TY - JOUR
AU - Kühn, Andrea, A
AB - Abstract Neuroimaging has seen a paradigm shift away from a formal description of local activity patterns towards studying distributed brain networks. The recently defined framework of the ‘human connectome’ enables global analysis of parts of the brain and their interconnections. Deep brain stimulation (DBS) is an invasive therapy for patients with severe movement disorders aiming to retune abnormal brain network activity by local high frequency stimulation of the basal ganglia. Beyond clinical utility, DBS represents a powerful research platform to study functional connectomics and the modulation of distributed brain networks in the human brain. We acquired resting-state functional MRI in 20 patients with Parkinson’s disease with subthalamic DBS switched on and off. An age-matched control cohort of 15 subjects was acquired from an open data repository. DBS lead placement in the subthalamic nucleus was localized using a state-of-the art pipeline that involved brain shift correction, multispectral image registration and use of a precise subcortical atlas. Based on a realistic 3D model of the electrode and surrounding anatomy, the amount of local impact of DBS was estimated using a finite element method approach. On a global level, average connectivity increases and decreases throughout the brain were estimated by contrasting on and off DBS scans on a voxel-wise graph comprising eight thousand nodes. Local impact of DBS on the motor subthalamic nucleus explained half the variance in global connectivity increases within the motor network (R = 0.711, P < 0.001). Moreover, local impact of DBS on the motor subthalamic nucleus could explain the degree to how much voxel-wise average brain connectivity normalized towards healthy controls (R = 0.713, P < 0.001). Finally, a network-based statistics analysis revealed that DBS attenuated specific couplings known to be pathological in Parkinson’s disease. Namely, coupling between motor thalamus and motor cortex was increased while striatal coupling with cerebellum, external pallidum and subthalamic nucleus was decreased by DBS. Our results show that resting state functional MRI may be acquired in DBS on and off conditions on clinical MRI hardware and that data are useful to gain additional insight into how DBS modulates the functional connectome of the human brain. We demonstrate that effective DBS increases overall connectivity in the motor network, normalizes the network profile towards healthy controls and specifically strengthens thalamo-cortical connectivity while reducing striatal control over basal ganglia and cerebellar structures. Parkinson's disease, deep brain stimulation, network modulation, resting-state functional MRI, brain connectivity Introduction Deep brain stimulation (DBS) is a highly efficacious and established treatment option for Parkinson’s disease (Deuschl et al., 2006) and a multitude of mechanisms of action have been proposed over the years (Lozano and Lipsman, 2013). Increasingly, these undergo a paradigm shift away from localized stimulation of specific areas towards global modulation of distributed brain networks (McIntyre and Hahn, 2010; Litvak et al., 2011; Vanegas Arroyave et al., 2016; Akram et al., 2017; Horn et al., 2017). To elucidate such network effects of DBS, modern-day neuroimaging methods become more and more useful (Horn, 2019). In addition, DBS could reversely be a powerful tool to study network changes as a function of precisely targeted stimuli. One of the few neuroimaging options to study the functional organization of the brain is functional MRI (fMRI). Intrinsic associations between subparts of the brain may be estimated using resting-state (rs-) fMRI and in this way, the ‘functional connectome’ of the brain can be explored (Biswal et al., 2010). When blood oxygenated level-dependent (BOLD) signals of two brain regions are correlated over time, these have been called functionally ‘connected’ in the literature (Fox et al., 2005; Friston, 2011), although this measure includes highly indirect connections (Varoquaux and Craddock, 2013). Until recently, it was not straightforward to acquire rs-fMRI data in patients with DBS implants, let alone with the stimulator switched on in the scanner. The reason was that no official certificate of device manufacturers allowed this practice and only limited pioneering work by a few specialized centres—the Jech and Foltynie groups should be mentioned among others—investigated changes of fMRI data under DBS in a so far limited fashion (Jech et al., 2001; Kahan et al., 2014) (Supplementary Table 1). In a first study of four patients, Jech et al. (2001) showed that the BOLD signals increase in ipsilateral subcortical structures under DBS. In a case report, Stefurak et al. (2003) then showed more distributed signal increases in (pre-) motor cortices, ventrolateral thalamus, putamen and cerebellum under effective DBS. Seminal work by Kahan et al. in 2014 described changes of direct, indirect and hyperdirect pathways of the basal ganglia—cortical loops under DBS using dynamic causal modelling. The same data were used to fit a computational mean-field model that was able to identify additional potential DBS targets beyond the classical subthalamic nucleus (STN) target used to treat Parkinson’s disease (Saenger et al., 2017). In a formal literature analysis, we identified further studies that used fMRI under active DBS in humans and animal models so far (Supplementary Table 1). In summary, STN-DBS in Parkinson’s disease may lead to increased overall functional connectivity in the premotor cortex (Mueller et al., 2013) and strengthened cortico-striatal and thalamo-cortical pathways in fMRI (Kahan et al., 2014). In sum, even though there was no official allowance for active DBS in fMRI until recently (see below), the concept was explored in small case numbers. One issue that has been neglected in prior studies is that slight changes of millimetres in lead placement may result in large differences in clinical improvement (Horn et al., 2019) and similarly, slight differences in connectivity profiles of DBS electrodes may be used to predict clinical improvement across patients, cohorts and DBS centres (Horn et al., 2017). Finally, small variations in lead placement can even explain changes of behaviour in motor (Neumann et al., 2018) and cognitive (Irmen et al., 2018) domains. Thus, we argue that it is crucial to incorporate DBS lead placement into an analysis of how they impact distributed brain networks. Instead, prior studies characterized fMRI changes in DBS on versus off contrasts and, by doing so, implicitly assumed that the DBS effect was equal in each patient. In reality, the impact of each DBS electrode on the target structure largely varies across patients and may be used as a regressor to better explain network changes. In the present study, we investigated a cohort of 20 patients with Parkinson’s disease at rest under DBS on and off conditions. We characterized changes in average connectivity (i.e. centrality) of brain regions and laid special focus on network changes as a function of the degree of motor STN-DBS modulation. Based on minor differences in DBS electrode placement, different amounts of motor STN volume were stimulated in each patient. As a result, we expected correspondingly differing changes in motor cortical activation that should be stronger or weaker as a function of electrode placement. Moreover, because differences in therapeutic effects are equally dependent on lead placement (Horn et al., 2017, 2019), we expected network properties to normalize towards healthy controls to a degree that is dependent on lead placement. We expected that an optimally placed lead would result in strong modulations in the motor network, normalizing towards the network properties found in healthy controls. In contrast, poorly placed leads would not result in strong motor network changes. Materials and methods Literature analysis To summarize previous studies in the still novel field of DBS-fMRI, a systematic medline literature analysis was conducted using the search-string (‘Deep brain stimulation’ OR ‘DBS’) AND (‘Parkinson’s disease’ OR ‘Parkinson’ OR ‘Parkinsonian’ OR ‘thalam*’ OR ‘pallid*’ OR ‘subthalamic’) AND (‘functional connectivity’ OR ‘resting-state fMRI’ OR ‘fMRI’). This resulted in 64 studies on 31 August 2018. Predefined inclusion criteria for studies were: (i) investigation of effects of DBS on subcortical targets such as thalamus, pallidum or subthalamic nucleus in Parkinson’s disease patients or (potentially healthy) animals using fMRI; and (ii) bilateral/unilateral stimulation of DBS targets during fMRI scan. Exclusion criteria were: (i) studies reporting safety of fMRI in DBS patients or animals without reporting structures affected by DBS; (ii) rs-fMRI studies in healthy humans; (iii) studies assessing medication effects without DBS; and (iv) meta-analyses and reviews. Relevant references were identified by title and abstract (n = 17) and judged further based on inclusion and exclusion criteria, originality, quality and actuality of the study resulting in n = 15 remaining papers. Another seven publications were added by screening reference lists of included articles. Results are summarized in Supplementary Table 1. Patients and imaging DBS cohort Twenty patients suffering from Parkinson’s disease who underwent DBS with target STN were included in the present study. Inclusion criteria were: at least >3 months of active STN-DBS, a magnetic resonance-conditional DBS system (in agreement with the Medtronic whole-body CE certificate; Medtronic, 2015, see below), no metal implants beside the DBS system, no largely predominant affective or cognitive non-motor symptoms of the disease, and no excessive tremor at rest that would make scanning impossible. The study was carried out in accordance with the Declaration of Helsinki and was approved by the internal review board of Charité – Universitätsmedizin Berlin (master vote #EA2/138/15). All patients had undergone DBS surgery for idiopathic Parkinson’s disease between April 2010 and April 2016 and received two quadripolar DBS electrodes (model 3389; Medtronic). Before surgery, patients received preoperative MRI and neuropsychological testing to exclude structural or psychiatric comorbidities. During surgery, microelectrode recordings were performed to verify lead placement. Clinical variables, including age, sex, l-DOPA response and l-DOPA equivalent dose (LEDD) at baseline were recorded. Postoperative improvement scores were recorded 12 months after surgery under medication OFF and DBS on/off conditions using clinical DBS stimulation parameters. After multiple studies had shown fMRI acquisition with implanted DBS devices to be safe (Jech et al., 2001; Carmichael et al., 2007; Sutton et al., 2008), and elaborate modelling and animal testing by Medtronic, their Activa® portfolio received an extended magnetic resonance conditional CE certificate for full-body MRI in 2015 (FDA approval followed in 2016). Under this extended certificate, it became officially feasible to acquire fMRI data under active DBS in humans (Medtronic, 2015). Limits, such as a maximal scanning time of 30 min and a B1+RMS value below 2.0 μT apply, with stimulation only allowed using bipolar settings. Thus, postoperatively, patients were scanned in a Siemens Magnetom Aera 1.5 T MRI. After a structural scan, an rs-fMRI scan of ∼9 min was carried out in the DBS on condition. Patients were then briefly taken out of the scanner to turn the impulse generator off. An experienced neurologist (G.W. or J.H.) waited for symptoms to reappear. After that, the same scan was repeated (DBS off condition). Scan parameters were as follows: T1 MP-RAGE: voxel size 1 × 1 × 1 mm, repetition time 2200 ms, echo time 2.63 ms. Rs-fMRI EPI scan: voxel size 3 × 3 × 3 mm, 24 slices with distance factor 30%, A > > P phase encoding, repetition time 2690 ms, echo time 40 ms, field of view readout 200 mm. Two hundred and ten volumes were acquired for each condition (total scan length of rs-fMRI scans 2 × 9.42 min). A brief diffusion MRI scan was acquired before the rs-fMRI scans but was not used in the present study. Total scan time was close to but below 30 min and the B1+RMS value was kept below 2.0 μT at all times (conforming to the magnetic resonance-conditional regulations of the Medtronic Activa CE-certificate). Normative control subjects Structural and rs-fMRI data from 15 age-matched healthy control subjects were obtained from the Parkinson’s Disease Progression Marker Initiative (PPMI; ppmi-info.org; Marek et al., 2011) database. These were processed in the same way as patients of the present study using Lead connectome software (Horn et al., 2014; Horn and Blankenburg, 2016) and had been made available in form of a normative group connectome in (Horn et al., 2017). Thus, details in acquisition and processing scheme are described elsewhere (Marek et al., 2011; Horn et al., 2017). Electrode localization and modelling the volume of tissue activated All data were processed in Lead-DBS software (www.lead-dbs.org; Horn and Kühn, 2015) using the enhanced default workflow of version 2.1.7 described in Horn et al. (2019). Within the pipeline, Lead-DBS uses a multitude of open source tools (see below and Horn et al., 2019 for details). Briefly, all (pre- and postoperative) MRIs were linearly co-registered to the preoperative T1 anchor modality using SPM (https://www.fil.ion.ucl.ac.uk/spm/software/spm12/; Friston et al., 2004). Registration between post- and preoperative T1 was further refined using the ‘brain shift correction’ module in Lead-DBS, which focuses on the subcortical target region of interest (Schönecker et al., 2009) and thus minimizes non-linear bias introduced when opening the skull during surgery. Data were normalized into standard space (ICBM 2009b NLIN ASYM; Fonov et al., 2009; henceforth referred to as ‘MNI space’) in a multispectral fashion (i.e. using T1- and T2-weighted MRIs) in parallel. This was done with the symmetric diffeomorphic registration algorithm implemented in Advanced Normalization Tools (http://stnava.github.io/ANTs/; Avants et al., 2008) using the ‘effective (low variance)’ preset with subcortical refinement as implemented in Lead-DBS (Horn et al., 2019). This exact normalization method was recently evaluated best-performing in a large comparative study (Ewert et al., 2019) and STN definitions obtained by the method rivalled those of manual expert segmentations in precision. Electrode trajectories and contacts were automatically pre-localized and manually refined using Lead-DBS. Based on the (bipolar) stimulation settings active in the scanner, the electric fields around the electrode were modelled using a finite element approach defined in Horn et al. (2017, 2019). Briefly, an adapted version of the FieldTrip/SimBio pipeline that is part of Lead-DBS (http://www.fieldtriptoolbox.org/; https://www.mrt.uni-jena.de/simbio/; Vorwerk et al., 2018) was used which uses a four-compartment anatomical model of the electrodes and surrounding tissue. All analyses were performed using the impact of the electric field on the STN but were repeated with a binarized version of it [in terms of a volume of tissue activated (VTA), which is a more commonly applied model in the field]. To approximate the VTA, the electric field gradient magnitude was thresholded at a heuristic value of 0.2 V/mm. This value has been frequently been used in similar context (Åström et al., 2009, 2015; Vasques et al., 2009; Horn et al., 2017). Estimating the impact of DBS on motor STN As mentioned above, in the DBS context, the VTA has most often been modelled in binary form (McIntyre and Grill, 2002; Butson et al., 2007; Åström et al., 2015; Dembek et al., 2017; Horn et al., 2017). However, a recent study showed superior results when directly using the electric field magnitude (instead of a binarized version; Horn et al., 2019). Building upon this, here, we estimated the factor ‘impact of DBS on the motor STN’ by multiplying the (unthresholded) electric field gradient magnitude map with the atlas-defined mask of the motor STN (Ewert et al., 2018) and summed up voxels of the resulting image. Intuitively, this grasps the amount of voltage gradient that is present within the motor STN. Of note, all results reported in the present study would hold true with similar (and significant) effect sizes when repeating analyses using a binarized VTA, instead. Definition of a (sensori-)motor network parcellation To estimate changes in the motor network of key regions within the basal ganglia–cerebellar–cortical loop, sensorimotor functional zones of cortex, striatum, thalamus, globus pallidus internal/external (GPi/GPe), substantia nigra and cerebellum were defined based on multiple brain atlases available to the authors. Table 1 summarizes which atlas components were used for this parcellation. When multiple atlases defined a parcel, definitions were summed up to include voxels defined by any of the atlases available. Finally, for each component, parcels were masked by a definition of the whole structure (e.g. the motor thalamus parcel was masked by a thalamic mask), which is also denoted in Table 1. Code that generates the final atlas is available online (www.github.com/leaddbs/leaddbs). Throughout this work, we refer to this network as ‘motor’ network, but it does include premotor and sensory domains. Table 1 Atlases that contributed to motor network definition Structure/Atlas Name Atlas reference(s) Atlas available in Lead-DBS Parcels used Cortex Automated Anatomical Labeling Atlas Tzourio-Mazoyer et al., 2002 + 1–2, 19–20, 57–58 AICHA Atlas Joliot et al., 2015 + 51–62, 71–76, 221–226 Hammersmith Atlas n30r83 Hammers et al., 2003; Gousias et al., 2008 + 50–51, 60–61 Harvard Oxford Thr25 2 mm Makris et al., 2006 + 13–14, 33–34, 51–52 JueBrain atlas thr25 2 mm Eickhoff et al., 2005 + 47–66, 91–92 Mindboggle 101, Desikan protocol Klein et al., 2012 + 1022, 2022, 1024, 2024 HMAT Atlas Mayka et al., 2006 + 1–12 Masked by GM component of ICBM 2009b NLIN space (Lead-DBS) Fonov et al., 2009 + – Striatum Striatal fMRI Networks 7 Choi et al., 2012 + Somatomotor Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by CIT168_Reinf_Learning Atlas Pauli et al., 2017 + Ca, Pu, NAc Thalamus Morel Atlas Krauth et al., 2010; Jakab et al., 2012; Morel et al., 2013 – VLa Thalamic fMRI Networks 7 Choi et al., 2012 – Somatomotor Masked by Morel Atlas Krauth et al., 2010; Jakab et al., 2012; Morel et al., 2013 – All thalamic components GPi DISTAL Atlas Ewert et al., 2018 + GPi_sensorimotor Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by DISTAL Atlas Ewert et al., 2018 + GPi GPe Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by DISTAL Atlas Ewert et al., 2018 + GPe Substantia nigra Nigral organization atlas Zhang et al., 2017 + Lateral_SNc_somatomotor Masked by DISTAL Atlas Ewert et al., 2018 + SN Cerebellum Cerebellar fMRI Networks 7 Buckner et al., 2011 + 2 Masked by SUIT Atlas Diedrichsen et al., 2006 + 1–34 Structure/Atlas Name Atlas reference(s) Atlas available in Lead-DBS Parcels used Cortex Automated Anatomical Labeling Atlas Tzourio-Mazoyer et al., 2002 + 1–2, 19–20, 57–58 AICHA Atlas Joliot et al., 2015 + 51–62, 71–76, 221–226 Hammersmith Atlas n30r83 Hammers et al., 2003; Gousias et al., 2008 + 50–51, 60–61 Harvard Oxford Thr25 2 mm Makris et al., 2006 + 13–14, 33–34, 51–52 JueBrain atlas thr25 2 mm Eickhoff et al., 2005 + 47–66, 91–92 Mindboggle 101, Desikan protocol Klein et al., 2012 + 1022, 2022, 1024, 2024 HMAT Atlas Mayka et al., 2006 + 1–12 Masked by GM component of ICBM 2009b NLIN space (Lead-DBS) Fonov et al., 2009 + – Striatum Striatal fMRI Networks 7 Choi et al., 2012 + Somatomotor Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by CIT168_Reinf_Learning Atlas Pauli et al., 2017 + Ca, Pu, NAc Thalamus Morel Atlas Krauth et al., 2010; Jakab et al., 2012; Morel et al., 2013 – VLa Thalamic fMRI Networks 7 Choi et al., 2012 – Somatomotor Masked by Morel Atlas Krauth et al., 2010; Jakab et al., 2012; Morel et al., 2013 – All thalamic components GPi DISTAL Atlas Ewert et al., 2018 + GPi_sensorimotor Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by DISTAL Atlas Ewert et al., 2018 + GPi GPe Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by DISTAL Atlas Ewert et al., 2018 + GPe Substantia nigra Nigral organization atlas Zhang et al., 2017 + Lateral_SNc_somatomotor Masked by DISTAL Atlas Ewert et al., 2018 + SN Cerebellum Cerebellar fMRI Networks 7 Buckner et al., 2011 + 2 Masked by SUIT Atlas Diedrichsen et al., 2006 + 1–34 Ca = caudate; GM = grey matter; PMd = dorsal premotor cortex; PMv = ventral premotor cortex; Pu = putamen; SNc = substantia nigra pars compacta; VLa = anterior ventrolateral nucleus of the thalamus. Open in new tab Table 1 Atlases that contributed to motor network definition Structure/Atlas Name Atlas reference(s) Atlas available in Lead-DBS Parcels used Cortex Automated Anatomical Labeling Atlas Tzourio-Mazoyer et al., 2002 + 1–2, 19–20, 57–58 AICHA Atlas Joliot et al., 2015 + 51–62, 71–76, 221–226 Hammersmith Atlas n30r83 Hammers et al., 2003; Gousias et al., 2008 + 50–51, 60–61 Harvard Oxford Thr25 2 mm Makris et al., 2006 + 13–14, 33–34, 51–52 JueBrain atlas thr25 2 mm Eickhoff et al., 2005 + 47–66, 91–92 Mindboggle 101, Desikan protocol Klein et al., 2012 + 1022, 2022, 1024, 2024 HMAT Atlas Mayka et al., 2006 + 1–12 Masked by GM component of ICBM 2009b NLIN space (Lead-DBS) Fonov et al., 2009 + – Striatum Striatal fMRI Networks 7 Choi et al., 2012 + Somatomotor Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by CIT168_Reinf_Learning Atlas Pauli et al., 2017 + Ca, Pu, NAc Thalamus Morel Atlas Krauth et al., 2010; Jakab et al., 2012; Morel et al., 2013 – VLa Thalamic fMRI Networks 7 Choi et al., 2012 – Somatomotor Masked by Morel Atlas Krauth et al., 2010; Jakab et al., 2012; Morel et al., 2013 – All thalamic components GPi DISTAL Atlas Ewert et al., 2018 + GPi_sensorimotor Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by DISTAL Atlas Ewert et al., 2018 + GPi GPe Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by DISTAL Atlas Ewert et al., 2018 + GPe Substantia nigra Nigral organization atlas Zhang et al., 2017 + Lateral_SNc_somatomotor Masked by DISTAL Atlas Ewert et al., 2018 + SN Cerebellum Cerebellar fMRI Networks 7 Buckner et al., 2011 + 2 Masked by SUIT Atlas Diedrichsen et al., 2006 + 1–34 Structure/Atlas Name Atlas reference(s) Atlas available in Lead-DBS Parcels used Cortex Automated Anatomical Labeling Atlas Tzourio-Mazoyer et al., 2002 + 1–2, 19–20, 57–58 AICHA Atlas Joliot et al., 2015 + 51–62, 71–76, 221–226 Hammersmith Atlas n30r83 Hammers et al., 2003; Gousias et al., 2008 + 50–51, 60–61 Harvard Oxford Thr25 2 mm Makris et al., 2006 + 13–14, 33–34, 51–52 JueBrain atlas thr25 2 mm Eickhoff et al., 2005 + 47–66, 91–92 Mindboggle 101, Desikan protocol Klein et al., 2012 + 1022, 2022, 1024, 2024 HMAT Atlas Mayka et al., 2006 + 1–12 Masked by GM component of ICBM 2009b NLIN space (Lead-DBS) Fonov et al., 2009 + – Striatum Striatal fMRI Networks 7 Choi et al., 2012 + Somatomotor Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by CIT168_Reinf_Learning Atlas Pauli et al., 2017 + Ca, Pu, NAc Thalamus Morel Atlas Krauth et al., 2010; Jakab et al., 2012; Morel et al., 2013 – VLa Thalamic fMRI Networks 7 Choi et al., 2012 – Somatomotor Masked by Morel Atlas Krauth et al., 2010; Jakab et al., 2012; Morel et al., 2013 – All thalamic components GPi DISTAL Atlas Ewert et al., 2018 + GPi_sensorimotor Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by DISTAL Atlas Ewert et al., 2018 + GPi GPe Basal Ganglia Functional Zones Draganski et al., 2008 – M1, PMd_PMv Masked by DISTAL Atlas Ewert et al., 2018 + GPe Substantia nigra Nigral organization atlas Zhang et al., 2017 + Lateral_SNc_somatomotor Masked by DISTAL Atlas Ewert et al., 2018 + SN Cerebellum Cerebellar fMRI Networks 7 Buckner et al., 2011 + 2 Masked by SUIT Atlas Diedrichsen et al., 2006 + 1–34 Ca = caudate; GM = grey matter; PMd = dorsal premotor cortex; PMv = ventral premotor cortex; Pu = putamen; SNc = substantia nigra pars compacta; VLa = anterior ventrolateral nucleus of the thalamus. Open in new tab Estimating changes in rs-fMRI based connectivity Rs-fMRI scans under DBS on and off conditions were preprocessed using Lead-Connectome (www.lead-connectome.org; Horn et al., 2014). The pipeline is described in detail elsewhere (Horn et al., 2014; Horn, 2015; Horn and Blankenburg, 2016) and largely follows the suggestions outlined in Weissenbacher et al. (2009). Briefly, rs-fMRI time series were detrended and motion parameters, as well as mean signals from white matter and CSF were added as nuisance regressors for noise removal. No global signal regression was performed. Signals were then band-pass filtered between 0.009 and 0.08 Hz. Finally, spatial smoothing with an isotropic 6 mm full-width at half-maximum Gaussian kernel was applied. As all tools involved in the present study, the preprocessing pipeline is openly available to ensure reproducibility (https://github.com/leaddbs/leaddbs/blob/master/connectomics/ea_preprocess_fmri.m). A voxel-wise parcellation scheme with 8000 nodes in grey matter [see ‘Voxelwise parcellations (Lead-DBS)’ under http://www.lead-dbs.org/helpsupport/knowledge-base/atlasesresources/cortical-atlas-parcellations-mni-space] was inversely registered to rs-fMRI volumes (via the T1 anchor modality) and signals were sampled from each node. Marginal correlations between time series were estimated, leading to two 8000 × 8000 adjacency matrices for DBS on and off conditions in each subject. Based on these, average connectivity between each node and all other 8000 − 1 nodes were calculated. Formally, this measure may be referred to as strength centrality (the weighted version of degree centrality; Rubinov and Sporns, 2010). However, we henceforth refer to it as ‘average connectivity’, as this seems a fitting intuitive term for what we calculate. Likewise, ‘changes in average connectivity’ denote the difference between DBS on and off conditions. Spearman correlations between local DBS impact and fMRI-derived measures were calculated. Random permutation (×5000) was conducted to obtain P-values. Subsequently, changes in connectivity between specific parts of the ‘motor network’ (Table 1) as a function of motor STN-DBS were estimated using the network-based statistics (NBS) approach as implemented in the GraphVar toolbox (http://www.rfmri.org/GraphVar; Kruschwitz et al., 2015). This involved non-parametric testing of DBS on/off differences in each link that were explained by the degree of motor STN-DBS stimulation. Overall significance was then estimated using the NBS approach on the whole graph matrix (Zalesky et al., 2010). Data availability The DBS MRI datasets generated and analysed during the current study are not publicly available because of data privacy regulations of patient data, but are available from the corresponding author upon reasonable request. The control cohort is available within the PPMI repository (www.ppmi-info.org). All code used to analyse the dataset is available within Lead-DBS/-Connectome software (https://github.com/leaddbs/leaddbs). Results Patients (n = 20) were 63.7 ± 6.6 years old at time of surgery and the sample included four females. Healthy control subjects were 59.5 ± 11.9 years old and 4 of 15 subjects were female. Patients were scanned 30 ± 21 months after surgery with a minimum of 4 months in one subject. All other subjects had been implanted at least 11 months at the time of the scan. Because of logistical reasons (see ‘Materials and methods’ section), patients were scanned in medication ON condition and first scanned in DBS on followed by the DBS off condition (with a brief interval of 5–15 min in which the impulse generator was turned off). Before surgery, patients had a levodopa response of 36.8 ± 11% [per cent improvement in medication ON versus OFF conditions; from 35.9 ± 7.6 to 18.9 ± 6.8 points on the Unified Parkinson’s Disease Rating Scale (UPDRS-III) scale; lacking scores of two patients]. LEDD of medication was 724.7 ± 441 mg (lacking data of one patient). Under effective DBS, motor symptoms improved by 43 ± 18% (from 38.8 ± 10.9 points before in DBS off to 22.0 ± 9.8 in DBS on; lacking scores of two patients). Note that stimulation parameters during rs-fMRI differed as they had to be in the bipolar setting and were not all taken at the same time of the 12-month postoperative evaluation. The aim was to best match unipolar long-term stimulation settings with the bipolar settings used during rs-fMRI acquisition. Nevertheless, for logistic reasons, a second UPDRS score could not be taken at the time of scanning and thus a matching UPDRS-III improvement for the DBS applied in the scanner is not available in this cohort. Electrodes were all placed in the subthalamic region (Fig. 1) although five patients had an overlap between their bilateral VTA and the motor parts of the bilateral STN (as defined by the DISTAL atlas; Ewert et al., 2018) below 4 mm3. In comparison, the average coverage of motor STN stimulation in the rest of the sample was 36.4 ± 20.5 mm3, which amounts to 50.6% of the bilateral nucleus volume. Congruently, clinical improvements in this subgroup of patients were significantly worse compared to the rest of the cohort (29.5 ± 14% versus 49.4 ± 17% improvement on the UPDRS-III scale; P = 0.039), although as mentioned above, VTAs used for clinical assessment and during the rs-fMRI experiment were not directly comparable. Figure 1 Open in new tabDownload slide DBS electrode placement in the sample of 20 patients. (A) 3D lead reconstructions together with STN (orange), GPi (green), GPe (blue) and red nucleus (red) as defined by the DISTAL atlas (Ewert et al., 2018). An axial section of the BigBrain dataset (Amunts et al., 2013) at z = −10 mm is shown. Right hemispheric volumes of activated tissue (B) and active (bipolar) stimulation contacts (C) were non-linearly flipped to the left hemisphere to show the combined stimulation volume of the group (white wireframes) and the active contacts (red dots) in relationship to the STN. Figure 1 Open in new tabDownload slide DBS electrode placement in the sample of 20 patients. (A) 3D lead reconstructions together with STN (orange), GPi (green), GPe (blue) and red nucleus (red) as defined by the DISTAL atlas (Ewert et al., 2018). An axial section of the BigBrain dataset (Amunts et al., 2013) at z = −10 mm is shown. Right hemispheric volumes of activated tissue (B) and active (bipolar) stimulation contacts (C) were non-linearly flipped to the left hemisphere to show the combined stimulation volume of the group (white wireframes) and the active contacts (red dots) in relationship to the STN. To investigate whether subject motion in the scanner differed between DBS on and off conditions, motion parameters were estimated using SPM12 and framewise displacements calculated following the approach of Power et al. (2014) (Fig. 2C). Average displacement values for DBS on and off conditions were not significantly different (0.26 ± 0.17 mm in on condition versus 0.30 ± 0.17 mm in off, P = 0.32) and within a typical range of tolerable movement (Ardekani et al., 2001). Figure 2 Open in new tabDownload slide BOLD signal around DBS electrodes and raw rs-fMRI volumes. (A and B) Resting-state time series sampled from the left VTA (red/yellow) and bilateral SMA (blue/cyan) in an example patient in DBS on condition. Despite presence of electrode artefacts, useful BOLD time series could be sampled from the vicinity of the electrode (i.e. the VTA) that show functional connectivity to cortical motor regions (such as the SMA). (A) 3D reconstruction and example time series from the left hemisphere. (B) Actual EPI volume with region of interest overlays. Susceptibility artefact of the lead extension can be seen on the axial slice. In this patient, correlation between the two contralateral VTA time series was R = 0.41, correlation between left/right VTA and SMA were R = 0.34/0.25. In the DBS off condition, values were similar (left to right VTA: R = 0.34; left/right VTA to SMA: R = 0.32 / 0.42). Sagittal and coronal sections of the BigBrain dataset (Amunts et al., 2013) shown as backdrops in A. Figure 2 Open in new tabDownload slide BOLD signal around DBS electrodes and raw rs-fMRI volumes. (A and B) Resting-state time series sampled from the left VTA (red/yellow) and bilateral SMA (blue/cyan) in an example patient in DBS on condition. Despite presence of electrode artefacts, useful BOLD time series could be sampled from the vicinity of the electrode (i.e. the VTA) that show functional connectivity to cortical motor regions (such as the SMA). (A) 3D reconstruction and example time series from the left hemisphere. (B) Actual EPI volume with region of interest overlays. Susceptibility artefact of the lead extension can be seen on the axial slice. In this patient, correlation between the two contralateral VTA time series was R = 0.41, correlation between left/right VTA and SMA were R = 0.34/0.25. In the DBS off condition, values were similar (left to right VTA: R = 0.34; left/right VTA to SMA: R = 0.32 / 0.42). Sagittal and coronal sections of the BigBrain dataset (Amunts et al., 2013) shown as backdrops in A. To investigate whether BOLD time series sampled directly from the region of the electrode/VTA could be used despite the prevalence of metal artefacts, we first calculated functional connectivity between bilateral VTAs and second between VTAs and supplementary motor area (SMA) in DBS off conditions. The SMA was chosen given its functional role and positive connectivity to the STN (Horn et al., 2017; Akram et al., 2018). In some patients, correlation values of R > 0.3 were prevalent between bilateral VTAs, suggesting the STN time series around the electrode are useable but variability in lead placement introduced strong variance across patients (average absolute R-values were 0.16 ± 0.11). Crucially, the connectivity between the two electrodes could be explained by lead placement: The more similar their impact on the motor STN, the higher their connectivity (R = 0.45 at P < 0.05). Functional connectivity values between bilateral VTAs and SMA were similar, R-values ranged up to 0.24 (average absolute R-values were 0.14 ± 0.06). Here, lead placement did not significantly correlate with connectivity between VTA and SMA (R = 0.31, P = 0.09). Figure 2 illustrates BOLD time series sampled from the VTA and the SMA of a representative patient and gives a visual impression about the BOLD signal sampled directly at and around the electrode artefact. Despite the apparent usefulness of the BOLD signal directly around the stimulation sites, the first and main analysis of this manuscript did not use this signal but rather investigated whole-brain connectivity changes within the motor network (cortex, cerebellum and basal ganglia). Average connectivity changes between DBS on and off conditions were estimated on a graph of 8000 nodes that covered the whole brain but were then sampled within the motor network (see Table 1 for a proper definition of the network). In other words, overall connectivity changes between each pair of voxels in the whole brain was defined and changes in the motor network were then correlated with impact of DBS on the motor STN. Across subjects, changes in average connectivity of the motor network were correlated with the impact of the DBS electrode on the motor STN (R = 0.71, P < 0.001). Figure 3 shows these results across the group and highlights two example subjects. In the first (Patient 9; Fig. 3, top), overlap between the VTA and motor STN is large, i.e. it represents a case with (near) optimally placed DBS electrodes. As expected, in this patient under DBS, average connectivity increases strongly and predominantly in the motor network. The contrary is the case in a patient where the leads are largely outside the STN where as expected little to no modulation of the motor network takes place (Patient 1; Fig. 3, bottom). Across the group, this relationship between DBS placement and the connectivity increase in the motor network was strong. In fact, 50% of the observed variance in motor network changes (which are exclusively informed by rs-fMRI) may be explained purely by lead placement (which is exclusively informed by structural MRI). Figure 3 Open in new tabDownload slide The amount of stimulated motor STN explains connectivity changes within the motor network across patients. The larger the weighted overlap between electric field and motor STN, the more connected the motor network becomes (on average to all other brain regions). Here, ‘motor network’ refers to sensorimotor zones (including premotor domains) in cortex, cerebellum, striatum, pallidum and thalamus (visible parts outlined by dashed lines; see methods). Example Patients 1 and 9 are marked in the scatterplot and their electrode placement (left) and cortical changes in connectivity induced by DBS (right) are shown. Crucially, cortical connectivity changes were calculated independently from electrode placement. Both electric field and cortical changes are visualized for the left hemisphere only but were similar on the right hemisphere. Figure 3 Open in new tabDownload slide The amount of stimulated motor STN explains connectivity changes within the motor network across patients. The larger the weighted overlap between electric field and motor STN, the more connected the motor network becomes (on average to all other brain regions). Here, ‘motor network’ refers to sensorimotor zones (including premotor domains) in cortex, cerebellum, striatum, pallidum and thalamus (visible parts outlined by dashed lines; see methods). Example Patients 1 and 9 are marked in the scatterplot and their electrode placement (left) and cortical changes in connectivity induced by DBS (right) are shown. Crucially, cortical connectivity changes were calculated independently from electrode placement. Both electric field and cortical changes are visualized for the left hemisphere only but were similar on the right hemisphere. This analysis concluded that changes of average connectivity throughout the motor network were dependent on how strongly the DBS electrode modulated the motor STN. This latter factor is dependent on electrode placement and the applied stimulation amplitudes (as known parameters) and on the electrode-tissue impedance (which was not measured and thus is an unknown parameter). To exclude the possibility that the impedance had a significant impact on the observed relationship, we performed a stochastic sensitivity analysis to account for the potential bias introduced by tissue impedance. Specifically, electric fields were recalculated, in each patient, after scaling the assumed tissue conductivity values by factors of 0.5, 0.75, 1, 1.5 and 2. Based on these additional electric fields that were calculated using adapted conductivity values, a 5000-fold permutation test was run and correlations were recalculated for each permuted combination. This analysis showed high robustness of the relationship towards the potential influence of the unknown factor tissue impedance. In fact, all combinations of VTAs under different impedance assumptions revealed constantly significant relationships, with the lowest R-value at 0.47 (P = 0.028), the highest at 0.75 (P < 0.001) and the average at R = 0.58 ± 0.04 (P = 0.004 ± 0.004). After this analysis, we investigated relationships between motor STN stimulation and connectivity changes of specific regions within the motor network. In post hoc uncorrected head-to-head comparisons, this relationship was significant for the cortical (R = 0.50, P = 0.024) and cerebellar (R = 0.51, P = 0.022) regions of interest, but not for remaining region of interest (striatum, thalamus, GPi, GPe, substantia nigra) with a positive but non-significant R-value for all regions. Subsequently, connectivity profiles seeding from bilateral VTAs in DBS on and off conditions were compared to the same maps informed by a normative connectome of age-matched healthy controls (Marek et al., 2011; Horn et al., 2017). DBS on and off maps (Fig. 4, top right and left maps) were each compared to healthy controls (Fig. 4, top middle map) by means of voxel-wise spatial correlation (as a similarity metric). Here, seed connectivity maps in DBS on condition were significantly more similar to the ones obtained when using rs-fMRI data of healthy controls both on a group level (Fig. 4, bottom right and left scatterplots) and single subject level (Fig. 4, bottom middle bar plots). This analysis shows that the overall connectivity of the DBS electrode ‘normalizes’ towards healthy controls under DBS. In a subsequent analysis, we compared whole-brain average connectivity estimates (voxel-wise strength centrality) changes based on DBS. These estimates became more similar to the ones obtained in healthy controls as a function of motor-STN stimulation. The same two patients (Patients 9 and 1; Fig. 5, top and bottom rows) are shown as examples. Here, the average connectivity metrics across the whole brain—and not just the motor network—became more similar to healthy controls as a clear function of motor STN stimulation. Similar to the analysis in Fig. 3, ∼50% of variance in how much the overall average connectivity ‘normalizes’ towards healthy controls could be explained just based on DBS electrode placement. In other words, the more optimal a DBS electrode was placed (as measured by overlap with the motor STN), the more normal the overall functional connectivity became (as measured by similarity to healthy controls). Figure 4 Open in new tabDownload slide Rs-fMRI functional connectivity profiles seeding from volumes of tissue activated (weighted sum of electric fields estimating the bilateral DBS stimulation volumes) to the rest of the brain. Patient specific data in DBS off (top left), on (top right) conditions as well as maps derived from healthy age-matched controls (top middle) were compared. Unthresholded T-maps calculated across the group of 20 patients are shown in the top row. Spatial correlation values (that assess similarity) across cortical voxels between each DBS condition and the healthy controls were calculated on a group (scatterplots) and individual level (bottom middle) and were significantly higher between on and healthy controls versus off and healthy controls. Figure 4 Open in new tabDownload slide Rs-fMRI functional connectivity profiles seeding from volumes of tissue activated (weighted sum of electric fields estimating the bilateral DBS stimulation volumes) to the rest of the brain. Patient specific data in DBS off (top left), on (top right) conditions as well as maps derived from healthy age-matched controls (top middle) were compared. Unthresholded T-maps calculated across the group of 20 patients are shown in the top row. Spatial correlation values (that assess similarity) across cortical voxels between each DBS condition and the healthy controls were calculated on a group (scatterplots) and individual level (bottom middle) and were significantly higher between on and healthy controls versus off and healthy controls. Figure 5 Open in new tabDownload slide The shift of whole brain average connectivity profiles towards that of healthy controls (central) is dependent on the amount of motor STN stimulated (left scatterplot). Whole brain average connectivity profiles in DBS off (left) and on (right) conditions of the same example patients as in Fig. 3 are shown. Their connectivity distribution is spatially correlated with the one in healthy controls (to assess similarity) and the difference between conditions is related to the amount of motor STN stimulation in each patient. As can be seen in Patient 9 (top row), DBS has a strong effect on the average connectivity profile. In contrast, profiles of Patient 1 remain largely unaltered. Figure 5 Open in new tabDownload slide The shift of whole brain average connectivity profiles towards that of healthy controls (central) is dependent on the amount of motor STN stimulated (left scatterplot). Whole brain average connectivity profiles in DBS off (left) and on (right) conditions of the same example patients as in Fig. 3 are shown. Their connectivity distribution is spatially correlated with the one in healthy controls (to assess similarity) and the difference between conditions is related to the amount of motor STN stimulation in each patient. As can be seen in Patient 9 (top row), DBS has a strong effect on the average connectivity profile. In contrast, profiles of Patient 1 remain largely unaltered. To investigate exactly which connections would increase or decrease further, we applied the network based statistics approach on DBS on versus off graphs between regions of interest defined in Table 1 (which include the motor domains of STN, GPi, GPe, striatum, thalamus, cerebellum, substantia nigra and cortex). Again, changes were investigated as a function of motor STN DBS impact (weighted overlap between electric field and motor STN). As expected, this showed a significant increase between thalamus and cortex and decreases between striatum and cerebellum, striatum and STN as well as STN and GPe (Fig. 6). Figure 6 Open in new tabDownload slide Specific connections in the motor network modulated by effective STN-DBS. Functional connectivity between motor thalamus and cortex (SMA, M1, S1) increases as a function of DBS impact on the motor STN. Instead, connectivity between motor striatum and cerebellum, motor striatum and motor STN as well as motor STN and motor GPe decreases. For exact definition of regions of interest within the motor network see Table 1. **P < 0.005, *P < 0.05 corrected for multiple comparisons using the network-based statistics approach as implemented in the GraphVar toolbox. Figure 6 Open in new tabDownload slide Specific connections in the motor network modulated by effective STN-DBS. Functional connectivity between motor thalamus and cortex (SMA, M1, S1) increases as a function of DBS impact on the motor STN. Instead, connectivity between motor striatum and cerebellum, motor striatum and motor STN as well as motor STN and motor GPe decreases. For exact definition of regions of interest within the motor network see Table 1. **P < 0.005, *P < 0.05 corrected for multiple comparisons using the network-based statistics approach as implemented in the GraphVar toolbox. In summary, results show that effective STN-DBS leads to an overall increase of connectivity in the motor network of the brain (Fig. 3), potentially mediated by a reduced basal ganglia output (and stronger thalamo-cortical interaction, Fig. 6). Second, STN-DBS seems to attenuate or ‘normalize’ both functional connectivity profiles seeding from the electrode (Fig. 4) and the overall average connectivity (i.e. centrality) estimates in the brain (Fig. 5). Discussion Four main conclusions may be drawn from the present study. First, the acquisition of postoperative rs-fMRI in DBS patients is feasible in a clinical setting and resulting data are both useful and useable to noninvasively study effects of DBS on distributed brain networks. Based on functional connectivity estimates to other regions of the motor network, even the BOLD signal sampled directly at and around the electrode exhibited a physiologically meaningful signal. Second, STN-DBS had a significant effect on brain connectivity throughout the motor network, specifically on its cortical and cerebellar subparts. Third, and most importantly, DBS induced changes on average connectivity throughout the motor network were strongly dependent on electrode placement. Specifically, electrodes with strong impact on the motor STN induce larger changes than the ones with weak or no impact on the motor STN (R = 0.7, P < 0.001). This correlation is crucial as it may show a direct modulatory effect of STN-DBS on the motor network. Finally, STN-DBS had the effect of ‘normalizing’ both connectivity profiles of both the electrodes but also average connectivity profiles of the brain towards profiles found in age-matched healthy control subjects. Again, this effect was dependent on electrode location; well-placed electrodes shifted the overall connectivity profiles more strongly towards controls than poorly placed electrodes. Clinical implications The main clinical finding shows that connectivity changes induced by DBS with optimally placed leads are being ‘normalized’ towards healthy controls. Overall, STN-DBS led to increased functional connectivity in the motor network and relative decoupling of basal ganglia pathways. Our findings are based on the largest DBS-fMRI sample to date and extended by explaining individual differences in network changes as a function of DBS electrode placement. This extension is crucial since it shows that fMRI could potentially be used to even guide DBS surgery and programming. Specifically, it could be feasible to acquire real-time rs-fMRI during surgeries that are based on intraoperative MRI. In such a setting, one could picture test stimulations during and after surgery. Lead placement and parameters could then be semi-automatically tuned to maximize impact of DBS on the motor network. Even without intraoperative MRI, before surgery, our findings could be used to guide DBS targeting by estimating the region in and around the STN that is most strongly functionally connected to the motor network. Such a concept has been briefly explored for essential tremor (Anderson et al., 2011) in the past, but could be extended to STN-DBS for Parkinson’s disease as well. Pathophysiological implications Beyond clinical implications of our findings, they may also shed light on the pathophysiology of Parkinson’s disease and the mechanism of action of DBS. Specifically, we showed an increase of functional connectivity between the motor thalamus and cortex as a function of effective DBS and a decrease of striato-cerebellar coupling. Both effects have partly been shown using fMRI before (Kahan et al., 2014). In particular, the increase of thalamo-cortical interaction by STN-DBS may be derived from classical basal-ganglia cortical loop models (Rodríguez-Oroz et al., 2009). In Parkinson’s disease, the dopamine deficit leads to increased activity of the indirect basal ganglia pathway which results in increased GPi/substantia nigra pars reticulata output and in turn inhibition of the ventrolateral thalamic nucleus. STN hyperactivity is a key characteristic of this pathological circuitry and by attenuating this hyperactivity with DBS the inhibition of the thalamo-cortical pathway is reduced (Rodríguez-Oroz et al., 2009; Kahan et al., 2014). Correspondingly, our results suggest an attenuation of coupling between the STN and striatum as well as the STN and GPe. Taking the above into account, one may speculate that DBS could ‘rescue’ the thalamocortical interaction due to a disruption of subthalamic control over GABAergic pallidothalamic efferents. In turn, this seems implemented by attenuating coupling between the overactive STN and the striatum. In this way, DBS may restore the hypodopaminergic loss of effective coding capacity within the thalamo-cortical network. Similarly, our findings suggest that striatal dominance over the cerebellum could be attenuated by effective DBS. Increased functional striato-cerebellar coupling was described in Parkinson’s disease patients before (Hou et al., 2016) and attributed to a compensatory effect that was associated to improved movement performance (Simioni et al., 2016). Be it pathological or compensatory, the increased coupling was decreased to a level found in healthy controls under dopaminergic medication. Our study now extends these findings by showing that STN-DBS seems to have a similar attenuating effect on increased striato-cerebellar coupling. DBS as an optimal tool to study the functional connectome of the brain Our study identified DBS as a promising tool to study changes of the functional connectome due to precise brain stimulation of focal subcortical areas. Currently, such a focal stimulation is not possible using other techniques in the human brain. Moreover, deeper structures are not accessible by non-invasive brain stimulation. Worldwide, DBS is an increasingly applied treatment option that is well established for severe movement disorders but indications extend to a growing number of psychiatric diseases (Lozano and Lipsman, 2013). Accordingly, a growing number of structures are targeted by DBS (Fox et al., 2014). Sometimes, the same disease can be treated by targeting varying brain structures. For instance, at least five targets are under investigation to treat obsessive compulsive disorder (de Koning et al., 2011) or treatment-refractory depression (Zhou et al., 2018). Even established diseases are treated with different targets (for instance both Parkinson’s disease and dystonia have been treated by stimulating STN, GPi or thalamus). It seems that DBS may modulate symptom-related brain networks that may be overlapping at the same node (e.g. the therapeutic network modulated in Parkinson’s disease and dystonia seems to overlap at both STN and GPi). Functional MRI analyses may help further strengthen this concept. Adding to the complexity, different targets can be used in the same disease to preferentially treat different symptoms (for instance STN being an obvious choice to treat most Parkinson’s disease symptoms while the ventral intermediate nucleus of the thalamus is predominantly used to exclusively treat parkinsonian tremor). DBS cohorts may now be studied using functional MRI (in DBS on setting currently approved for some systems by Medtronic only) as was done here and in a few previous studies (Table 1). Needless to say, beyond acquiring fMRI data under stimulation in the resting-state, changes in task-fMRI paradigms could equally be studied. We argue that DBS-fMRI may soon become a new field of research that may be crucial to investigating the functional architecture of the human brain. Based on this present study, we conclude that much focus should be put upon precise localization of DBS electrode placement. Specifically, studying on versus off group results without taking electrode placement into account may lead to erroneous conclusions. The reason is that small changes in placement do result in clinically meaningful changes in motor outcome (Horn et al., 2019; Yao et al., 2019) or behavioural response (Irmen et al., 2018; Neumann et al., 2018). In the present study we show that the same is true for functional response patterns. Limitations Several limitations apply for careful interpretation of our results. First and foremost, despite several publications in the past (Table 1), the field of fMRI under DBS is still quite new and most publications were case reports or based on low n cohorts of ∼10 patients. Further, the impact of DBS induced artefacts on the rs-fMRI signal has not been investigated in detail and more methodological work is needed to address potential issues, in the future. To our knowledge, several laboratories worldwide are currently investigating methodological and signal-processing questions related to fMRI under DBS. In our study, the BOLD signal sampled from directly around the electrode was at times highly correlated to the SMA (with a maximum R of 0.7 that is unlikely to originate from noise; Fig. 2), a region that is coupled to the STN and not (directly) impacted by metal artefacts. This made us decide to use the BOLD signal of the electrode itself in one analysis (Fig. 4). However, we refrained from using the signal in the most central analyses from which we draw our main conclusions (Figs 2 and 5). Similarly, the DBS lead extensions typically induce an artefact in cortical structures (as seen in Fig. 2) that can currently not be avoided. No software algorithms have been introduced to correct for such artefacts given these types of datasets are still comparably new. To reduce the theoretical bias induced by the extension artefact, we chose to use a coarse parcellation of the cortex (while finer parcellations may run into the risk that single parcels are completely filled by the artefact). A crucial parameter that was investigated here for the first time in an fMRI DBS context was the impact of the stimulation on the motor STN. Several limitations apply for the model used to derive this measure. First, lead reconstruction and patient registration to the STN atlas may be biased. To this end, we use a state-of-the-art DBS imaging pipeline that has been explicitly designed for the task. In fact, a recent study could show that using our pipeline, automatic registrations between patient space and an STN atlas are not significantly different from manual expert segmentations of the nucleus (Ewert et al., 2019). The study involved >11 000 non-linear warps in >100 brains and tested six multispectral deformation algorithms with several parameter settings each to fit patient data to the STN atlas. Second, the atlas of the STN itself might be biased. To this end, we used a modern atlas that was explicitly created for our pipeline (Ewert et al., 2018) and is based on the newest multispectral high-resolution MNI template available (Fonov et al., 2009). Recently, others confirmed accuracy of the atlas based on intraoperative microelectrode recordings and showed that it matched electrophysiological data better than three other atlases of the STN (Nowacki et al., 2018). Finally, the motor functional zone of the atlas was defined using diffusion weighted imaging data acquired on specialized MR hardware (Setsompop et al., 2013) and cross-validated using both healthy controls and Parkinson’s disease data (Ewert et al., 2018). A third potential limitation is the impact of the electric field on the STN. In the past, the VTA by DBS (McIntyre and Grill, 2002) has been modelled as a binary region around the electrode (McIntyre et al., 2004; Butson et al., 2007; Mädler and Coenen, 2012; Dembek et al., 2017). However, the degree of mesoscopic tissue modulation may not be a binary (‘all or nothing’) effect but instead show probabilistic properties, with areas closer to the electrode being more strongly modulated than more distant ones (Eisenstein et al., 2014). In line with this assumption, we were recently able to predict slightly more variance in clinical outcome using a weighted over a binary model (Horn et al., 2019). This led us to adopt the same principle here, i.e. to calculate the sum over the electric field distribution within the motor STN domain. Still, as mentioned earlier, all results shown here would hold significant when repeating the analyses using a binary VTA instead. Finally, clinical improvement data were not acquired at the time of imaging due to logistic reasons. Surrogate clinical improvement estimates were collected during clinical routine (see ‘Results’ section) but the VTAs used during the 12-month visit were substantially different from the VTAs used in the fMRI experiment. Thus, our study describes network effects of DBS on the brain but cannot draw direct conclusions on how those affect clinical outcome. Relatedly, our patients were scanned in the medication ON condition for logistic reasons and to reduce additional motion artefacts. Thus, an effect of dopaminergic medication cannot be excluded. The two scans directly followed each other (i.e. were performed under comparable medication) and we mainly draw conclusions that are dependent on electrode placement throughout the manuscript. Still, further studies are needed to draw inferences in the clinical and medication domain. Conclusions Our study exemplifies the use of invasive brain stimulation to study and modulate the functional connectome of the human brain. The study shows the promise of using invasive neuromodulation in the exemplary case of Parkinson’s disease and hints at the promise to broaden this novel field of functional neuroimaging under precisely targeted and focal brain stimulation. More specifically, we demonstrate that the acquisition of postoperative rs-fMRI under STN-DBS in a clinical setting is feasible and resulting data are useful to non-invasively study DBS effects. DBS may attenuate pathological basal ganglia output leading to changes in average connectivity within the motor network that were strongly dependent on electrode placement. Finally, STN-DBS seemed to have the effect of ‘normalizing’ brain connectivity towards that of healthy controls. Similar studies in populations under DBS with different targets and/or in different diseases will extend our knowledge on the impact of focal brain stimulation on distributed brain networks. Abbreviations Abbreviations BOLD blood oxygenated level-dependent DBS deep brain stimulation rs-fMRI resting-state functional MRI SMA supplementary motor area STN subthalamic nucleus VTA volume of tissue activated Acknowledgement We would like to thank Lea Waller for support on using the GraphVar toolbox. Funding The study was supported by DFG grant SPP 2041 to A.A.K. and DFG Emmy Noether grant 410169619 to A.H. Data used in the preparation of this article were obtained from the Parkinson’s Progression Markers Initiative (PPMI) database (www.ppmi-info.org/data). For up-to-date information on the study, visit www.ppmi-info.org. PPMI—a public-private partnership—is funded by the Michael J. Fox Foundation for Parkinson’s Research and funding partners, see www.ppmi-info.org/fundingpartners. Competing interests A.H. reports one-time lecture fee from Medtronic. A.A.K. reports personal fees and non-financial support from Medtronic, personal fees from Boston Scientific, personal fees from Ipsen Pharma, grants and personal fees from St. Jude Medical outside the submitted work. G.W. recieved travel grants from Boston Scientific. P.K. reports one-time lecture fee for Medtronic. F.I., J.H., N.L., W-J.N., G.B. and M.S. have nothing to disclose. References Akram H , Dayal V , Mahlknecht P , Georgiev D , Hyam J , Foltynie T et al. . Connectivity derived thalamic segmentation in deep brain stimulation for tremor . Neuroimage Clin 2018 ; 18 : 130 – 42 . Google Scholar Crossref Search ADS PubMed WorldCat Akram H , Sotiropoulos SN , Jbabdi S , Georgiev D , Mahlknecht P , Hyam J et al. . Subthalamic deep brain stimulation sweet spots and hyperdirect cortical connectivity in Parkinson’s disease . NeuroImage 2017 ; 158 : 332 – 45 . Google Scholar Crossref Search ADS PubMed WorldCat Amunts K , Lepage C , Borgeat L , Mohlberg H , Dickscheid T , Rousseau M-É et al. . BigBrain: an ultrahigh-resolution 3D human brain model . Science 2013 ; 340 : 1472 – 5 . Google Scholar Crossref Search ADS PubMed WorldCat Anderson JS , Dhatt HS , Ferguson MA , Lopez-Larson M , Schrock LE , House PA et al. . Functional connectivity targeting for deep brain stimulation in essential tremor . Am J Neuroradiol 2011 ; 32 : 1963 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Ardekani BA , Bachman AH , Helpern JA . A quantitative comparison of motion detection algorithms in fMRI . Magn Reson Imaging 2001 ; 19 : 959 – 63 . Google Scholar Crossref Search ADS PubMed WorldCat Avants BB , Epstein CL , Grossman M , Gee JC . Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain . Med Image Anal 2008 ; 12 : 26 – 41 . Google Scholar Crossref Search ADS PubMed WorldCat Åström M , Diczfalusy E , Martens H , Wårdell K . Relationship between neural activation and electric field distribution during deep brain stimulation . IEEE Trans Biomed Eng 2015 ; 62 : 664 – 72 . Google Scholar Crossref Search ADS PubMed WorldCat Åström M , Zrinzo LU , Tisch S , Tripoliti E , Hariz MI , Wårdell K . Method for patient-specific finite element modeling and simulation of deep brain stimulation . Med Biol Eng Comput 2009 ; 47 : 21 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Biswal BB , Mennes M , Zuo X-N , Gohel S , Kelly C , Smith SM et al. . Toward discovery science of human brain function . Proc Natl Acad Sci 2010 ; 107 : 4734 – 9 . Google Scholar Crossref Search ADS PubMed WorldCat Buckner RL , Krienen FM , Castellanos A , Diaz JC , Yeo BTT . The organization of the human cerebellum estimated by intrinsic functional connectivity . J Neurophysiol 2011 ; 106 : 2322 – 45 . Google Scholar Crossref Search ADS PubMed WorldCat Butson CR , Cooper SE , Henderson JM , McIntyre CC . Patient-specific analysis of the volume of tissue activated during deep brain stimulation . NeuroImage 2007 ; 34 : 661 – 70 . Google Scholar Crossref Search ADS PubMed WorldCat Carmichael DW , Pinto S , Limousin-Dowsey P , Thobois S , Allen PJ , Lemieux L et al. . Functional MRI with active, fully implanted, deep brain stimulation systems: safety and experimental confounds . NeuroImage 2007 ; 37 : 508 – 17 . Google Scholar Crossref Search ADS PubMed WorldCat Choi EY , Yeo BTT , Buckner RL . The organization of the human striatum estimated by intrinsic functional connectivity . J Neurophysiol 2012 ; 108 : 2242 – 63 . Google Scholar Crossref Search ADS PubMed WorldCat de Koning PP , Figee M , van den Munckhof P , Schuurman PR , Denys D . Current status of deep brain stimulation for obsessive-compulsive disorder: a clinical review of different targets . Curr Psychiatry Rep 2011 ; 13 : 274 – 82 . Google Scholar Crossref Search ADS PubMed WorldCat Dembek TA , Barbe MT , Åström M , Hoevels M , Visser-Vandewalle V , Fink GR et al. . Probabilistic mapping of deep brain stimulation effects in essential tremor . Neuroimage Clin 2017 ; 13 : 164 – 73 . Google Scholar Crossref Search ADS PubMed WorldCat Deuschl G , Schade-Brittinger C , Krack P , Volkmann J , Schäfer H , Bötzel K et al. . A randomized trial of deep-brain stimulation for Parkinson’s disease . N Engl J Med 2006 ; 355 : 896 – 908 . Google Scholar Crossref Search ADS PubMed WorldCat Diedrichsen J . A spatially unbiased atlas template of the human cerebellum . NeuroImage 2006 ; 33 : 127 – 38 . Google Scholar Crossref Search ADS PubMed WorldCat Draganski B , Kherif F , Klöppel S , Cook PA , Alexander DC , Parker GJM et al. . Evidence for segregated and integrative connectivity patterns in the human Basal Ganglia . J Neurosci 2008 ; 28 : 7143 – 52 . Google Scholar Crossref Search ADS PubMed WorldCat Eickhoff SB , Stephan KE , Mohlberg H , Grefkes C , Fink GR , Amunts K et al. . A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data . 2005 ; 25 : 1325 – 35 . WorldCat Eisenstein SA , Koller JM , Black KD , Campbell MC , Lugar HM , Ushe M et al. . Functional anatomy of subthalamic nucleus stimulation in Parkinson disease . Ann Neurol 2014 ; 76 : 279 – 95 . Google Scholar Crossref Search ADS PubMed WorldCat Ewert S , Horn A , Finkel F , Li N , Kühn AA , Herrington TM . Optimization and comparative evaluation of nonlinear deformation algorithms for atlas-based segmentation of DBS target nuclei . NeuroImage 2019 ; 184 : 586 – 98 . Google Scholar Crossref Search ADS PubMed WorldCat Ewert S , Plettig P , Li N , Chakravarty MM , Collins DL , Herrington TM et al. . Toward defining deep brain stimulation targets in MNI space: a subcortical atlas based on multimodal MRI, histology and structural connectivity . NeuroImage 2018 ; 170 : 271 – 82 . Google Scholar Crossref Search ADS PubMed WorldCat Fonov VS , Evans AC , McKinstry RC , Almli CR , Collins DL . Unbiased nonlinear average age-appropriate brain templates from birth to adulthood . NeuroImage 2009 ; 47 : S102 . Google Scholar Crossref Search ADS WorldCat Fox MD , Buckner RL , Liu H , Chakravarty MM , Lozano AM , Pascual-Leone A . Resting-state networks link invasive and noninvasive brain stimulation across diverse psychiatric and neurological diseases . Proc Natl Acad Sci 2014 ; 111 : E4367 – 75 . Google Scholar Crossref Search ADS PubMed WorldCat Fox MDM , Snyder AZA , Vincent JLJ , Corbetta MM , Van Essen DCD , Raichle MEM . The human brain is intrinsically organized into dynamic, anticorrelated functional networks . Proc Natl Acad Sci USA 2005 ; 102 : 9673 – 8 . Google Scholar Crossref Search ADS PubMed WorldCat Friston KJ , Holmes AP , Worsley KJ , Poline JP , Frith CD , Frackowiak RSJ . Statistical parametric maps in functional imaging: a general linear approach . Hum Brain Mapp 2004 ; 2 : 189 – 210 . Google Scholar Crossref Search ADS WorldCat Friston KJ . Functional and effective connectivity: a review . Brain Connect 2011 ; 1 : 13 – 36 . Google Scholar Crossref Search ADS PubMed WorldCat Gousias IS , Rueckert D , Heckemann RA , Dyet LE , Boardman JP , Edwards AD et al. . Automatic segmentation of brain MRIs of 2-year-olds into 83 regions of interest . NeuroImage 2008 ; 40 : 672 – 84 . Google Scholar Crossref Search ADS PubMed WorldCat Hammers A , Allom R , Koepp MJ , Free SL , Myers R , Lemieux L et al. . Three-dimensional maximum probability atlas of the human brain, with particular reference to the temporal lobe . Hum Brain Mapp 2003 ; 19 : 224 – 47 . Google Scholar Crossref Search ADS PubMed WorldCat Horn A . A structural group-connectome in standard stereotactic (MNI) space . Data Brief 2015 ; 5 : 292 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Horn A . The impact of modern-day neuroimaging on the field of deep brain stimulation . Curr Opin Neurol 2019 ; 32 : 511 – 20 . Google Scholar Crossref Search ADS PubMed WorldCat Horn A , Blankenburg F . Toward a standardized structural-functional group connectome in MNI space . NeuroImage 2016 ; 124 : 310 – 22 . Google Scholar Crossref Search ADS PubMed WorldCat Horn A , Kühn AA . Lead-DBS: a toolbox for deep brain stimulation electrode localizations and visualizations . NeuroImage 2015 ; 107 : 127 – 35 . Google Scholar Crossref Search ADS PubMed WorldCat Horn A , Li N , Dembek TA , Kappel A , Boulay C , Ewert S et al. . Lead-DBS v2: Towards a comprehensive pipeline for deep brain stimulation imaging . NeuroImage 2019 ; 184 : 293 – 316 . Google Scholar Crossref Search ADS PubMed WorldCat Horn A , Ostwald D , Reisert M , Blankenburg F . The structural-functional connectome and the default mode network of the human brain . NeuroImage 2014 ; 102 : 142 – 51 . Google Scholar Crossref Search ADS PubMed WorldCat Horn A , Reich M , Vorwerk J , Li N , Wenzel G , Fang Q et al. . Connectivity predicts deep brain stimulation outcome in Parkinson disease . Ann Neurol 2017 ; 82 : 67 – 78 . Google Scholar Crossref Search ADS PubMed WorldCat Hou Y , Yang J , Luo C , Ou R , Song W , Liu W et al. . Patterns of striatal functional connectivity differ in early and late onset Parkinson’s disease . J Neurol 2016 ; 263 : 1993 – 2003 . Google Scholar Crossref Search ADS PubMed WorldCat Irmen F , Horn A , Meder D , Neumann W-J , Plettig P , Schneider G-H et al. . Sensorimotor subthalamic stimulation restores risk-reward trade-off in Parkinson’s disease . Mov Disord 2018 ; 368 : 610 – 11 . WorldCat Jakab A , Blanc R , Berenyi EL , Szekely G . Generation of individualized thalamus target maps by using statistical shape models and thalamocortical tractography . Am J Neuroradiol 2012 ; 33 : 2110 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat Jech R , Mueller K , Urgošík D , Sieger T , Holiga Š , Růžička F et al. . The subthalamic microlesion story in Parkinson’s disease: electrode insertion-related motor improvement with relative cortico-subcortical hypoactivation in fMRI . PLoS ONE 2012 ; 7 : e49056 . Google Scholar Crossref Search ADS PubMed WorldCat Jech R , Urgošík D , Tintěra J , Nebuželský A , Krásenský J , Liščák R et al. . Functional magnetic resonance imaging during deep brain stimulation: a pilot study in four patients with Parkinson’s disease . Mov Disord 2001 ; 16 : 1126 – 32 . Google Scholar Crossref Search ADS PubMed WorldCat Joliot M , Jobard G , Naveau M , Delcroix N , Petit L , Zago L et al. . AICHA: An atlas of intrinsic connectivity of homotopic areas . J Neurosci Methods 2015 ; 254 : 46 – 59 . Google Scholar Crossref Search ADS PubMed WorldCat Kahan J , Urner M , Moran R , Flandin G , Marreiros A , Mancini L et al. . Resting state functional MRI in Parkinson’s disease: the impact of deep brain stimulation on ‘effective’ connectivity . Brain 2014 ; 137 : 1130 – 44 . Google Scholar Crossref Search ADS PubMed WorldCat Klein A , Tourville J . 101 Labeled brain images and a consistent human cortical labeling protocol . Front Neurosci 2012 ; 6 : 171. WorldCat Krauth A , Blanc R , Poveda A , Jeanmonod D , Morel A , Székely G . A mean three-dimensional atlas of the human thalamus: generation from multiple histological data . NeuroImage 2010 ; 49 : 2053 – 62 . Google Scholar Crossref Search ADS PubMed WorldCat Kruschwitz JD , List D , Waller L , Rubinov M , Walter H . GraphVar: a user-friendly toolbox for comprehensive graph analyses of functional brain connectivity . J Neurosci Methods 2015 ; 245 : 107 – 15 . Google Scholar Crossref Search ADS PubMed WorldCat Litvak V , Jha A , Eusebio A , Oostenveld R , Foltynie T , Limousin P et al. . Resting oscillatory cortico-subthalamic connectivity in patients with Parkinson’s disease . Brain 2011 ; 134 : 359 – 74 . Google Scholar Crossref Search ADS PubMed WorldCat Lozano AM , Lipsman N . Probing and regulating dysfunctional circuits using deep brain stimulation . 2013 ; 77 : 406 – 24 . WorldCat Makris N , Goldstein JM , Kennedy D , Hodge SM , Caviness VS , Faraone SV et al. . Decreased volume of left and total anterior insular lobule in schizophrenia . Schizophrenia Res 2006 ; 83 : 155 – 71 . Google Scholar Crossref Search ADS WorldCat Marek K , Jennings D , Lasch S , Siderowf A , Tanner C , Simuni T et al. . The Parkinson progression marker initiative (PPMI) . Prog Neurobiol 2011 ; 95 : 629 – 35 . Google Scholar Crossref Search ADS PubMed WorldCat Mayka MA , Corcos DM , Leurgans SE , Vaillancourt DE . Three-dimensional locations and boundaries of motor and premotor cortices as defined by functional brain imaging: a meta-analysis . NeuroImage 2006 ; 31 : 1453 – 74 . Google Scholar Crossref Search ADS PubMed WorldCat Mädler B , Coenen VA . Explaining clinical effects of deep brain stimulation through simplified target-specific modeling of the volume of activated tissue . Am J Neuroradiol 2012 ; 33 : 1072 – 80 . Google Scholar Crossref Search ADS PubMed WorldCat McIntyre CC , Grill WM . Extracellular stimulation of central neurons: influence of stimulus waveform and frequency on neuronal output . J Neurophysiol 2002 ; 88 : 1592 – 604 . Google Scholar Crossref Search ADS PubMed WorldCat McIntyre CC , Grill WM , Sherman DL , Thakor NV . Cellular effects of deep brain stimulation: model-based analysis of activation and inhibition . J Neurophysiol 2004 ; 91 : 1457 – 69 . Google Scholar Crossref Search ADS PubMed WorldCat McIntyre CC , Hahn PJ . Network perspectives on the mechanisms of deep brain stimulation . Neurobiol Dis 2010 ; 38 : 329 – 37 . Google Scholar Crossref Search ADS PubMed WorldCat Medtronic. Medtronic Announces European Approval of the First and Only Full-Body MR Conditional Deep Brain Stimulation Systems . newsroom.medtronic.com 2015 . WorldCat Morel A . Stereotactic Atlas of the Human Thalamus and Basal Ganglia . Boca Raton : CRC Press ; 2013 . Google Preview WorldCat COPAC Mueller K , Jech R , Růžička F , Holiga Š , Ballarini T , Bezdicek O et al. . Brain connectivity changes when comparing effects of subthalamic deep brain stimulation with levodopa treatment in Parkinson’s disease . Neuroimage Clin 2018 ; 19 : 1025 – 35 . Google Scholar Crossref Search ADS PubMed WorldCat Mueller K , Jech R , Schroeter M . Deep-brain stimulation for Parkinson’s disease . N Engl J Med 2013 ; 368 : 482 – 4 . Google Scholar Crossref Search ADS PubMed WorldCat Neumann W-J , Schroll H , de Almeida Marcelino AL , Horn A , Ewert S , Irmen F et al. . Functional segregation of basal ganglia pathways in Parkinson’s disease . Brain 2018 ; 12 : 366 . WorldCat Nowacki A , Nguyen TA-K , Tinkhauser G , Petermann K , Debove I , Wiest R et al. . Accuracy of different three-dimensional subcortical human brain atlases for DBS -lead localisation . Neuroimage Clin 2018 ; 20 : 868 – 74 . Google Scholar Crossref Search ADS PubMed WorldCat Pauli WM , Nili AN , Tyszka JM . A high-resolution probabilistic in vivo atlas of human subcortical brain nuclei . bioRxiv 2017 : 211201 . WorldCat Power JD , Schlaggar BL , Petersen SE . Studying brain organization via spontaneous fMRI signal . Neuron 2014 ; 84 : 681 – 96 . Google Scholar Crossref Search ADS PubMed WorldCat Rodríguez-Oroz MC , Jahanshahi M , Krack P , Litvan I , Macias R , Bezard E et al. . Initial clinical manifestations of Parkinson’s disease: features and pathophysiological mechanisms . Lancet Neurol 2009 ; 8 : 1128 – 39 . Google Scholar Crossref Search ADS PubMed WorldCat Rubinov M , Sporns O . Complex network measures of brain connectivity: uses and interpretations . NeuroImage 2010 ; 52 : 1059 – 69 . Google Scholar Crossref Search ADS PubMed WorldCat Saenger VM , Kahan J , Foltynie T , Friston K , Aziz TZ , Green AL et al. . Uncovering the underlying mechanisms and whole-brain dynamics of deep brain stimulation for Parkinson’s disease . Sci Rep 2017 ; 7 : 9882 . Google Scholar Crossref Search ADS PubMed WorldCat Schönecker T , Kupsch A , Kühn AA , Schneider G-H , Hoffmann KT . Automated optimization of subcortical cerebral MR imaging-atlas coregistration for improved postoperative electrode localization in deep brain stimulation . Am J Neuroradiol 2009 ; 30 : 1914 – 21 . Google Scholar Crossref Search ADS PubMed WorldCat Setsompop K , Kimmlingen R , Eberlein E , Witzel T , Cohen-Adad J , McNab JA et al. . Pushing the limits of in vivo diffusion MRI for the human connectome project . NeuroImage 2013 ; 80 : 220 – 33 . Google Scholar Crossref Search ADS PubMed WorldCat Simioni AC , Dagher A , Fellows LK . Compensatory striatal-cerebellar connectivity in mild-moderate Parkinson’s disease . Neuroimage Clin 2016 ; 10 : 54 – 62 . Google Scholar Crossref Search ADS PubMed WorldCat Stefurak T , Mikulis D , Mayberg H , Lang AE , Hevenor S , Pahapill P et al. . Deep brain stimulation for Parkinson’s disease dissociates mood and motor circuits: a functional MRI case study . Mov Disord 2003 ; 18 : 1508 – 16 . Google Scholar Crossref Search ADS PubMed WorldCat Sutton R , Kanal E , Wilkoff BL , Bello D , Luechinger R , Jenniskens I et al. . Safety of magnetic resonance imaging of patients with a new Medtronic EnRhythm MRI SureScan pacing system: clinical study design . Trials 2008 ; 9 : 2878 . Google Scholar Crossref Search ADS WorldCat Tzourio-Mazoyer N , Landeau B , Papathanassiou D , Crivello F , Etard O , Delcroix N et al. . Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain . NeuroImage 2002 ; 15 : 273 – 89 . Google Scholar Crossref Search ADS PubMed WorldCat Vanegas Arroyave N , Lauro PM , Huang L , Hallett M , Horovitz SG , Zaghloul KA et al. . Tractography patterns of subthalamic nucleus deep brain stimulation . Brain 2016 ; 139 : 1200 – 10 . Google Scholar Crossref Search ADS PubMed WorldCat Varoquaux G , Craddock RC . Learning and comparing functional connectomes across subjects . NeuroImage 2013 ; 80 : 405 – 15 . Google Scholar Crossref Search ADS PubMed WorldCat Vasques X , Cif L , Hess O , Gavarini S , Mennessier G , Coubes P . Stereotactic model of the electrical distribution within the internal globus pallidus during deep brain stimulation . J Comput Neurosci 2009 ; 26 : 109 – 18 . Google Scholar Crossref Search ADS PubMed WorldCat Vorwerk J , Oostenveld R , Piastra MC , Magyari L , Wolters CH . The FieldTrip-SimBio pipeline for EEG forward solutions . Biomed Eng Online 2018 ; 17 : 37 . Google Scholar Crossref Search ADS PubMed WorldCat Weissenbacher A , Kasess C , Gerstl F , Lanzenberger R , Moser E , Windischberger C . Correlations and anticorrelations in resting-state functional connectivity MRI: a quantitative comparison of preprocessing strategies . NeuroImage 2009 ; 47 : 1408 – 16 . Google Scholar Crossref Search ADS PubMed WorldCat Yao C , Horn A , Li N-F , Lu Y , Fu Z , Wang N et al. . Post-operative electrode location and clinical efficacy of subthalamic nucleus deep brain stimulation in Meige syndrome . Parkinsonism Relat Disord 2019 ; 58 : 40 – 45 . Google Scholar Crossref Search ADS PubMed WorldCat Younce JR , Albaugh DL , Shih Y-YI . Deep brain stimulation with simultaneous fMRI in rodents . J Vis Exp 2014 ; e51271. doi: 10.3791/51271. WorldCat Young CK , Brown AR , Robinson JHB , Tuor UI , Dunn JF , Bland BH et al. . Functional MRI response and correlated electrophysiological changes during posterior hypothalamic nucleus deep brain stimulation . NeuroImage 2011 ; 56 : 35 – 44 . Google Scholar Crossref Search ADS PubMed WorldCat Zalesky A , Fornito A , Bullmore ET . Network-based statistic: identifying differences in brain networks . NeuroImage 2010 ; 53 : 1197 – 207 . Google Scholar Crossref Search ADS PubMed WorldCat Zhang Y , Larcher KM-H , Misic B , Dagher A . Anatomical and functional organization of the human substantia nigra and its connections . Elife 2017 ; 6 : e26653 . Google Scholar Crossref Search ADS PubMed WorldCat Zhou C , Zhang H , Qin Y , Tian T , Xu B , Chen J et al. . A systematic review and meta-analysis of deep brain stimulation in treatment-resistant depression . Prog Neuropsychopharmacol Biol Psychiatry 2018 ; 82 : 224 – 32 . Google Scholar Crossref Search ADS PubMed WorldCat © The Author(s) (2019). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Deep brain stimulation induced normalization of the human functional connectome in Parkinson’s disease
JF - Brain
DO - 10.1093/brain/awz239
DA - 2019-10-01
UR - https://www.deepdyve.com/lp/oxford-university-press/deep-brain-stimulation-induced-normalization-of-the-human-functional-loBjPr0HHO
SP - 3129
VL - 142
IS - 10
DP - DeepDyve
ER -