Connectivity Patterns of Deep Brain Stimulation of the Subthalamic Nucleus in Parkinson’s Disease
Silvina G Horovitz1, Nora Vanegas-Arroyave1,2, Ling Huang2, Peter M Lauro2, Paul A Taylor3,4,5, Mark Hallett1, Kareem A Zaghloul6, and Codrin Lungu2

1Human Motor Control Section, National Institute of Neurological Disorders and Stroke, NIH, Bethesda, MD, United States, 2Office of the Clinical Director, National Institute of Neurological Disorders and Stroke, NIH, Bethesda, MD, United States, 3Scientific and Statistical Computing Core, National Institutes of Health, Bethesda, MD, United States, 4Department of Human Biology, Faculty of Health Sciences, University of Cape Town, MRC/UCT Medical Imaging Research Unit, Cape Town, South Africa, 5African Institute for Mathematical Sciences, Muizenberg, South Africa, 6Surgical Neurology Branch, National Institute of Neurological Disorders and Stroke, NIH, Bethesda, MD, United States


Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is an effective surgical treatment for Parkinson’s Disease (PD). However, its mechanism is unclear. We have developed a pipeline for processing diffusion tensor imaging (DTI) data in DBS patients, and applied it to analyze 22 PD patients implanted with bilateral STN-DBS. With this approach, we have identified the motor nuclei of the thalamus and the superior frontal cortex as the most common targets and predictors of clinical benefits.


Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is an effective surgical treatment for Parkinson’s Disease (PD). The therapeutic mechanism of STN-DBS remains unclear. Diffusion tensor imaging (DTI) is an MRI based method that allows the evaluation of white matter pathways in vivo. We hypothesize that the investigation of relevant networks using DTI will add to the understanding of the mechanism of action for DBS. Here we present a method for DBS-DTI analysis and study the structural connectivity patterns of clinically effective STN stimulation in PD patients.


We studied the stimulator-brain connectivity patterns in twenty-two patients with idiopathic PD (12 females, age 58.8±8.6 years, disease duration 13.4±6.3 years). All underwent standard STN-DBS implantation (DBS lead Model 3389, Medtronic, Minneapolis, Minnesota, USA) bilaterally. One month after surgery, patients underwent initial monopolar screening of the four electrodes of each lead. T1-, T2- and diffusion-weighted (b=1000 s/mm2, 32 directions) MR images were collected on at 3T (Philips Achieva XT) preoperatively, and T1-weighted MR (1.5T) (Philips Achieva XT), and CT images (Siemens SOMATOM Definition Flash) were collected postoperatively. We developed a publicly available pipeline for DBS-DTI data analysis within the AFNI environment (DBSproc1). After pre-processing images in TORTOISE2, our pipeline performs the following actions (Figure 1):

1) multi-modal image co-registration within native patient space,

2) electrode reconstruction from T2-space-linked postoperative CT images,

3) estimation of volume of tissue activated (VTA) for each electrode based on stimulation voltage and impedance,

4) computation of connectivity between the VTA and brain ROIs from FreeSurfer3 segmentation assessed with probabilistic tractography using FATCAT4 (sample tractography result in Figure 2).

For clinically effective electrodes, we identified the most prevalent ROIs for subcortical and cortical regions performing a binary analysis of connections (>1% of total streamlines from the VTA). We then assessed whether connectivity strength to these regions could be an indicator of clinical effectiveness. Specifically, for each ROI and within each lead, we used a simple connection-based classifier to identify whether the electrode with highest connectivity to the ROI indicates clinical effectiveness.


At three and six months after surgery, all patients demonstrated significant benefits (pre- vs post-operative, p<0.0001) with STN-DBS, as illustrated by reductions in UPDRS-IV (complications of therapy assessment) and LEDD (requirement of dopaminergic medications) scores. An initial screening session identified 123 clinically effective electrodes in 40 hemispheres. Four hemispheres were undetermined because the contralateral body-side was asymptomatic. Proper registration and electrode localization were successful in all patients (quality assessment in Figure 3).

Binary analysis of the probabilistic tractography of the VTA showed that 96% of clinically effective electrodes had connections to the thalamus and 98% to the brainstem. Within the cortical regions, the superior frontal gyrus (SFG) was the most prevalent connection (56%)(Figure 4).

Connection-based classifiers were tested for 38 leads that contained at least one electrode with clinical benefits. Classification accuracy (highest connectivity was beneficial electrode) was 87% for the thalamus, 71% for the brainstem and 94% for the SFG.

In a post-hoc analysis, we assessed connectivity between the VTA and thalamic nuclei (derived from a Talairach-Tournoux atlas) for all of the clinically effective electrodes using tractography. 95% of clinically effective electrodes reached at least one motor thalamic nucleus (93% ventrolateral and 85% ventroanterior nucleus).


Thalamus and SFG were identified as the most “desirable” targets in STN-DBS. Our findings are in line with previous studies suggesting that the modulation of white matter structures in the vicinity of the dorsal STN may be implicated in its mechanism of action. In agreement with electrophysiological and PET studies,5,6 projections to specific motor subnuclei of the thalamus might be particularly implicated in the motor benefits observed with STN-DBS. Further brainstem segmentation/parcellation might allow the differentiation of individual subnuclei associated with relevant motor circuits in the basal ganglia. Further research is warranted.


NINDS Intramural Program


1.Lauro PM, Vanegas-Arroyave N, Huang L, Taylor PA, Zaghloul KA, Lungu C, Saad ZS, Horovitz SG. DBSproc: an open source process for DBS electrode localization and tractographic analysis. Hum Brain Mapp. 2015.

2.Pierpaoli C, Walker L, Irfanoglu MO, Barnett A, Basser P, Chang LC, Koay C, Pajevic S, Rohde G, Sarlls J, Wu M. TORTOISE: an integrated software package for processing of diffusion MRI data. ISMRM 18th annual meeting, 2010; Stockholm, Sweden, #1597.

3.Taylor PA, Saad ZS. FATCAT: (an efficient) functional and tractographic connectivity analysis toolbox. Brain Connect. 2013;3:525-535.

4.Fischl B. FreeSurfer. NeuroImage. 2012; 62:774-781.

5.Xu W, Russo GS, Hashimoto T, Zhang J, Vitek JL. Subthalamic nucleus stimulation modulates thalamic neuronal activity. J Neurosci. 2008;28:11916-11924.

6.Bradberry TJ, Metman LV, Contreras-Vidal JL, van den Munchkhof P, Hosey LA, Thompson JL, Schulz GM, Lenz F, Pahwa R, Lyons KE. Common and unique responses to dopamine agonist therapy and deep brain stimulation in Parkinson’s disease: an H215O PET study. Brain Stimul. 2012;5:605-615.


Flowchart illustrating the processing pipeline of DBSproc. Bold text represent input images in original space. Rectangle boxes represent major processing steps.

Single subject example illustrating reconstructed DBS lead and the tractography pattern of a clinically effective electrode. VTA is enlarged for visualization and shown in black. The contralateral thalamus is in red and the superior frontal gyrus is in yellow.

Single subject example illustrating the quality check for image alignment and electrode reconstruction. Right (A) and left (B) hemisphere electrode. The postop T1w image transformed to target T2w space was used as an underlay, with red overlay representing VTAs and the green overlay representing the subthalamic nucleus.

Histogram illustrating the most prevalent regions that VTA of the clinically effective electrodes were connected to. Brainstem showed 98% hits, followed by the thalamus (96%). The most common hit for cortical regions was the superior frontal gyrus with 56% hits. STN and SN results might be influenced by electrode proximity.

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)