PARTICLE

 Save as PDF

Abstract

Aim: Many resting-state functional magnetic resonance imaging (rs-fMRI) studies have provided evidence that repetitive transcranial magnetic stimulation (rTMS) exerts treatment effects via functional connectivity (FC) from a superficial stimulation target to a deep effective region. The dorsolateral subthalamic nucleus (DL-STN) is an effective target in deep brain stimulation surgery for Parkinson’s disease (PD), but its targeting highly depends on well-trained neurosurgeons and is not easily used for FC-guided rTMS. We aimed to devise a method for automatically localizing the DL-STN, and further develop a one-stop plug-in of rs-fMRI FC analysis to assist future individualized FC-guided rTMS.

Methods: Based on structural and iron-sensitive MRI of 78 participants, two raters defined the DL-STN coordinates with very high reliability. The averaged coordinates in the standard Montreal Neurological Institute (MNI) space were: left DL-STN, x: -11.89 ± 0.82, y: -14.51 ± 1.00, and z: -6.40 ± 1.01 and the right DL-STN, x: 12.53 ± 0.77, y: -13.97 ± 0.86, and z: -6.57 ± 0.99. As the individual distances from the averaged coordinates were less than 3 mm (within one voxel for most rs-fMRI studies) for all 78 participants, we defined the average coordinates as AutoSTN. We then developed a one-stop plug-in named Connectivity and Coordinates Converting Assistant Toolbox (CC-CAT) and performed AutoSTN FC analysis.

Results: The AutoSTN seed showed significant FC with the motor cortices in all participants.

Conclusion: The AutoSTN-based rs-fMRI FC could guide future rTMS on PD. The one-stop plug-in CC-CAT can be used for any FC-guided rTMS treatment.

Keywords

Dorsolateral subthalamic nucleus, automatic targeting, Parkinson’s disease, functional connectivity, repetitive transcranial magnetic stimulation

INTRODUCTION

Parkinson’s disease (PD), one of the most frequent neurodegenerative disorders, is clinically characterized by akinesia, rigidity, resting tremor, and gait instability and is accompanied by various nonmotor complications, such as depression, cognitive decline, and sleeping problems, all of which could lead to significant family and social burdens^[1,2]. Although dopaminergic drugs could effectively control motor symptoms at the early stage of PD, nonmotor symptoms might be worsened, and additional drug-related motor complications could emerge with long-term dopaminergic treatment^[3,4]. Deep brain stimulation (DBS) surgery is a well-established and effective antiparkinsonian method for those patients who are refractory to medications, but it has a high cost and depends on strict patient selection criteria^[5].

As a noninvasive technique, repetitive transcranial magnetic stimulation (rTMS) has been recommended for the treatment of PD^[5–7]. While the commonly used figure “8” coil offers a focused stimulation approach, its focus could only reach 2~4 cm in depth^[8–10], i.e., only the superficial cortex could be stimulated directly. However, the core pathology of most brain disorders, including PD^[11], lies in the deep brain regions. Effectively modulating deep brain activity through the superficial cortex is essential for rTMS treatment. Wang et al. proposed a functional connectivity (FC)-guided rTMS approach with stimulation of the left parietal cortex which has FC with the hippocampus. The authors successfully modulated the FC of the hippocampus and significantly improved subjects’ associative memory performance^[12]. This FC-guided rTMS approach might also be a promising therapy for patients with brain disorders.

The sensorimotor subdivision of the subthalamic nucleus (STN), a crucial node of the cortico-basal ganglia motor pathway^[13,14], is an effective deep brain target in DBS surgery for PD patients^[15–17]; the effectiveness of targeting this region indicates that the dorsolateral (DL)-STN might be a viable deep brain region in the FC-guided rTMS approach.

However, the small size, biconvex shape, oblique orientation in 3D planes, and special junctional position among deep brain regions of the DL-STN^[18–20] make it a challenge to precisely target this region. To address the issue in DBS surgery, a method based on the anatomical landmarks of the red nucleus (RN) is widely used^[21,22]. This method greatly depends on well-trained neurosurgeons and is not suitable for FC-guided rTMS.

Therefore, the current study developed an “AutoSTN” method to help automatically target DL-STN coordinates. Furthermore, a one-stop plug-in FC analytic toolbox named the “Connectivity and Coordinates Converting Assistant Toolbox (CC-CAT)” was developed to automatically target the DL-STN in native space and to serve as an individualized potential superficial stimulation target in rTMS therapy.

METHODS

Participants and data acquisition

The research was conducted at the Center for Cognition and Brain Disorders (CCBD) of Hangzhou Normal University (HZNU) and subjected to a rigorous review by the local ethics committee (No. 20140508). Recruited participants excluded those with a history of neurological and psychiatric disorders. For each subject, a written informed consent was signed before participation in this study. Eighty subjects were included in this study. Of these, two were excluded due to poor MRI image quality; thus, 78 subjects (60.9 ± 7.7 years old, 54 males) were finally included.

For all participants, a resting-state functional magnetic resonance imaging (rs-fMRI), 3D-enhanced susceptibility-weighted angiography (ESWAN), and 3D T1-weighted imaging scanning were collected on a 3T GE scanner (MR750, GE Medical Systems, Milwaukee, WI).

During rs-fMRI collection, the subjects were instructed to keep their eyes closed, avoid falling asleep, and remain as still as possible. The rs-fMRI images were scanned with the following parameters: repetition time (TR) = 2,000 ms, echo time (TE) = 30 ms, interleaved acquisition with a total of 43 slices, flip angle (FA) = 90°, collecting matrix = 64 × 64, field of view (FOV) = 220 × 220 mm², spatial resolution = 3.44 × 3.44 × 3.20 mm³, and total scanning time = 8 min. ESWAN was performed as follows: TR = 54 ms, 16 echoes, first echo time (TE1) = 3 ms, echo separation = 3 ms, FA = 12°, collecting matrix = 256 × 256, FOV = 256 × 256 mm², and spatial resolution = 1 × 1 × 1 mm³. For the 3D T1 anatomical image, it was collected as follows: TR = 8.1 ms, TE = 3.1 ms, a total of 176 sagittal slices with thickness = 1 mm, FA = 8°, and FOV = 250 × 250 mm².

Data preprocessing

Individual DL-STN coordinate targeting based on anatomical landmarks

Data preprocessing for DL-STN coordinate targeting for each subject was performed based on AFNI (https://afni.nimh.nih.gov/) and SPM12 (http://www.fil.ion.ucl.ac.uk/spm/). First, we reoriented the T1 image manually to anterior commissure-posterior commissure (AC-PC) stereotactic space using 3drefit in AFNI. Then, ESWAN image was coregistered to the T1 image using SPM. After that, DL-STN coordinate targeting was conducted two times (1st and 2nd) by two independent raters (ZN and YJ) on the coregistered ESWAN images as follows.

DL-STN coordinate targeting was conducted on the coregistered ESWAN image in AC-PC aligned stereotactic space according to the steps of defining DL-STN in DBS surgery^[21,22] using AFNI. The details are shown in Figure 1.