowen Yang a, Xiu Wang a, Jiajie Mo a, Zilin Li a, Wenhan Hu a,b, Chao Zhang a,b, Baotian Zhao a, Dongmei Gao a, Xin Zhang a,b, Liangying Zou a,b, Xuemin Zhao c, Zhihao Guo a, Jianguo Zhang a,b,⁎, Kai Zhang a,b,⁎
PMCID: PMC10197096  PMID: 37182459

Highlights

  • •

    Patients with PD showed specific abnormal ALFF patterns compared to healthy controls.

  • •

    UPDRS-III improvement was correlated with ALFF in cerebellum and primary motor cortex.

  • •

    Regression and classification models were used to predict STN-DBS efficacy in PD.

  • •

    Regression model showed a correlation between the predicted and observed DBS outcomes.

  • •

    The ROC analysis revealed an AUC of 0.94 in differentiating DBS responders.

Keywords: Parkinson's disease, Deep brain stimulation, Resting-state functional magnetic resonance imaging, Machine learning

Abstract

Background

This study aims to investigate the altered spontaneous neural activity in patients with Parkinson's disease (PD) revealed by amplitudes of low-frequency fluctuations (ALFF) of resting-state fMRI, and the feasibility of using ALFF as neuroimaging predictors for motor improvement after bilateral subthalamic nucleus (STN) deep brain stimulation (DBS).

Methods

Fourty-four patients and 44 healthy controls were included in this study. First, the ALFF of patients with PD was compared with that of controls; then significant clusters were correlated with motor improvement after DBS (unified Parkinson's disease rating scale (UPDRS-III)) and other clinical variables. Second, regression and classification of the machine learning models were conducted to predict motor improvement after DBS. Receiver operating characteristic (ROC) analysis was used to evaluate the performance of the classification model.

Results

Compared with healthy controls, patients with PD showed increased ALFF in the bilateral motor area and decreased ALFF in the bilateral temporal cortex and cerebellum. The Hoehn-Yahr stages correlated with ALFF within the bilateral cerebellum (p = 0.021), and UPDRS-III improvement correlated with ALFF in the left (p < 0.001) and right (p = 0.005) motor areas. The regression model showed a significant correlation between the predicted and observed UPDRS-III changes (R = 0.65, p < 0.001). The ROC analysis revealed an area under the curve (AUC) of 0.94 which differentiated moderate and superior DBS responders.

Conclusion

The results revealed altered ALFF patterns in patients with PD and their correlations with clinical variables. Both binary and continuous ALFF can potentially serve as predictive biomarkers for DBS response.

1. Introduction

Subthalamic nucleus deep brain stimulation (STN-DBS) is a widely used surgical therapy for motor dysfunction in Parkinson's disease (PD) (Deuschl et al., 2006, Fasano and Lozano, 2015, Follett et al., 2010). Considering the variability in motor function improvement after deep brain stimulation (DBS), exploring reliable DBS outcome predictors is beneficial for selecting candidates for this therapy. Although some clinical predictors have been recognized, such as levodopa responses (Cavallieri et al., 2021, Yang et al., 2021), age (Ballarini et al., 2019), and electrode positions (Johnsen et al., 2010), their clinical applicability may be constrained by their unsatisfactory predictive accuracies. Thus, it is of the utmost importance to investigate more precise predictors of the efficacy of this therapy.

Previous research has also shown that DBS outcome correlates with some neuroimaging biomarkers, most of which use structural MRI and PET-CT scans (Fasano et al., 2012). Through structural MRI, researchers have found that patients with greater cortical thickness in the visual-motor areas present more significant motor improvement after STN-DBS (Frizon et al., 2020). Structural MRI measurements have also revealed associations of neuropsychiatric symptoms with cognitive function in patients treated with STN-DBSs (Mosley et al., 2020, Weinkle et al., 2018). Additionally, a PET-CT study demonstrated that the modulation effect on the metabolic networks correlated with the effects of STN-DBS (Feigin et al., 2007). Many DBS targets are highly functionally connected, implying that DBS may modulate large-scale brain networks to achieve the desired effects. Compared to other scanning methods, resting-state functional MRI (rs-fMRI) may be more appropriate for delineating the properties of large-scale brain networks, that reflect spontaneous neural activity in blood oxygen level-dependent (BOLD) signals (Biswal et al., 1995). Combined with the reported association of fMRI with the clinical variables of PD (Wong et al., 2020), fMRI can potentially become a predictor of DBS outcomes.

The efficacy of fMRI in predicting DBS outcomes remains to be explored (Loh et al., 2022). An fMRI study demonstrated that the location of effective stimulation correlated with STN-DBS outcomes (Horn et al., 2017). In addition, one study showed postsurgical motor outcomes were positively correlated with preoperative functional connectivity between the internal globus pallidus and STN (Younce et al., 2021). The above findings only confirmed the relationship between fMRI and DBS outcomes, rather than predicting DBS outcomes using fMRI. However, patient-specific prediction of DBS outcomes by fMRI still needs to be investigated. In addition, the functional connectivity method highly relies on the region of interest (ROI) hypothesis (Lv et al., 2018), which is highly biased and limited to clinical application. In contrast, this study applied ALFF, a measurement independent of the hypothesis (Zang et al., 2007). In contrast to functional connectivity reflecting the functional integration between different brain regions(Lv et al., 2018, Wurina et al., 2012), the ALFF focuses on local spontaneous neural activity. ALFF measures the total power of the BOLD signal within the low-frequency range between 0.01 and 0.1 Hz, which will remove physiological high-frequency noise. Although the specific mechanism of ALFF is still unclear, it showed remarkably high temporal stability(Küblböck et al., 2014) and long-term (approximately 6 months) test–retest reliability. With these advances, the ALFF was used in the present study (Zuo and Xing, 2014).

In this study, we measured the ALFF using rs-fMRI to investigate the altered patterns of spontaneous neural activity in patients with PD and its predictive value for the motor efficacy of STN-DBS. Initially, the altered ALFF patterns in patients with PD were determined by inter-group comparison with controls. Significant clusters from the inter-group comparison were then correlated with clinical variables. Finally, regression and classification machine learning models were used to predict motor improvement after STN-DBS in patients with PD.

2. Methods

2.1. Participants

Fourty-four patients with PD and healthy controls were retrospectively included, and six other patients were excluded during candidate selection. The inclusion criteria for patients were as follows: (1) patients who met the UK Brain Bank criteria for PD (Hughes et al., 1992); (2) patients who underwent 3D T1-weighted magnetization-prepared rapid gradient echo imaging (MPRAGE), rs-fMRI, and cognitive testing; (3) patients who completed the postoperative CT scan, to confirm the accuracy of the electrode positions; (4) patients who underwent presurgical evaluation (items listed in section 2.2). The UPDRS-III (Goetz et al., 2008) was assessed before and after the levodopa challenge test (Albanese et al., 2001). The exclusion criteria for patients were as follows: (1) the Euclidean distance between the lead tip positions and intended target coordinates was higher than 3.0 mm; (2) frame displacement was higher than 3.0 mm; (3) the imaging has magnetic field distortion. The exclusion criteria for healthy controls included a history of head injury, psychiatric or neurological disease, other major medical diseases, and alcohol or drug dependency or abuse. This study was approved by the Ethics Committee of Beijing Tiantan Hospital in Beijing, China, in accordance with the Declaration of Helsinki. Informed consent was obtained from all the participants.

2.2. Surgical procedures and clinical evaluation

Surgical procedures were performed as described previously (Fan et al., 2020). The DBS electrodes (Model 3389; Medtronic Inc., Minneapolis, Minnesota, USA) were implanted and fixed under local anesthesia using a Leksell micro-stereotactic system. Intraoperative macro-stimulation and microelectrode recordings were used to confirm electrode position. The electrodes were connected to the pulse generator. The electrode position was reconstructed using postoperative CT scanning. The distance between the center of the lead tip and the closest voxel within the STN according to the DISTAL Minimal atlas was calculated and compared between two groups with different surgical outcomes (Ewert et al., 2018). Three subjects were excluded because the Euclidean distance between lead tip positions and the intended target coordinates was higher than 3.0 mm.

Motor function was evaluated in two preoperative statuses (med-off and med-on) and two postoperative statuses (med-off/stimulation-on and med-on/stimulation-on) at 1-month follow-up (M1). Notably, motor improvement of DBS was determined by the change rates in the UPDRS-III scores between the postoperative and preoperative med-off statuses. A levodopa challenge test was administered to assess motor function in the med-off and med-on states. The med-off state was assessed at least 12 h after the medication withdrawal. In contrast, the med-on state assessment was conducted 60 min after taking regular medications plus an additional 50/12.5 mg of Madopar. The clinical assessment also included symptom-related scales (freezing of gait questionnaire (FOGQ), Parkinson's disease questionnaire (PDQ), cognitive test batteries, mini-mental state examination (MMSE), Montreal cognitive assessment (MoCA), and neuropsychological assessment (e.g., the Hamilton anxiety rating scale (HAMA), and Hamilton depression rating scale (HDRS)). The levodopa equivalent daily dose (LEDD) was recorded before the DBS operation and at a 1-month follow-up. One month after the DBS operation, the patients returned to the hospital to switch on the pulse generator. After a 3–7 days observation period, the patients were introduced to the evaluation center for an overall motor and neuropsychological assessment.

2.3. MRI data acquisition and post-processing

MRI scans were acquired using a 3 T Siemens Prisma scanner (Siemens Medical System, South Iselin, NJ, USA). The scanning series included a 3D T1 magnetization prepared rapid gradient echo sequence and resting-state fMRI. The following parameters were used to obtain BOLD-EPI images: 48 axial slices with a thickness of 3.0 mm, in-plane matrix 74 × 74, field of view (FOV) of 1554 × 1554 mm2, repetition time (TR) of 750 ms, echo time (TE) of 30 ms; frames of 400, total scan time of 5 mins. The following parameters were used to obtain 3D T1 images: 176 sagittal slices, slice thickness of 1.0 mm, 256 × 256 in-plane matrix, TR of 1560 ms, TE of 1.70 ms, inversion time of 778 ms, and FOV of 240 × 240 mm2. The scans were obtained without sedation or levodopa while the patients were awake and with their eyes closed.

DPARSF software was used for rsfMRI data processing (Chao-Gan and Yu-Feng, 2010). The following steps were performed for the rsfMRI images processing: (1) removing the first ten volumes to reach a steady state and eliminate the influence of head motion; (2) slice timing; (3) frame displacement-based scrubbing, during which the frame displacements threshold for “bad” time points was set as 0.2 mm; (4) regressing out the physiologic signal through global signal regression; (5) spatial normalization onto the Montreal Neurological Institute space (resampling voxel size: 3 × 3 × 3 mm3); (6) removal of linear trends; (7) bandpass filtering (0.01–0.1 Hz); and (8) spatial smoothing (full width at half maximum of 4 mm Gaussian kernel). The results were normalized by calculating the Z-scores. Fig. 1A shows the preprocessing guidelines used in this study. In terms of quality control of fMRI, two subjects were excluded because of the inhomogeneity of the magnetic field, and one subject was excluded because of frame displacement was greater than the threshold (3.0 mm).

Fig. 1.

Fig. 1

Open in a new tab

Schematic outline of the study. A. Calculation steps of the amplitude of low-frequency fluctuations (ALFF). 3D, T1, and rs-fMRI were co-registered, and data were processed through the steps in the flowchart. The power within 0.01–0.1 Hz was extracted. The time series chart presented the calculation process of extracting power within the low-frequency bans. B. Inter-group comparison between patients with PD and controls. The results show five significant clusters. A correlation between significant clusters and clinical variables was conducted. C. Implementation of two predictive models; the regression and the classification models. The left column shows the ratio of training and testing datasets and the cross-validation method. Corr, correlation; SVM, support vector machine; GradBoost, gradient boosting; AdaBoost, adaptive boosting; CatBoost, categorical boosting; MAE, mean absolute error; AUC, area under the curve; ROC, receiver operating characteristic curve.

2.4. ALFF abnormality patterns and relevancies with clinical variables

A two-sample t-test was used to compare the ALFF of patients with PD with that of controls, with a two-tailed p-value of < 0.05, which was considered significant. The covariances included the motion parameters, age, sex, and preoperative levodopa dose. The correlations between the significant clusters derived from the above group comparison and the 11 clinical variables were then examined using partial correlations (Spearman's correlation). The clinical variables included age at surgery, disease duration, Hoehn-Yahr (H-Y) stage, preoperative LEDD, FOGQ, HAMA, HDRS, PDQ, MMSE, MoCA, and change ratio of UPDRS-III improvement.

2.5. Predictive models for motor improvement of DBS

Two machine-learning models were applied to predict the motor improvement of STN-DBS. The targeted UPDRS-III improvement scores were used as continuous values in the regression model and binary values in the classification model, which was the primary distinction between the two models.

2.5.1. Regression model

The ensemble method used in the regression model consisted of a two-layer structure. The first layer comprised six meta-models, whereas the second layer comprised the least squares regression model. The meta-models of the first layer included least squares linear regression, ridge regression, k-adjacency regression, adaptive boosting (AdaBoost) regression, gradient boosting (GradBoost) regression, and a linear support vector machine (SVM). The average probability values from the first layer were passed on to the second layer.

Cross-validation was used to evaluate predictive performance, with inner and outer 10-fold cross-validation (10F-CV). The feature engineering process was severely constrained inside the training set. The Z-score was calculated for the normalization of characteristics. The correlation coefficient between UPDRS-III improvement and each regressive factor was calculated and transformed to the F-value of analysis of variance (ANOVA), and finally to a p-value. Principal component analysis (PCA) was used for feature reduction. The model structure is shown in Fig. 1C.

2.5.2. Classification model

The patients were divided into two responsiveness subgroups (superior and moderate) based on a UPDRS-III score improvement of 50% (Deuschl et al., 2019). Patients were grouped into superior responders if improvement of UPDRS-III was greater than 50% and moderate responders if it was < 50%. The Category Boosting (CatBoost) algorithm was used to construct the predictive classification model, reduce over-fitting and enhance accuracy and generalizability. As mentioned in the previous section, the cross-validation, the feature selection, and the normalization steps were the same as those used in the regression model methods (Fig. 1C).

2.6. Correlation between clinical covariance and DBS response

The correlation between UPDRS-change and the four clinical factors was analyzed to avoid the potential influence of preoperative clinical factors on the predictive model. The four variables included age at surgery, H-Y stages, LEDD, and disease duration. A partial correlation (Spearman's test) was conducted, and the remaining factors were controlled as covariates.

2.7. Statistical analysis

Data with normal distribution were presented as mean ± SD, while data with skewed distribution were presented as median (range). An independent t-test was used for continuous variables, the chi-square test for binary variables, and the Mann-Whitney U test for ordinal variables in group comparisons of demographic data. A two-tailed, smaller p-value at p < 0.05 was considered significant. Multiple corrections were carried out to the inter-group comparisons of the fMRI results using the Gaussian random field (GRF) (p < 0.05). A p < 0.0045 (0.05/11) was considered significant after multiple Bonferroni corrections in the correlation analysis between clinical variables and significant ALFF clusters in PD. All statistical analyses were performed using SPSS version 23 software (IBM Corp., Armonk, NY, USA).

To estimate the regression model, Spearman's r between the predicted and actual values and the mean absolute error (MAE) were calculated. Using the mean and standard deviation of the 10-fold cross-validation (10F-CV), we performed receiver operating characteristics (ROC) analysis to calculate the specificity, sensitivity, accuracy, and area under the curve (AUC).

3. Results

3.1. Demographic data

The study included 25 moderate and 19 superior responders, and 44 healthy controls. The clinical characteristics of the patients presented are in Table 1. There was no significant difference in age or sex between the patients and healthy controls. The clinical variables in the two responsive groups did not differ significantly. The UPDRS-III scores of superior responders in the preoperative med-OFF state (57.47 ± 20.31) were higher than those of moderate responders (43.64 ± 17.85, p = 0.02). We used partial correlation rather than correlation in the correlation analysis to eliminate the potential impact of UPDRS differences in baseline UPDRS scores on UPDRS change ratios. There was no significant difference in head motion during the rs-fMRI between the PD and control groups(p = 0.39). The distance between the lead tip positions and intended target coordinates did not show a significant difference between the moderate and superior responders (Table S1) (p = 0.72 for the left and p = 0.20 for the right). DBS lead placement was roughly similar across the two groups with different UPDRS-III improvements (Fig S1).

Table 1.

Demographic characteristics.

PD patients Moderate Responders Superior Responders p-value
(Two PD groups)		 Healthy Control p-value (PD vs control)
Number 44 25 19 44
Female 17 9 8 0.68 19 0.66
Age at surgery 61.16 ± 10.00 61.64 ± 11.19 60.53 ± 8.42 0.71 60.39 ± 4.85 0.65
Disease duration 9.52 ± 4.36 9.60 ± 4.71 9.42 ± 3.98 0.89
Hoehn-Yahr stage 3.13 ± 0.70 3.14 ± 0.70 3.13 ± 0.72 0.62
UPDRS-III-ON 22.02 ± 11.02 19.40 ± 11.09 25.47 ± 10.18 0.07
UPDRS-III-OFF 49.61 ± 19.96 43.64 ± 17.85 57.47 ± 20.31 0.02
FOGQ 13.54 ± 5.08 14.12 ± 8.54 12.79 ± 9.21 0.63
HAMA 18.41 ± 9.39 17.56 ± 10.79 19.53 ± 7.27 0.48
HDRS 17.71 ± 8.28 16.88 ± 9.09 18.79 ± 7.16 0.44
PDQ-39 67.15 ± 26.33 65.28 ± 25.32 69.63 ± 28.10 0.25
MMSE 25.32 ± 5.08 26.28 + 3.36 24.05 ± 6.61 0.19
MoCa 21.86 ± 6.63 22.16 ± 6.49 21.47 ± 6.97 0.74
LEDD at DBS 761.01 ± 308.11 789.11 ± 320.68 724.02 ± 295.17 0.49
LEDD at 12 months 733.25 ± 412.25 810.22 ± 483.59 631.97 ± 274.10 0.13
Mean FD (mm) 0.16 ± 0.06 0.16 ± 0.08 0.16 ± 0.05 0.39
Open in a new tab

UPDRS, Unified Parkinson's Disease Rating Scale; FOG-Q, freezing of gait questionnaire; HAMA, Hamilton Anxiety Rating Scale; HDRS, Hamilton Depression Rating Scale; PDQ, Parkinson's disease questionnaire; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; LEDD, levodopa equivalent daily dose. Data are presented as mean ± standard deviation; FD, frame displacement.

3.2. PD abnormality patterns of ALFF and correlations with clinical variables

The PD group had five clusters with significant ALFF values (Fig. 2A). In detail, the PD had decreased ALFF values compared to controls within the bilateral cerebellum, left middle temporal gyrus, and right superior temporal gyrus, and increased ALFF within the bilateral primary motor area (Table 2). The H-Y stage was positively correlated with ALFF in the bilateral cerebellum (R = 0.348, p = 0.021), while UPDRS-III improvement was positively correlated with ALFF clusters within the left (R = 0.486, p < 0.001) or right (R = 0.417, p = 0.005) motor areas (Fig. 2B, 2C). Other significant clusters from the group comparison were not correlated with motor improvement, other clinical symptoms, or psychological scales.

Fig. 2.

Fig. 2

Open in a new tab

Altered ALFF in PD patients and its correlation with clinical variables. A. Representative t-map slices showing significant ALFF differences between PD patients and controls. Intergroup comparisons were conducted, followed by multiple corrections using the Gaussian random field (GRF) method (p < 0.05). Warm colour scheme: PD > control; cold colour scheme: PD < control. B. Correlation matrix between the five significant clusters and 11 clinical variables. A p < 0.005 (0.05/11) was considered significant with Bonferroni multiple corrections. C-E. A linear correlation diagram showing three significant correlation pairs between ALFF clusters and clinical variables. The x-axis indicates the average of normalized ALFF values for individuals. ALFF, the amplitude of low-frequency fluctuations; H-Y Stages, Hoehn-Yahr Stages; UPDRS, Unified Parkinson's Disease Rating Scale; FOG-Q, freezing of gait questionnaire, HAMA, Hamilton Anxiety Rating Scale; HDRS, Hamilton Depression Rating Scale; PDQ, Parkinson's disease questionnaire; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; LEDD levodopa equivalent daily dose; L, left; R, right; CBM, cerebellum; PCG, precentral gyrus.

Table 2.

Significant ALFF clusters comparing PD patients to controls.

Side Cluster Locations Voxel Number Coordinates T Value Of Peak
Cluster1 Left Cerebellar Tonsil 102 (-18,-42,-54) −3.779
Cluster2 Left Precentral Gyrus 170 (-48,-3,30) 4.041
Cluster3 Right Precentral Gyrus 1500 (27,-21,63) 4.802
Cluster4 Left Middle Temporal Gyrus 182 (-39,-63,6) −5.027
Cluster5 Right Superior Temporal Gyrus 90 (45,-57,0) −4.127
Open in a new tab

3.3. The regression and classification tasks in the predictive model

The regression model was assessed using Spearman's correlation coefficient and MAE. The predicted and observed UPDRS-III improvements significantly correlated in the ensemble model (Fig. 3B; r = 0.65, p < 0.01; MAE = 0.10 ± 0.03).

Fig. 3.

Fig. 3

Open in a new tab

Predictive model features and performance. A. Feature importance in the machine learning models. The map shows that the brain regions contributed to the classification model. The colour represents the weighted features, while only features with a weighted z-score higher than 0.6 are shown. B. The performance of the regression model. C. Density plot of the predictive distribution. D. The receiver operating characteristic (ROC) curve for the classification. E. The chart demonstrates measurements of the classification model.

The classification model had an accuracy of 88.0%, a specificity of 80.0%, and a sensitivity of 93.3% (Fig. 3E). Additionally, the classification model had an average AUC of 0.937 (Fig. 3D). The density plot indicated that the classification model could distinguish between superior and moderate responders (Fig. 3C). Regarding feature importance, the significant clusters involved the bilateral precentral gyrus, postcentral gyrus, temporal cortex, cuneus, precuneus, post-cingulate gyrus, and right thalamus. However, only the brain regions with feature importance larger than 0.6 are shown.

3.4. Correlation between clinical covariance and DBS response

Clinical variables and UPDRS-III scores did not show a statistically significant correlation (p < 0.05) (Fig. 4). As a result, we inferred that these clinical factors do not affect the performance of the predictive models, confirming the predictive value of the ALFF for DBS responses.

Fig. 4.

Fig. 4

Open in a new tab

Correlations between clinical variables and motor improvement of DBS. No clinical variable was significantly correlated with DBS response. A-D. The horizontal axis represents the improvement rate of UPDRS-III after DBS, and the vertical axis represents the clinical variables. Partial correlation (Spearman's) coefficients (r) and p-values are displayed. UPDRS, Unified Parkinson's Disease Rating Scale; H-Y Stages, Hoehn-Yahr Stages; LEDD, levodopa equivalent daily dose.

4. Discussion

We investigated altered ALFF in patients with PD and its relationships with clinical characteristics. Furthermore, we used a new approach to predict improvement in motor function resulting from DBS at a patient-specific level. Notably, the two predictive models confirmed the potential of fMRI as a biomarker in selecting participants for STN-DBS, thus demonstrating the significance of our study.

4.1. The altered ALFF patterns and their correlation with clinical variables

In a group comparison, patients with PD showed increased ALFF in the bilateral motor area and decreased ALFF in the bilateral temporal cortex and cerebellum. The motor area is a critical node in the basal ganglia-cortex-cerebellum circuit (Caligiore et al., 2016). Consistent with our results, a structural MRI study showed that patients with non-demented PD patients have cortical thinning in the sensorimotor cortex compared to healthy controls (Burciu and Vaillancourt, 2018). Furthermore, overactive BOLD in the primary motor cortex was associated with upper limb rigidity (Yu et al., 2007). Most patients in our cohort exhibited akinetic and rigidity symptoms, which may explain the altered ALFF in the primary motor cortex and cerebellum. In terms of specific components of the UPDRS-III that drove the significant correlations with ALFF, this study does not give a direct answer, because further subtype analysis was hindered by limited sample size. Previous literature has inconsistent findings for the correlation between fMRI signals and rigidity. For example, research suggested that connectivity between the cerebellum and multiple other regions could predict clinical rigidity scores (Baradaran et al., 2013), while another research showed that BOLD activation in the cerebellum was not correlated with rigidity (Yu et al., 2007).

We performed a correlation analysis for significant ALFF clusters and clinical variables to investigate the association between fMRI abnormalities and PD pathophysiology. Our results indicated a correlation between reduced ALFF in the bilateral cerebellum and disease severity, as determined by the H-Y stages. Similarly, cerebellar's connectivity with default mode networks is correlated with motor function severity (Palmer et al., 2020). Additionally, our findings showed a correlation between an increased ALFF in the bilateral motor cortex and improved motor function after DBS. In a previous study, decreased functional connectivity between the stimulation location and the motor area correlated with DBS efficacy (Horn et al., 2017). These findings suggested that the effect of DBS on PD may involve abnormal spontaneous activity, as revealed by fMRI within a large-scale brain network.

4.2. The prediction models for the motor outcome of STN-DBS

In the previous section, we identified abnormal ALFF patterns and their correlation with clinical variables, illustrating ALFF's potential to predict DBS's motor improvement. In this section, we describe two predictive models for the motor outcome of DBS. To evaluate the predictive performance of the machine learning models more robustly, we conducted both prediction and classification tasks using ALFF as continuous and binary values, respectively.

4.3. The prediction models

The prediction models showed a significant correlation between predicted and observed UPDRS-III improvement after DBS, with the correlation coefficient of the model of 0.65 and MAE of 0.10. The brain regions with feature importance of the model higher than 0.6 included the bilateral precentral and postcentral gyri, temporal cortex, precuneus, post-cingulate gyrus, cuneus, thalamus, and cerebellum. These regions are critical for the motor pathway in patients with PD (Albin et al., 1989). These results suggest that PD may represent a disease with disturbances in broad networks as opposed to local pathology alone. Consistent with previous findings, most regions play essential roles in PD symptoms and neurophysiological assessments (Burciu and Vaillancourt, 2018).

Disruption of the cortical-basal ganglia-thalamo-cortical neural network is crucial for the pathological mechanisms of PD (Burciu and Vaillancourt, 2018). The functional abnormality within the primary motor cortex has been illustrated in PD patients (Choe et al., 2013, Dirkx et al., 2016, Hu et al., 2015). These studies used regional homogeneity, ALFF, and voxel-mirrored homotopic connectivity measurements. Abnormal activity in motor areas may be implicated in the pathogenesis of PD symptoms, such as the onset of movements (Wu et al., 2011), resting tremors (Dirkx et al., 2016), and rigidity of movement (Hu et al., 2015). Our results suggest the predictive value of the spontaneous activity of the primary motor cortex for DBS outcomes, providing evidence for the critical role of the motor area in the pathophysiological substrates of PD. Indeed, STN-DBS was associated with increased BOLD activation in the primary motor cortex as demonstrated in some studies (Knight et al., 2015, Loh et al., 2022), while other studies demonstrated significantly decreased activation in this region (Boutet et al., 2021). How does fMRI signal in the primary motor area responds to STN stimulation is a crucial issue to explain the underlying mechanisms of DBS treatment, which is still unclear. The regression model used in this study is the ensemble method, which combines 6 meta-models, which means that the relationship between the ALFF value and clinical improvement is nonlinear. Therefore, it is not applicable to directly explain the improvement of motor symptoms by the increase or decrease of ALFF.

The temporal cortex may be involved in the abnormal neural networks in PD. A previous study reported a correlation between gait freezing and functional connectivity within the right occipitotemporal gyrus (Tessitore et al., 2012). Considering the high freezing of gait scores of the participants in this study, the higher feature importance in the bilateral temporal cortex is reasonable. The model features used in this study involved the bilateral occipital lobes. Accordingly, fMRI and structural MRI studies have suggested an association between the occipital cortex activity and the progression of PD (Fioravanti et al., 2015, Hou et al., 2018). The ALFF in the bilateral cerebellum contributed prominently to our predictive model. Patients with PD with postural instability and gait disorders (PIGD) consistently exhibit a lower ALFF in the posterior cerebellar lobe (Chen et al., 2015). Notably, the participants in our cohort showed more frequent PIGD symptoms than tremors.

A prominent feature of the predictive model is the thalamus. According to task-fMRI studies, the thalamus is activated when comparing off and on conditions (Kraft et al., 2009, Mueller et al., 2020). The bilateral precuneus cortex of the default mode (DMN) network is one of the regions with high feature importance in this study. Dopaminergic drugs can reverse the decrease in DMN network connectivity (Berman et al., 2016), explaining the contribution of ALFF within the DMN network to our predictive model. Based on these findings, it is plausible that our model shows that brain regions in a large-scale network have a predictive value for DBS responses.

4.4. Clinical variable correlation with STN-DBS response

We correlated DBS outcome with other variables to avoid potential effect of other variables on the predictive models. Previous studies have demonstrated the association between DBS motor improvement and clinical variables, including age and disease duration (Frizon et al., 2020, Habets et al., 2020, Su et al., 2017). In contrast, our findings showed no significant correlation between DBS-induced motor improvement and other clinical variables. The discrepancies between our results and other studies could be attributed to the selection criteria for the participants. Our study excluded participants with poor preoperative dopaminergic responses, and those with potential unsatisfactory surgical outcomes assessed by preoperative UPDRS.

4.5. Strengths and limitations

Although some studies have shown the predictive value of fMRI for the motor outcome of DBS (Horn et al., 2017, Li et al., 2019, Shang et al., 2020, Younce et al., 2021), this study has two main advantages. First, most other studies only implied a linear correlation between DBS outcome and fMRI features rather than predicting motor improvement at a patient-specific level. In contrast, the present study predicted DBS outcomes at a patient-specific level. Second, we implemented a regression model using continuous scores and a classification model using binary scores, providing a more robust and repetitive predictive system with higher clinical utility. Third, it is important to note that the test–retest reliability of ALFF potentially increases the potential value of ALFF to become a predictive marker. In detail, shorter TRs (323 ms) have previously been associated with higher ALFF reliability scores compared with longer TRs (2000 ms) (Cahart et al., 2022). Specifically, the TRs of this study were 750 ms.

Specific methodological considerations were considered for this study. Previous studies have reported that the efficacy of DBS is related to clinical factors such as levodopa response and electrode location (Cavallieri et al., 2021, Johnsen et al., 2010). Therefore, some studies have used clinical factors as predictors in their models. This study used only fMRI in the model for the following reasons: first, our result indicated that the DBS outcome was not affected by clinical factors such as age and disease duration. Second, the primary purpose of this study was to predict DBS outcomes using preoperative data. The electrode locations could only be acquired after surgery and thus were not included as predictors in this study. In addition, we excluded participants with inaccurate electrodes to eliminate the disturbance of electrode locations on the DBS outcome.

The limitations of this study include its relatively small sample size. To address this problem, we strictly constrained the feature engineering of the predictive model within the training set to avoid overfitting. Additionally, the cohorts were not divided into subgroups according to symptoms, which could have precluded further investigation with specific PD subtypes. The unbalanced composition of the PD subtypes also hindered subgroup analysis. Patients with severe tremors were rarely enrolled in this cohort because tremors can affect preoperative fMRI scans without medication. Due to the COVID-19 pandemic and related policies, not all patients undergoing STN-DBS attend the hospital postoperatively, resulting in incomplete long-term follow-up data.

In conclusion, our results demonstrated that disease severity and motor improvement in STN-DBS are related to the altered spontaneous neural activity of ALFF in patients with PD. Furthermore, we found that continuous and binary whole-brain ALFF maps could predict motor improvement in STN-DBS at the patient-specific level. With continued studies, the ALFF is expected to be a valuable biomarker for presurgical evaluation, especially for identifying DBS candidates.

5. Funding information

This work was supported by the National Natural Science Foundation of China (82071457 (Kai Zhang), 82271495 (Kai Zhang), 82201603 (Xiu Wang), 82201600 (Baotian Zhao)), and Capital's Funds for Health Improvement and Research (2022–1-1071 (Kai Zhang), 2020–2-1076 (Wenhan Hu)).

CRediT authorship contribution statement

Bowen Yang: Conceptualization, Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Xiu Wang: Data curation, Writing – review & editing. Jiajie Mo: Writing – review & editing. Zilin Li: Data curation, Writing – review & editing. Wenhan Hu: Funding acquisition. Chao Zhang: Funding acquisition. Baotian Zhao: Writing – review & editing. Dongmei Gao: Data curation. Xin Zhang: Data curation. Liangying Zou: Data curation. Xuemin Zhao: Data curation. Zhihao Guo: Writing – review & editing. Jianguo Zhang: Project administration. Kai Zhang: Conceptualization, Funding acquisition, Project administration, Supervision, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.nicl.2023.103430.

Contributor Information

Jianguo Zhang, Email: zjguo73@126.com.

Kai Zhang, Email: zhangkai62035@163.com.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Supplementary data 1
mmc1.docx (2MB, docx)

Data availability

Data will be made available on request.

References

  1. Albanese A., Bonuccelli U., Brefel C., Chaudhuri K.R., Colosimo C., Eichhorn T., Melamed E., Pollak P., Van Laar T., Zappia M. Consensus statement on the role of acute dopaminergic challenge in Parkinson's disease. Mov. Disord. 2001;16(2):197–201. doi: 10.1002/mds.1069. [DOI] [PubMed] [Google Scholar]
  2. Albin R.L., Young A.B., Penney J.B. The functional anatomy of basal ganglia disorders. Trends Neurosci. 1989;12(10):366–375. doi: 10.1016/0166-2236(89)90074-x. [DOI] [PubMed] [Google Scholar]
  3. Ballarini T., Mueller K., Albrecht F., Růžička F., Bezdicek O., Růžička E., Roth J., Vymazal J., Jech R., Schroeter M.L. Regional gray matter changes and age predict individual treatment response in Parkinson's disease. Neuroimage Clin. 2019;21 doi: 10.1016/j.nicl.2018.101636. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Baradaran N., Tan S.N., Liu A., Ashoori A., Palmer S.J., Wang Z.J., Oishi M.M.K., McKeown M.J. Parkinson's disease rigidity: relation to brain connectivity and motor performance. Front. Neurol. 2013;4 doi: 10.3389/fneur.2013.00067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Berman B.D., Smucny J., Wylie K.P., Shelton E., Kronberg E., Leehey M., Tregellas J.R. Levodopa modulates small-world architecture of functional brain networks in Parkinson's disease. Mov. Disord. 2016;31(11):1676–1684. doi: 10.1002/mds.26713. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Biswal B., Zerrin Yetkin F., Haughton V.M., Hyde J.S. Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magn. Reson. Med. 1995;34(4):537–541. doi: 10.1002/mrm.1910340409. [DOI] [PubMed] [Google Scholar]
  7. Boutet A., Madhavan R., Elias G.J.B., Joel S.E., Gramer R., Ranjan M., Paramanandam V., Xu D., Germann J., Loh A., Kalia S.K., Hodaie M., Li B., Prasad S., Coblentz A., Munhoz R.P., Ashe J., Kucharczyk W., Fasano A., Lozano A.M. Predicting optimal deep brain stimulation parameters for Parkinson's disease using functional MRI and machine learning. Nat. Commun. 2021;12(1) doi: 10.1038/s41467-021-23311-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Burciu R.G., Vaillancourt D.E. Imaging of Motor Cortex Physiology in Parkinson's Disease. Mov. Disord. 2018;33(11):1688–1699. doi: 10.1002/mds.102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Cahart M.-S., O'Daly O., Giampietro V., Timmers M., Streffer J., Einstein S., Zelaya F., Dell'Acqua F., Williams S.C.R. Comparing the test-retest reliability of resting-state functional magnetic resonance imaging metrics across single band and multiband acquisitions in the context of healthy aging. Hum. Brain Mapp. 2023;44(5):1901–1912. doi: 10.1002/hbm.26180. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Caligiore D., Helmich R.C., Hallett M., Moustafa A.A., Timmermann L., Toni I., Baldassarre G. Parkinson's disease as a system-level disorder. NPJ Parkinsons Dis. 2016;2(1) doi: 10.1038/npjparkd.2016.25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cavallieri F., Fraix V., Bove F., Mulas D., Tondelli M., Castrioto A., Krack P., Meoni S., Schmitt E., Lhommée E., Bichon A., Pélissier P., Chevrier E., Kistner A., Seigneuret E., Chabardès S., Moro E. Predictors of Long-Term Outcome of Subthalamic Stimulation in Parkinson Disease. Ann. Neurol. 2021;89(3):587–597. doi: 10.1002/ana.25994. [DOI] [PubMed] [Google Scholar]
  12. Chao-Gan Y., Yu-Feng Z. DPARSF: A MATLAB Toolbox for “Pipeline” Data Analysis of Resting-State fMRI. Front. Syst. Neurosci. 2010;4:13. doi: 10.3389/fnsys.2010.00013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Chen H.-M., Wang Z.-J., Fang J.-P., Gao L.-Y., Ma L.-Y., Wu T., Hou Y.-N., Zhang J.-R., Feng T. Different patterns of spontaneous brain activity between tremor-dominant and postural instability/gait difficulty subtypes of Parkinson's disease: a resting-state fMRI study. CNS Neurosci. Ther. 2015;21(10):855–866. doi: 10.1111/cns.12464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Choe I.-H., Yeo S., Chung K.-C., Kim S.-H., Lim S. Decreased and increased cerebral regional homogeneity in early Parkinson's disease. Brain Res. 2013;1527:230–237. doi: 10.1016/j.brainres.2013.06.027. [DOI] [PubMed] [Google Scholar]
  15. Deuschl G., et al. A randomized trial of deep-brain stimulation for Parkinson's disease. N. Engl. J. Med. 2006;355:896–908. doi: 10.1056/NEJMoa060281. [DOI] [PubMed] [Google Scholar]
  16. Deuschl G., Follett K.A., Luo P., Rau J., Weaver F.M., Paschen S., Steigerwald F., Tonder L., Stoker V., Reda D.J. Comparing two randomized deep brain stimulation trials for Parkinson's disease. J. Neurosurg. 2020;132(5):1376–1384. doi: 10.3171/2018.12.JNS182042. [DOI] [PubMed] [Google Scholar]
  17. Dirkx M.F., den Ouden H., Aarts E., Timmer M., Bloem B.R., Toni I., Helmich R.C. The Cerebral Network of Parkinson's Tremor: An Effective Connectivity fMRI Study. J. Neurosci. 2016;36(19):5362–5372. doi: 10.1523/JNEUROSCI.3634-15.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Ewert S., Plettig P., Li N., Chakravarty M.M., Collins D.L., Herrington T.M., Kühn A.A., Horn A. Toward defining deep brain stimulation targets in MNI space: A subcortical atlas based on multimodal MRI, histology and structural connectivity. Neuroimage. 2018;170:271–282. doi: 10.1016/j.neuroimage.2017.05.015. [DOI] [PubMed] [Google Scholar]
  19. Fan S.-Y., Wang K.-L., Hu W., Eisinger R.S., Han A., Han C.-L., Wang Q., Michitomo S., Zhang J.-G., Wang F., Ramirez‐Zamora A., Meng F.-G. Pallidal versus subthalamic nucleus deep brain stimulation for levodopa-induced dyskinesia. Ann. Clin. Transl. Neurol. 2020;7(1):59–68. doi: 10.1002/acn3.50961. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Fasano A., Daniele A., Albanese A. Treatment of motor and non-motor features of Parkinson's disease with deep brain stimulation. Lancet Neurol. 2012;11(5):429–442. doi: 10.1016/S1474-4422(12)70049-2. [DOI] [PubMed] [Google Scholar]
  21. Fasano A., Lozano A.M. Deep brain stimulation for movement disorders: 2015 and beyond. Curr. Opin. Neurol. 2015;28:423–436. doi: 10.1097/WCO.0000000000000226. [DOI] [PubMed] [Google Scholar]
  22. Feigin A., Kaplitt M.G., Tang C., Lin T., Mattis P., Dhawan V., During M.J., Eidelberg D. Modulation of metabolic brain networks after subthalamic gene therapy for Parkinson's disease. PNAS. 2007;104(49):19559–19564. doi: 10.1073/pnas.0706006104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. V. Fioravanti F. Benuzzi L. Codeluppi S. Contardi F. Cavallieri P. Nichelli F. Valzania MRI correlates of Parkinson's disease progression: a voxel based morphometry study Parkinsons Dis 2015 2015 1 8 378032. [DOI] [PMC free article] [PubMed]
  24. Follett K.A., Weaver F.M., Stern M., Hur K., Harris C.L., Luo P., Marks W.J., Rothlind J., Sagher O., Moy C., Pahwa R., Burchiel K., Hogarth P., Lai E.C., Duda J.E., Holloway K., Samii A., Horn S., Bronstein J.M., Stoner G., Starr P.A., Simpson R., Baltuch G., De Salles A., Huang G.D., Reda D.J. Pallidal versus subthalamic deep-brain stimulation for Parkinson's disease. N. Engl. J. Med. 2010;362(22):2077–2091. doi: 10.1056/NEJMoa0907083. [DOI] [PubMed] [Google Scholar]
  25. Frizon L.A., Gopalakrishnan R., Hogue O., Floden D., Nagel S.J., Baker K.B., Isolan G.R., Stefani M.A., Machado A.G. Cortical thickness in visuo-motor areas is related to motor outcomes after STN DBS for Parkinson's disease. Parkinsonism Relat. Disord. 2020;71:17–22. doi: 10.1016/j.parkreldis.2020.01.006. [DOI] [PubMed] [Google Scholar]
  26. Goetz C.G., Tilley B.C., Shaftman S.R., Stebbins G.T., Fahn S., Martinez-Martin P., Poewe W., Sampaio C., Stern M.B., Dodel R., Dubois B., Holloway R., Jankovic J., Kulisevsky J., Lang A.E., Lees A., Leurgans S., LeWitt P.A., Nyenhuis D., Olanow C.W., Rascol O., Schrag A., Teresi J.A., van Hilten J.J., LaPelle N. Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov. Disord. 2008;23(15):2129–2170. doi: 10.1002/mds.22340. [DOI] [PubMed] [Google Scholar]
  27. Habets J.G.V., Janssen M.L.F., Duits A.A., Sijben L.C.J., Mulders A.E.P., De Greef B., Temel Y., Kuijf M.L., Kubben P.L., Herff C. Machine learning prediction of motor response after deep brain stimulation in Parkinson's disease-proof of principle in a retrospective cohort. PeerJ. 2020;8 doi: 10.7717/peerj.10317. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Horn A., Reich M., Vorwerk J., Li N., Wenzel G., Fang Q., Schmitz-Hübsch T., Nickl R., Kupsch A., Volkmann J., Kühn A.A., Fox M.D. Connectivity Predicts deep brain stimulation outcome in Parkinson disease. Ann. Neurol. 2017;82(1):67–78. doi: 10.1002/ana.24974. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Hou Y., et al. Impaired topographic organization in cognitively unimpaired drug-naive patients with rigidity-dominant Parkinson's disease. Parkinsonism Relat. Disord. 2018;56:52–57. doi: 10.1016/j.parkreldis.2018.06.021. [DOI] [PubMed] [Google Scholar]
  30. Hu X., Zhang J., Jiang X., Zhou C., Wei L., Yin X., Wu Y.a., Li J., Zhang Y., Wang J. Decreased interhemispheric functional connectivity in subtypes of Parkinson's disease. J. Neurol. 2015;262(3):760–767. doi: 10.1007/s00415-014-7627-x. [DOI] [PubMed] [Google Scholar]
  31. Hughes A.J., Daniel S.E., Kilford L., Lees A.J. Accuracy of clinical diagnosis of idiopathic Parkinson's disease: a clinico-pathological study of 100 cases. J. Neurol. Neurosurg. Psychiatry. 1992;55(3):181–184. doi: 10.1136/jnnp.55.3.181. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Johnsen E.L., Sunde N., Mogensen P.H., Østergaard K. MRI verified STN stimulation site–gait improvement and clinical outcome. Eur. J. Neurol. 2010;17(5):746–753. doi: 10.1111/j.1468-1331.2010.02962.x. [DOI] [PubMed] [Google Scholar]
  33. Knight E.J., Testini P., Min H.-K., Gibson W.S., Gorny K.R., Favazza C.P., Felmlee J.P., Kim I., Welker K.M., Clayton D.A., Klassen B.T., Chang S.-Y., Lee K.H. Motor and Nonmotor Circuitry Activation Induced by Subthalamic Nucleus Deep Brain Stimulation in Patients With Parkinson Disease: Intraoperative Functional Magnetic Resonance Imaging for Deep Brain Stimulation. Mayo Clin. Proc. 2015;90(6):773–785. doi: 10.1016/j.mayocp.2015.03.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kraft E., Loichinger W., Diepers M., Lule D., Schwarz J., Ludolph A.C., Storch A. Levodopa-induced striatal activation in Parkinson's disease: a functional MRI study. Parkinsonism Relat. Disord. 2009;15(8):558–563. doi: 10.1016/j.parkreldis.2009.02.005. [DOI] [PubMed] [Google Scholar]
  35. Küblböck M., Woletz M., Höflich A., Sladky R., Kranz G.S., Hoffmann A., Lanzenberger R., Windischberger C. Stability of low-frequency fluctuation amplitudes in prolonged resting-state fMRI. Neuroimage. 2014;103:249–257. doi: 10.1016/j.neuroimage.2014.09.038. [DOI] [PubMed] [Google Scholar]
  36. Li X., Xiong Y., Liu S., Zhou R., Hu Z., Tong Y., He L.e., Niu Z., Ma Y.u., Guo H. Predicting the Post-therapy Severity Level (UPDRS-III) of Patients With Parkinson's Disease After Drug Therapy by Using the Dynamic Connectivity Efficiency of fMRI. Front. Neurol. 2019;10 doi: 10.3389/fneur.2019.00668. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Loh A., Gwun D., Chow C.T., Boutet A., Tasserie J., Germann J., Santyr B., Elias G., Yamamoto K., Sarica C., Vetkas A., Zemmar A., Madhavan R., Fasano A., Lozano A.M. Probing responses to deep brain stimulation with functional magnetic resonance imaging. Brain Stimul. 2022;15(3):683–694. doi: 10.1016/j.brs.2022.03.009. [DOI] [PubMed] [Google Scholar]
  38. H. Lv Z. Wang E. Tong L.M. Williams G. Zaharchuk M. Zeineh A.N. Goldstein-Piekarski T.M. Ball C. Liao M. Wintermark Resting-State Functional MRI: Everything That Nonexperts Have Always Wanted to Know. [DOI] [PMC free article] [PubMed]
  39. Mosley P.E., Paliwal S., Robinson K., Coyne T., Silburn P., Tittgemeyer M., Stephan K.E., Perry A., Breakspear M. The structural connectivity of subthalamic deep brain stimulation correlates with impulsivity in Parkinson's disease. Brain. 2020;143(7):2235–2254. doi: 10.1093/brain/awaa148. [DOI] [PubMed] [Google Scholar]
  40. K. Mueller D. Urgošík T. Ballarini Š. Holiga H.E. Möller F. Růžička J. Roth J. Vymazal M.L. Schroeter E. Růžička R. Jech Differential effects of deep brain stimulation and levodopa on brain activity in Parkinson's disease Brain Commun 2 1 2020 2020. [DOI] [PMC free article] [PubMed]
  41. Palmer W.C., Cholerton B.A., Zabetian C.P., Montine T.J., Grabowski T.J., Rane S. Resting-State Cerebello-Cortical Dysfunction in Parkinson's Disease. Front. Neurol. 2020;11 doi: 10.3389/fneur.2020.594213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Shang R., He L.e., Ma X., Ma Y.u., Li X. Connectome-Based Model Predicts Deep Brain Stimulation Outcome in Parkinson's Disease. Front. Comput. Neurosci. 2020;14 doi: 10.3389/fncom.2020.571527. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Su X.-L., Luo X.-G., Lv H., Wang J., Ren Y., He Z.-Y. Factors predicting the instant effect of motor function after subthalamic nucleus deep brain stimulation in Parkinson's disease. Transl Neurodegener. 2017;6(1) doi: 10.1186/s40035-017-0084-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Tessitore A., Amboni M., Esposito F., Russo A., Picillo M., Marcuccio L., Pellecchia M.T., Vitale C., Cirillo M., Tedeschi G., Barone P. Resting-state brain connectivity in patients with Parkinson's disease and freezing of gait. Parkinsonism Relat. Disord. 2012;18(6):781–787. doi: 10.1016/j.parkreldis.2012.03.018. [DOI] [PubMed] [Google Scholar]
  45. Weinkle L.J., Hoyt B., Thompson J.A., Sillau S., Tanabe J., Honce J., Klepitskaya O. Association of MRI Measurements with Cognitive Outcomes After STN-DBS in Parkinson's Disease. Mov Disord Clin Pract. 2018;5(4):417–426. doi: 10.1002/mdc3.12643. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Wong J.K., Middlebrooks E.H., Grewal S.S., Almeida L., Hess C.W., Okun M.S. A Comprehensive Review of Brain Connectomics and Imaging to Improve Deep Brain Stimulation Outcomes. Mov. Disord. 2020;35(5):741–751. doi: 10.1002/mds.28045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Wu T., Long X., Wang L., Hallett M., Zang Y., Li K., Chan P. Functional connectivity of cortical motor areas in the resting state in Parkinson's disease. Hum. Brain Mapp. 2011;32(9):1443–1457. doi: 10.1002/hbm.21118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Wurina, Zang Y.-F., Zhao S.-G. Resting-state fMRI studies in epilepsy. Neurosci. Bull. 2012;28(4):449–455. doi: 10.1007/s12264-012-1255-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Yang B., Wang X., Mo J., Li Z., Gao D., Bai Y., Zou L., Zhang X., Zhao X., Wang Y., Liu C., Zhao B., Guo Z., Zhang C., Hu W., Zhang J., Zhang K. The amplitude of low-frequency fluctuation predicts levodopa treatment response in patients with Parkinson's disease. Parkinsonism Relat. Disord. 2021;92:26–32. doi: 10.1016/j.parkreldis.2021.10.003. [DOI] [PubMed] [Google Scholar]
  50. Younce J.R., et al. Resting-State Functional Connectivity Predicts STN DBS Clinical Response. Mov. Disord. 2021;36:662–671. doi: 10.1002/mds.28376. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Yu H., Sternad D., Corcos D.M., Vaillancourt D.E. Role of hyperactive cerebellum and motor cortex in Parkinson's disease. Neuroimage. 2007;35(1):222–233. doi: 10.1016/j.neuroimage.2006.11.047. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Yu-Feng Z., Yong H.e., Chao-Zhe Z., Qing-Jiu C., Man-Qiu S., Meng L., Li-Xia T., Tian-Zi J., Yu-Feng W. Altered baseline brain activity in children with ADHD revealed by resting-state functional MRI. Brain and Development. 2007;29(2):83–91. doi: 10.1016/j.braindev.2006.07.002. [DOI] [PubMed] [Google Scholar]
  53. Zuo X.N., Xing X.X. Test-retest reliabilities of resting-state FMRI measurements in human brain functional connectomics: a systems neuroscience perspective. Neurosci. Biobehav. Rev. 2014;45:100–118. doi: 10.1016/j.neubiorev.2014.05.009. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary data 1
mmc1.docx (2MB, docx)

Data Availability Statement

Data will be made available on request.

Articles from NeuroImage : Clinical are provided here courtesy of Elsevier

RESOURCES