Connectivity-derived segmentation of the STN in PD patients; correlation with location of stimulation and motor improvement after DBS
Varvara Mathiopoulou
MSc Brain & Cognitive Sciences, UVA
Maarten Bot, MD, PhD Academic Medical Center (AMC)
Abstract
Deep brain stimulation (DBS) of the dorsolateral STN suppresses motor symptoms in Parkinson’s Disease (PD) patients. High-field 7 Tesla T2 MRI provides superior visualisation of the dorsolateral subthalamic nucleus (STN), however it does not indicate the subdivision containing highest connections to cortical motor areas. 7T DWI data enables visualisation of connections between motor, limbic and associative cortical areas and their associated STN subdivisions. The segmented subdivisions can provide motor STN identification and possibly optimisation of DBS therapy. The aim of the current study is to evaluate the STN segmentation using 7 Tesla T2 and DWI MRI, and to implement it in DBS for PD. We manually delineated the STN and cortical (motor, frontal, temporal) areas in native space for 28 PD patients, and we
subsequently used these areas for probabilistic segmentation of motor, associative and limbic STN subdivisions. Intraoperative conebeam CT scans were used for electrode localisation. Subsequently, pre- and post-DBS UPDRS-III scores were used for
evaluating correspondence between location of stimulation and the different segmented STN subdivisions. The segmentation showed both connectivity overlap as well as
distinct differences between the areas. For all 56 STN’s, motor connections were highest in the dorsolateral STN, and frontal connections in the ventromedial part of the
nucleus. Temporal connections were less numbered and mostly situated in medial STN, surrounded by frontal connections. The active electrode contact was located in close proximity to the motor subdivision, mostly slightly anterior to this area. No significant correlation with UPDRS motor outcome was found. The implementation of 7T T2 and DWI for STN segmentation in DBS for PD is feasible and may be used for optimising DBS therapy.
Connectivity-derived segmentation of the subthalamic nucleus in Parkinson’s Disease patients; correlation with location of stimulation and motor improvement
after Deep Brain Stimulation Introduction
Parkinson’s Disease (PD) is a neurodegenerative disorder with motor, behavioral, cognitive, and psychiatric symptoms (McGregor & Nelson, 2019). The subthalamic nucleus (STN) has been a well-established target in Deep Brain Stimulation (DBS), for suppressing motor symptoms (Benabid, 2003; Johnson, Miocinovic, McIntyre, & Vitek, 2008). The STN receives input from the primary motor cortex, and the supplementary motor area (SMA) to its dorsolateral part, which is therefore considered the theoretic optimal DBS target for Parkinson’s disease (Akram et al., 2017; Avecillas-Chasin, Rascón-Ramírez, & Barcia, 2016; Bot et al., 2018).
Accurate DBS electrode placement within the STN is facilitated by using
micro-electrode recordings (MER) and T2 weighted MRI. Identification on MRI can be challenging because the nucleus is small, and is difficult to distinguish from adjacent sustantia nigra (SN), as both are hypointense on T2 weighted MRI. As the different subdivisions of the STN are not readily identifiable on T2 weighted imaging, MER may be conducted during DBS surgery for identification of the sensorimotor subdivision.
Although DBS therapy is considered to be most effective within the dorsolateral STN, in current literature there is dispute considering the internal structure of this nucleus. According to the conventional theory, the nucleus has a tripartite internal division; the sensorimotor area on its dorsolateral part, the cognitive on its central part, and the limbic on its medial tip (Lambert et al., 2012). However, this topographical organization of the nucleus has recently been challenged, revealing gradient border zones, in comparison to a clear tripartite topology (Alkemade et al., 2019; Alkemade & Forstmann, 2014; Keuken et al., 2012).
An effective way to depict the subdivisions of the STN based on cortical connections is through diffusion weighted imaging (DWI). DWI is an in vivo noninvasive technique that measures the diffusion of water molecules in the brain
(Sotiropoulos et al., 2013). Visualisations of white matter tracts connecting STN to specific cortical areas can be used for delineating its functional zones.
Recent studies using probabilistic 3T tractography showed the dorsolateral STN to be connected mainly to cortical motor areas, and found a correlations between STN subdivisions and DBS therapy effectiveness. (Avecillas-Chasin, Alonso-Frech, Nombela, Villanueva, & Barcia, 2019; Vassal et al., 2019). Duchin et al. (2018) compared MER STN and MRI STN length using 7T T2 imaging and found similar lengths. Also,
correlation was found between the volume of tissue activated (VTA) within their 7T 3D STN model and motor improvement after DBS. Last, Plantinga et al. (2018) were able to segment the STN in 7T and demonstrated its three subregions with a certain amount of overlap.
3T imaging often fails to clearly delineate small structures in comparison to 7T MRI due to the weaker field strength (Alkemade et al., 2020). This underlines the possible advantage of high-resolution MRI for DBS, as accurate visualization of fine subcortical structures is considered essential.
In the current study we perform connectivity-derived segmentation of the STN in 7T MRI data of patients that underwent DBS. More specifically, we aim at depicting the different subdivisions of the STN using its various cortical connections. Subsequently, the location of the different subdivisions relative to the active electrode contact will be evaluated. For this, pre and post DBS UPDRS motor scores will be used.
We formulate the following hypotheses: The segmentation of the STN based on its connectivity will result in identifiable subdivisions; a dorsolateral part connecting with the cortical motor areas (M1 and SMA), a ventromedial part connecting with prefrontal cortical areas, and a medial part connecting to temporal cortical areas. We expect the subdivision to show both connectivity overlap, as well as distinct differences between these parts. Last, we predict that the distance between the electrode contact used for DBS therapy will correlate with improvement in UPDRS scores.
Methods Patients
Information was collected from all patients who underwent MER-guided DBS implantation for PD at the Academic Medical Center of Amsterdam between the years January 2019 and February 2020 from whom a complete set of 7 Tesla T2 and DWI was available. All patients underwent formal neurological assessment both pre and
post-operatively, and only patients that did not have any contraindications to surgery and gave no indications of cognitive dysfunctions proceeded to operation. When
dopamine agonists were provided, these were reduced gradually and were stopped three days before the operation, while Levodopa was stopped 12 hours before the surgery.
Preoperative T2 and Diffusion Weighted MRI acquisition
Patients underwent 7.0 T (Philips Healthcare, Cleveland, OH, USA) pre-operative T2 and DW imaging with a 32 channel receive coil, using the following image
acquisition parameters:
T2-Weighted MRI: 3D sagittal with Turbo spin echo (TSE) imaging, TR = 3000 ms, TE = 324 ms, FOV = 250 x 250 x 190, Flip angle = 100, voxel size = 0.7 mm, scan duration = 7 mins 33 sec.
Diffusion Weighted Imaging: TR = 6084 ms, TE = 70 ms, b= 1000 s/mm2, 32
directions, FOV = 140 x 177 x 110, voxel size = 1.5 mm. Diffusion data were acquired with reverse phase encoding which resulted in pairs of images with distortions in opposite directions (left to right and right to left).
Surgical Procedure & Electrode Contact Localization
The current surgical procedure of DBS placement in our center is described in detail elsewhere (Bot et al., 2018). Before October 2019, patients participated in a study comparing awake versus asleep DBS. Allocation was done randomly. From
October 2019 patients were operated under general anesthesia. In all patients MER was performed for dorsolateral STN identification. Both the Leksell G-frame and Vantage
(Elekta AB, Stockholm, Sweden) system were used. All patients received direct placement of the infra-clavicular stimulator.
STN identification was done using 7 Tesla T2 weighted imaging. The STN was considered the hypointense convex shaped area lateral to the red nucleus on axial and coronal orientated imaging. On axial orientated imaging the anterior border of the maximum diameter of RN was used for optimization (Fig. 1). DBS electrode placement was verified using O-arm Imaging System (Medtronic Inc, Minneapolis, Minnesota, USA). The theoretic optimal effective location was defined at 2.8 mm lateral, 1.7 mm anterior, 2.5 mm superior to the medial STN border (Bot et al., 2018).
Figure 1 . Axial midbrain section showing identification of the left STN at maximal
diameter of the RN in 7.0 Tesla T2 MRI sequence. The STN is located anterolateral to the RN. The horizontal line defines the anterior border of the RN. Both lines coincide at the medial STN border.
Diffusion Preprocessing
T2 and Diffusion Weighted Imaging Scans were converted from DICOM (Digital Imaging and Communications in Medicine) files to NiFTI volumes using MRIcron’s DCM2NIIX converter (Li, Morgan, Ashburner, Smith, & Rorden, 2016). The corresponding diffusion gradient direction values and vectors were also extracted.
Afterwards, a number of preprocessing steps were applied to the data. First, image denoising to enhance the data quality and Gibbs artifact removal were applied using the MRtrix3 software package (Tournier et al., 2019). Gibbs artifacts result from
the Fourier transformation (form frequency to image domain). Furthermore, we applied the following corrections using FSL v5.0; Topup was used to correct for susceptibility induced distortions (geometric mismatch between the structural and the diffusion image), and brain extraction was performed using BET. Eddy was used for correcting eddy currents and movement in the diffusion data. DTIFIT was used for fitting a diffusion tensor model in each voxel. BEDPOSTX was ran to fit the probabilistic diffusion model on the corrected data (Akram et al., 2018; Liebrand et al., 2019).
CT & MRI co-registration
The intra-operative O-arm CT scan depicting the electrodes was co-registered using FLIRT to 7T T2 or 3T T1 weighted images. Coordinates of electrode contact on CT were determined by using the center of the active contact used for chronic
stimulation. 7T Diffusion weighted images were linearly registered with FLIRT to the structural images with 12 DOF and mutual information as the cost function.
STN segmentation
Region of interest definition. Regions of interest were delineated manually in native (patient) space. The right and left STN were masked using the ITKsnap tool (version 3.8.0; Yushkevich et al., 2006) on diffusion data with the T2 volume overlapped. This ensured that the masking was done directly in diffusion space. By using axial, coronal and sagittal orientated imaging with optimised image contrast, the nucleus was depicted as the hypointense structure anterolateral to the red nucleus (axial plane) and superior to the substantia nigra (coronal plane). Voxels containing >50% of intensity were included as part of the STN (Fig. 2). The STN mask was used as seed area, and its cortical connections were used to guide the selection of cortical regions; motor areas (M1 and SMA), associative (prefrontal cortex), and limbic (temporal). Masking of the cortical areas was done on diffusion data using the masking tool of FSLview.
Figure 2 . Segmentation of the STN in two patients (1, 2) in T2 7T structural images.
Left planes show before, and right planes show after the segmentation. Top planes of each patient are axial slices, bottom planes of each patient are coronal. STN is located anterolateral to the red nucleus (RN) and superior to the substantia nigra (SN).
Pictures show that different contrasts were needed for optimal delineation of the STN, as well as the intra-subject variability of the STN in size and shape.
Probabilistic Tractography. Tractography of the STN was ran using
PROBTRACKX Probabilistic Tracking of FSL (v5.0), by defining a single mask as the
seed space (right or left STN), and the default parameters; number of samples = 5000, curvature threshold = 0.2, maximum number of steps = 2000, step length = 0.5 mm, volume fraction threshold = 0.01. Waypoint was applied independently to both
directions. Probabilistic tractography creates streamlines in an iterative manner in each seed voxel, reconstructing paths of minimum impediment of water diffusion in white matter.
Connectivity derived segmentation of the STN resulted by running
PROBTRACKX Probabilistic Tracking using the ipsilateral cortical ROIs as
classification targets. Seed voxels of the STN were classified according to the
probability of being connected to each cortical ROI. This gave output images, where each voxel in the seed mask had value equal to the number of samples beginning from this voxel and ending to the target area, creating areas within the STN that showed higher connectivity to each cortical ROI. After segmenting the STN in motor, frontal, and temporal, we visually inspected all patients, looking at the STN at the depth of the widest diameter of the red nucleus. A threshold of 25% was applied for each STN area.
Outcome measures
Motor improvement. Motor improvement was evaluated using preoperative and six months post-operative Unified Parkinson’s Disease Rating Scale Part 3 (UPDRS-III; Fish, 2011) scores. Patients were assessed in a total of six conditions; Preoperatively OFF/ON medication, and in four conditions postoperatively ON/OFF medication and ON/OFF stimulation. Right side and left side scores were noted for each condition. Percentage of improvement between two conditions was calculated as follows: (score difference between original and new value * 100) / original value.
UPDRS motor score assessment was performed at baseline and six months after surgery. Standardized assessments were done during off-drug and on-drug phase, six-month assessment was done with the stimulator turned on. Left and right-sided
motor scores (UPDRS part III items 3.3 -3.8) were separately evaluated. Improvement in UPDRS motor score between off-drug/on-stimulation after six months, and baseline off-drug was used to categorise contralateral body sides into three groups: (1)
non-responding (less than 30%), (2) responding (between 30% and 70%) and (3) optimally responding (more than 70%).
Distance between active contact and motor STN. From the
connectivity-derived segmentation of the STN, we extracted for each hemisphere the coordinates of the voxel with the maximum seed connectivity to the motor and frontal cortical areas, applying a threshold of at least 80%. After co-registering the CT scan as described above, we calculated the Euclidean distance between the voxel with the higher probability of connections to motor/frontal cortex and the (center of) active electrode coordinates using the following formula: q(∆X)2+ (∆Y )2+ (∆Z)2.
Euclidean distance represents the absolute error in 3D space; small values indicate that the electrode’s active contact was situated closer to the area in question, whereas larger values indicate that the electrode’s active contact was located further from it. In Figure 3 we show two examples of electrode contact location and the voxel with the maximum seed connectivity to the different cortical areas.
Figure 3 . Distance between electrode contact and left motor/frontal STN area in two
different patients (A, B) in coronal (top) and axial (bottom) planes. The location of the active electrode contact is the center of the two green crossing lines, whereas the voxel with higher seed connectivity to the motor cortex (A) and the frontal cortex (B) are shown in red. Distance between the two was calculated using the euclidean error.
Results Patients
From a total of 36 patients, 28 patients (8 female) were included in the study; seven were excluded due to distortions in the images, and in one patient image quality was insufficient for STN border identification. Mean age of patients included for further analyses was 61.55 [9.7] years, and the mean age at the time of the surgery was 60.72 [9.78] years.
Twenty-one patients underwent DBS surgery under general anesthesia and seven awake. There was a significant improvement in pre-operative UPDRS-III scores: mean score OFF medication = 46.55 [13.92], and ON medication 19.03 [10.13]; mean
improvement = 58.84% [18.12], (p < .001) . The sample of the post-operative scores was smaller (N =22), because for six patients six-month post-operative UPDRS scores were not available. There was a significant improvement with DBS (Off medication -On stimulation Off medication condition): mean change = -50.18% [17.79], (p < .001). See Figure 4 for graphs of mean scores in all conditions.
N = 29 N = 29 *** *** −58.84% [18.12] −58.84% [18.12] 19.03 [10.13] 19.03 [10.13] 46.55 [13.92] 46.55 [13.92] 0 20 40 60
Off Medication ON Medication
Conditions
Mean Score
Mean Pre−operative UPDRS−III Scores
(a) N = 22 N = 22 N = 22 N = 22 40.36 [12.02] 40.36 [12.02] 40.36 [12.02] 40.36 [12.02] 20.95 [10.87]20.95 [10.87]20.95 [10.87]20.95 [10.87] *** *** *** *** −50.08% [23.42] −50.08% [23.42] −50.08% [23.42] −50.08% [23.42] 31.05 [16.73] 31.05 [16.73] 31.05 [16.73] 31.05 [16.73] 14.95 [8.88]14.95 [8.88]14.95 [8.88]14.95 [8.88] *** *** *** *** −64.02% [17.51] −64.02% [17.51] −64.02% [17.51] −64.02% [17.51] 0 20 40 60
Med OFF/DBS OFF Med OFF/DBS ON Med ON/DBS OFF Med ON/DBS ON
Conditions
Mean Score
Mean Post−operative UPDRS−III Scores
N = 22 N = 22 N = 22 16.57 [10.93], N = 4 16.57 [10.93], N = 4 16.57 [10.93], N = 4 45.35 [11.38], N = 11 45.35 [11.38], N = 11 45.35 [11.38], N = 11 76.67 [5.53], N = 7 76.67 [5.53], N = 7 76.67 [5.53], N = 7 0 25 50 75 100
Optimal Responders Average Responders Poor responders Response to DBS
Mean Score
Mean post−operative UPDRS−III in three subgroups
(c)
Figure 4 . Mean UPDRS-III scores. Graph 4a: Mean UPDRS-III scores pre-operatively
Off and On Medication. Graph 4b: Mean UPDRS-III scores post-operatively in four conditions; Off medication – Off stimulation, Off medication – On stimulation, On medication – Off stimulation, On medication – On stimulation. Mean change in statistically significant bars is displayed in percentage of change in comparison to baseline (pre-operative off medication condition). Graph 4c: Mean UPDRS motor scores Off medication - On stimulation post-operatively in optimally (left bar), average (middle), and poorly (right) responding patients. In the bottom of all bars the mean scores and their standard deviations are displayed. Errors indicate the standard deviations of the mean.
STN segmentation and tractography
The mean size of the segmented STN was 168.51 mm3 [57.67] for the right nucleus and 137.68 mm3 [33.43] for the left. There was a statistically significant difference between the two nuclei (p = .006) . There was a great variation in shape and size, both within and between patients (Figure 2). Running probabilistic tractography with each STN as seed region, indicated that the nucleus was consistently connected to the primary motor cortex (M1), the SMA, the prefrontal cortex, and a temporal region (Figure 5).
Figure 5 . STN segmentation was based on its connections to the cortex. Here we
present the tracts of the right STN of one patient. Left; coronal view of associative (frontal) connections. Middle; axial view of motor connections. Right; sagittal view of limbic (temporal) connections. The majority of the tracts connect to the temporal lobe, the M1 and SMA, and the frontal lobe.
After running connectivity-based seed classification, connections were found
within the entire nucleus to the M1, the SMA and the frontal lobe, and few connections to the temporal lobe. Therefore, we used a threshold of 25% for M1, SMA and frontal, and a threshold of 70-80% for temporal. This gave the following mean percentages of seed voxels within each nucleus: For the right STN the motor connections consisted of the 27.42% [19.16] of the nucleus, the frontal 48.42% [19.44], and the temporal 5.44% [3.41]. For the left STN the motor connections consisted of the 32.55% [13.15], the frontal 48.22% [24.21], and the temporal 7.35% [7.32]. Highest density in motor
connections was observed in the dorsolateral STN, frontal in the ventromedial part, and temporal connections in medial or ventromedial part of the STN (Figure 6).
Electrode placement & correlation with clinical outcome
The horizontal line coinciding with the anterior RN border corresponded in 15 patients to the dorsolateral STN, in nine patients to the area between the dorsolateral and the ventral STN, while in one patient, the stimulation guideline coincided with the ventromedial STN (see Figure 6 for an overview of nine patients). This indicated that the majority of the patients were indeed stimulated in the motor STN, whereas in a considerable number of patients (ten) the electrode was anteriorly to the area with highest seed connectivity to motor areas.
Measuring the Euclidean distance of the active contact coordinates and the voxel with the highest seed motor/frontal connectivity, the following mean distances were noted: 4.35 mm [2.1] to the right motor; 4.82 mm [1.99] to the left motor area; 4.68 mm [2.72] to the right frontal; 3.88 mm [1.39] to the left frontal area. Comparing this
distance and the clinical outcome as measured by the UPDRS-III score, there was not statistically significant correlation between motor improvement and distance to the right motor STN (p = .296) or the left motor STN (p = .933). Also, there was no significant correlation between right side improvement and the left STN distance (p = .466), or the left side improvement and the right STN distance (p = .151).
All patients received high frequency stimulation (100/130 Hz). The pulse width was 60 µs for all patients. No correlation was found between the distances and the stimulation amplitude in right (p = .155) or left (p = .717) STN.
Figure 6 . Connectivity derived segmentation of the STN in nine patients in coronal
(top) and axial (bottom) section, on the level of widest diameter of RN. Frontal
connections are depicted in green, motor in blue, and temporal in red. In all STN’s, the frontal connections are located in the ventromedial part of the nucleus, motor
connections in the dorsolateral, and temporal in ventral or ventromedial STN. Green horizontal line coincides with the anterior border of the red nucleus, which was used for determining the target for electrode placement. Using the red nucleus for guiding STN target localization results in different areas of the STN segmentation. For instance, in patient 6 the target is in motor STN; in patient 7 the target is in frontal STN, and in patient 8 the right target is in motor, and left in frontal.
Discussion
In the current study we determined the structural connectivity of the STN in PD patients using 7.0 Tesla imaging of PD. The STN was segmented based on its
connections to the cortex, and correlation with electrode placement and clinical outcome was also evaluated.
STN identification was well feasible in 7T T2 weighted images as a hypointense structure anterolateral to the RN and dorsal to the SN. Identifying the STN is
facilitated using 7T in comparison to lower field strengths (Alkemade et al., 2020; Bot et al., 2018). The left STN was found to be significantly smaller than the right, although during the segmentation process the size of the STN was not known. The found difference in segmentation size can be further evaluated by using two independent researchers and by calculating the inter-rater reliability (Keuken & Forstmann, 2015). Nevertheless, manual segmentation seems to be the golden standard for such subcortical structures, making the findings individualized and highly applicable in clinical practice.
Regarding the connectivity-derived segmentation of the STN, we identified, as hypothesized, a motor subdivision in the dorsolateral part (connecting to the M1 and the SMA), an associative frontal area in the ventromedial, and a limbic temporal area in the medial or ventromedial part of the nucleus. These findings are in line with previous research on the topographical organization of the STN (Lambert et al., 2012). However, a significant overlap was observed between the zones (especially motor and frontal), a finding that supports recent studies that challenge the conventional model of a strict tripartite organization of the STN (Alkemade et al., 2019; Alkemade &
Forstmann, 2014; Keuken et al., 2012). Moreover, a variation in location and size of these subdivisions was noted both between and within patients. This variation could be of importance when the level of maximal RN diameter is used for STN target
identification in DBS surgery, and it underlines the significance of implementing patient-specific anatomical models in DBS.
The electrode contact was hypothesized to be located in the motor area of the STN. As is shown in Figure 6, the electrode was often not located exactly within the
area with highest density of motor connections, but slightly more anterior; within the part of the nucleus with highest density of frontal connections. Possibly 7T STN segmentation can optimize electrode placement when using the anterior border of the RN.
The distance between the electrode contact to the motor/frontal STN was
measured with the absolute error (Euclidean distance) of the active contact coordinates and the coordinates of the voxel that had the highest seed motor/frontal connectivity. There was no correlation between this distance and the clinical outcome post-surgery as measured by the UPDRS-III score. Investigating whether this distance impacts the stimulation parameters (i.e. whether in cases where the electrode was located further, higher stimulation amplitude was needed) did not result in significant findings. Multiple possibilities could explain these findings. First, since in ten patients the electrodes were placed more anterior, it would be interesting to include cognitive scores (e.g. mood disorders, cognitive difficulties) for further research, as this can possibly show a correlation with electrode location and distance to highest density of associative connections. Second, even though there is no significant correlation between the distance of the electrode and UPDRS scores, it is possible that the improvement post-operative would be larger, if the electrode was located within the motor STN. Third, it has been reported before (Akram et al., 2018) that different motor symptoms (tremor, bradykinesia, rigidity) have distinct optimal stimulation points, therefore it could be interesting to examine the relation of the electrode distance with regard to the dominant motor symptoms of each patient. However, given the noticeable overlap of the connections and the relatively big size of the electrode, it might be difficult to evaluate this. Last, a methodological consideration is whether the distance between the current contact and the voxel with the highest seed connectivity is a too conservative approach; in most patients this voxel was located in the most posterior motor part of the STN, increasing the distance to the electrode contact even in cases when it was arguably located within the motor STN subdivision. This could possibly have influenced the found correlations. Some alternatives for measuring electrode distance would be
modeling the Voxel of Tissue Activated (VTA), defining the distance of the electrode upon entrance in the STN (Duchin et al., 2018), and measuring the distance between the active contact and the closest border of the motor STN.
Limitations
There are a number of limitations to be considered for future investigations. First, we found a significant difference between the right and the left nucleus that was not sufficiently interpreted; this discrepancy could be solved by having two independent raters in manual segmentation. Second, we had a limited sample of post-operative scores, which made it difficult to separate patients between poor, average, and optimal responders, and draw sufficient statistically conclusions. In future studies when more post-operative scores will be readily available, the correlation to clinical outcome might be more enlightening. Future research could also investigate whether there is a
correlation between the distance of the electrode to the motor/frontal STN and neuropsychological scores post-operatively, in order to further examine the relation of the electrode contact location and unwanted side-effects. Last, measuring the distance of the electrode contact to the voxel with the highest density of motor connections might have been a fairly conservative approach that makes this measurement not informative in combination with clinical outcome.
Conclusion
The findings of the current study have significant implications for understanding the optimal stimulation location within the STN for PD patients. 7T imaging is proven very promising in accurately mapping subcortical structures that are not readily
depicted in MRI, and especially the STN. The findings of this study complement those of recent literature that investigates clinical applications of high-resolution brain imaging, showing insights in patient specific STN anatomy, which can be used in evaluation of DBS electrode localization. This can possibly improve STN targeting for DBS, as well as motor outcome after surgery in PD patients.
References
Akram, H., Dayal, V., Mahlknecht, P., Georgiev, D., Hyam, J., Foltynie, T., . . . others (2018). Connectivity derived thalamic segmentation in deep brain stimulation for tremor. NeuroImage: Clinical, 18 , 130–142. doi:
https://doi.org/10.1016/j.nicl.2018.01.008
Akram, H., Sotiropoulos, S. N., Jbabdi, S., Georgiev, D., Mahlknecht, P., Hyam, J., . . . others (2017). Subthalamic deep brain stimulation sweet spots and hyperdirect cortical connectivity in parkinson’s disease. Neuroimage, 158 , 332–345. doi: https://doi.org/10.1016/j.neuroimage.2017.07.012
Alkemade, A., de Hollander, G., Miletic, S., Keuken, M. C., Balesar, R., de Boer, O., . . . Forstmann, B. U. (2019). The functional microscopic neuroanatomy of the human subthalamic nucleus. Brain Structure and Function, 224 (9), 3213–3227. doi: https://doi.org/10.1007/s00429-019-01960-3
Alkemade, A., & Forstmann, B. U. (2014). Do we need to revise the tripartite
subdivision hypothesis of the human subthalamic nucleus (stn)? Neuroimage, 95 , 326–329. doi: https://doi.org/10.1016/j.neuroimage.2014.03.010
Alkemade, A., Isaacs, B., Mulder, M., Groot, J., Van Berendonk, N., Lute, N., . . . Forstmann, B. (2020). 3 versus 7 tesla magnetic resonance imaging for parcellations of subcortical brain structures. bioRxiv. doi:
https://doi.org/10.1101/2020.07.02.184275
Avecillas-Chasin, J. M., Alonso-Frech, F., Nombela, C., Villanueva, C., & Barcia, J. A. (2019). Stimulation of the tractography-defined subthalamic nucleus regions correlates with clinical outcomes. Neurosurgery, 85 (2), E294–E303. doi: https://doi.org/10.1093/neuros/nyy633
Avecillas-Chasin, J. M., Rascón-Ramírez, F., & Barcia, J. A. (2016). Tractographical model of the cortico-basal ganglia and corticothalamic connections: Improving our understanding of deep brain stimulation. Clinical Anatomy, 29 (4), 481–492. doi: https://doi.org/10.1002/ca.22689
in neurobiology, 13 (6), 696–706. doi: https://doi.org/10.1016/j.conb.2003.11.001
Bot, M., Schuurman, P. R., Odekerken, V. J., Verhagen, R., Contarino, F. M., De Bie, R. M., & van den Munckhof, P. (2018). Deep brain stimulation for parkinson’s disease: defining the optimal location within the subthalamic nucleus. Journal of
Neurology, Neurosurgery & Psychiatry, 89 (5), 493–498. doi:
http://dx.doi.org/10.1136/jnnp-2017-316907
Duchin, Y., Shamir, R. R., Patriat, R., Kim, J., Vitek, J. L., Sapiro, G., & Harel, N. (2018). Patient-specific anatomical model for deep brain stimulation based on 7 tesla mri. PloS one, 13 (8), e0201469. doi:
https://doi.org/10.1371/journal.pone.0201469
Fish, J. (2011). Unified parkinson’s disease rating scale. In J. S. Kreutzer, J. DeLuca, & B. Caplan (Eds.), Encyclopedia of clinical neuropsychology (pp. 2576–2577). New York, NY: Springer New York. Retrieved from
https://doi.org/10.1007/978-0-387-79948-31836 doi: 10.1007/978-0-387-79948-31836
Johnson, M. D., Miocinovic, S., McIntyre, C. C., & Vitek, J. L. (2008). Mechanisms and targets of deep brain stimulation in movement disorders. Neurotherapeutics,
5 (2), 294–308. doi: https://doi.org/10.1016/j.nurt.2008.01.010
Keuken, M. C., & Forstmann, B. U. (2015). A probabilistic atlas of the basal ganglia using 7 t mri. Data in brief , 4 , 577–582. doi:
https://doi.org/10.1016/j.dib.2015.07.028
Keuken, M. C., Uylings, H., Geyer, S., Schäfer, A., Turner, R., & Forstmann, B. U. (2012). Are there three subdivisions in the primate subthalamic nucleus?
Frontiers in Neuroanatomy, 6 , 14. doi: https://doi.org/10.3389/fnana.
2012.00014
Lambert, C., Zrinzo, L., Nagy, Z., Lutti, A., Hariz, M., Foltynie, T., . . . Frackowiak, R. (2012). Confirmation of functional zones within the human subthalamic nucleus: patterns of connectivity and sub-parcellation using diffusion weighted imaging.
2011.11.082
Li, X., Morgan, P. S., Ashburner, J., Smith, J., & Rorden, C. (2016). The first step for neuroimaging data analysis: Dicom to nifti conversion. Journal of neuroscience
methods, 264 , 47–56. doi: https://doi.org/10.1016/j.jneumeth.2016.03.001
Liebrand, L., Caan, M., Schuurman, P., van den Munckhof, P., Figee, M., Denys, D., & van Wingen, G. (2019). Individual white matter bundle trajectories are associated with deep brain stimulation response in obsessive-compulsive disorder. Brain
stimulation, 12 (2), 353–360. doi: https://doi.org/10.1016/j.brs.2018.11.014
McGregor, M. M., & Nelson, A. B. (2019). Circuit mechanisms of parkinson’s disease.
Neuron, 101 (6), 1042–1056. doi: https://doi.org/10.1016/j.neuron.2019.03.004
Plantinga, B. R., Temel, Y., Duchin, Y., Uludağ, K., Patriat, R., Roebroeck, A., . . . others (2018). Individualized parcellation of the subthalamic nucleus in patients with parkinson’s disease with 7t mri. Neuroimage, 168 , 403–411. doi:
https://doi.org/10.1016/j.neuroimage.2016.09.023
Sotiropoulos, S. N., Jbabdi, S., Xu, J., Andersson, J. L., Moeller, S., Auerbach, E. J., . . . others (2013). Advances in diffusion mri acquisition and processing in the human connectome project. Neuroimage, 80 , 125–143. doi:
https://doi.org/10.1016/j.neuroimage.2013.05.057
Tournier, J.-D., Smith, R., Raffelt, D., Tabbara, R., Dhollander, T., Pietsch, M., . . . Connelly, A. (2019). Mrtrix3: A fast, flexible and open software framework for medical image processing and visualisation. NeuroImage, 202 , 116137. doi: https://doi.org/10.1016/j.neuroimage.2019.116137
Vassal, F., Dilly, D., Boutet, C., Bertholon, F., Charier, D., & Pommier, B. (2019). White matter tracts involved by deep brain stimulation of the subthalamic nucleus in parkinson’s disease: a connectivity study based on preoperative
diffusion tensor imaging tractography. British Journal of Neurosurgery, 1–9. doi: https://doi.org/10.1080/02688697.2019.1701630
Yushkevich, P. A., Piven, J., Hazlett, H. C., Smith, R. G., Ho, S., Gee, J. C., & Gerig, G. (2006). User-guided 3d active contour segmentation of anatomical structures:
significantly improved efficiency and reliability. Neuroimage, 31 (3), 1116–1128. doi: https://doi.org/10.1016/j.neuroimage.2006.01.015
