A Protocol for Robot-assisted Neuronavigated Transcranial Magnetic Stimulation Targeting Individualized Brain Networks in Bipolar Disorder

Summary

This study proposes a method to ameliorate cognitive impairment during remission of bipolar disorder by robotic neuronavigated transcranial magnetic stimulation.

Abstract

Transcranial magnetic stimulation (TMS) is a non-invasive neuromodulation technique capable of modulating cortical plasticity and has been increasingly applied in the treatment of psychiatric disorders. However, conventional TMS often targets a single brain region and may suffer from imprecise coil positioning, potentially limiting its therapeutic efficacy. Robotic neuronavigated transcranial magnetic stimulation (RNNMS) is a precise TMS intervention modality, which is based on the structural magnetic resonance imaging (MRI) data of the human brain to formulate individualized stimulation targets and improve the accuracy of stimulation with the help of robotic arm guidance. Through brain imaging analysis, we selected stimulation target based on brain regions with abnormal functional connectivity in bipolar disorder (BD) patients, and customized intervention targets based on the structural magnetic resonance of each patient, and used robotic neuronavigated TMS to precisely intervene on the corresponding targets of the patients to reduce the possibility of ineffective stimulation due to coil shift during the whole intervention process, thus improving the cognitive impairments of the patients with BD. This protocol describes our approach to targeting patients based on their brain network images and demonstrates in detail the procedure for performing robotic neuronavigated TMS interventions on patients.

Introduction

Repetitive transcranial magnetic stimulation (rTMS) is currently the most widely used non-invasive neuromodulation technique to stimulate cortical areas of the brain, mainly through brief magnetic field pulses. Currently, rTMS has been approved by the Food and Drug Administration (FDA) for the treatment of major depressive disorder (MDD). rTMS has a response and remission rate of < 50% and < 20%, respectively, in MDD patients unresponsive to medication¹. Several studies on depression TMS in bipolar disorder (BD) have achieved positive efficacy, but the number of studies addressing cognitive impairment in BD is small and needs to be further explored².

Although the parameters of rTMS can be adjusted, such as the site, mode, frequency, intensity, and number of pulses of stimulation, the parameter patterns mostly used in previous studies as well as in current clinical applications are more fixed, such as mostly choosing 10 Hz stimulation of the left dorsolateral prefrontal cortex (DLPFC), theta burst stimulation (TBS), and 1 Hz stimulation of the right DLPFC³. The selection of stimulation targets in the cortex is an important factor in transcranial magnetic stimulation (TMS), and effective cortical targets may vary depending on the disease as well as different individuals. It has been shown that the efficacy of rTMS in refractory depression can be significantly improved by personalized stimulation using neuronavigation based DLPFC⁴, which also suggests the importance of individualized targets for TMS efficacy.

With the development of functional neuroimaging, changes in the activity of certain brain regions associated with, for example, hallucinations, anxiety, and depression have been found to have the potential to serve as therapeutic targets for these symptoms in studies and analyses of a number of psychiatric symptoms⁵. Moreover, the stimulatory effects of TMS are not limited to the stimulated area, but may also affect brain regions distant from the stimulation site by modulating functional connectivity between brain regions⁶. In contrast, functional connectivity within and between resting-state networks of the brain may also be altered in neuropsychiatric disease states⁷, and these altered brain networks may undergo partial normalization after the condition improves^8,9. The selection of the DLPFC as the target for conventional TMS targeting techniques, and the fact that different regions within the DLPFC map to different distributed brain networks¹⁰, is also likely to reveal the variability in treatment response between different patients in conventional TMS. Identifying rTMS targets based on connectivity between different brain regions may provide a more individualized approach¹¹. For example, one study localized the DLPFC coordinate that was most anti-correlated with the subgenual cingulate cortex (SGC) and used neuronavigation to optimize antidepressant response¹². On the one hand, current clinical interventions for BD patients still mostly follow the traditional targets of MDD, and on the other hand, existing protocols are more centered on mood symptoms in BD patients and less explored for the targets of cognitive impairment in BD, so further exploration of the corresponding targets is still needed.

In addition to the choice of target location, the ability of TMS to achieve a stimulus intensity consistent with expectations is also closely related to whether it is off-target or not; it has been found that elastic caps positioned to place coils reach cortical targets with significantly lower stimulation, with some individuals receiving only 48.6% of the expected target electric field, whereas neuronavigation targeted stimulation achieves a more accurate and precise localization of the TMS to produce the target cortical level with the expected stimulation intensity¹³. During treatment, sometimes the patient's head position may move¹⁴, which may also result in a discrepancy between the actual stimulation target and the expected target position, thus failing to achieve the expected stimulation intensity¹³. The use of robotic neuronavigation can be calibrated during TMS interventions and keep the stimulation position as accurate as possible, but the efficacy of interventions for bipolar depression in this modality needs to be explored more.

In this article, we describe a safe, promising, and more precise transcranial magnetic therapy modality based on functional connectivity networks to identify targets for intervention using robotic neuronavigation with transcranial magnetic stimulation to ameliorate cognitive impairment in patients with bipolar disorder.

Protocol

This study was reviewed and approved by the Ethics Committee of the First Affiliated Hospital, Zhejiang University School of Medicine, in accordance with the Declaration of Helsinki, and this study has been registered with the Chinese Clinical Trial Registry (http://www.chictr.org.cn/enIndex.aspx), with the registration number ChiCTR2000030675. Here, the patients were recruited from the psychiatric outpatient clinic of the First Affiliated Hospital, Zhejiang University School of Medicine. All participants were thoroughly informed about the purpose, procedures, potential risks, and benefits of the study, and written informed consent was obtained prior to participation.

1 Participants, eligibility, and randomization

1. Inclusion criteria: Include patients who fulfill the following criteria:
   1. Patients who are 16-65 years of age.
   2. Patients who are diagnosed with BD by a psychiatrist according to the Mini-International Neuropsychiatry Interview (MINI).
   3. Patients who are on stable medication.
   4. Patients who have a Young Mania Rating Scale (YMRS) score ≤ 6¹⁵ and 17-item Hamilton Depression Rating Scale (HDRS-17) score ≤ 7.
   5. Patients who are in clinical remission formore than three months¹⁶.
   6. Patients who have self-reported cognitive impairment and perceptual deficits with a Self-reported Cognitive Disorders and Perceptual Deficits Questionnaire-Depression (PDQ-D) score ≥ 17^17,18.
   7. Patients who are right-handed.
   8. Patients who have at least 9 years of education.
2. Exclusion criteria: Exclude patients who have the following conditions:
   1. Patients who have comorbidity with any other mental disorder as defined in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5).
   2. Patients with a history of severe neurological disorders, such as epilepsy or traumatic brain injury.
   3. Patients who have significant and unstable medical conditions.
   4. Patients who have a history of drug or alcohol abuse.
   5. Pregnant or breastfeeding women.
   6. Patients who have generalized colorblindness, hypochromia, or hearing impairment.
   7. Patients who are currently taking medication with antidepressants, anticholinergics, and other drugs.
   8. Patients with other medical conditions.
   9. Patients who are unable to undergo magnetic resonance imaging (MRI) or have contraindications for rTMS.
3. Randomization and allocation concealment
   1. Generate a random number sequence using a computer program to assign patients to active or sham rTMS groups.
   2. Prepare the randomization sequence and keep it with an independent researcher who is not involved in recruitment, assessment, or treatment.
   3. Conceal group allocation using sealed, opaque envelopes, and open the envelope only at the time of intervention.
   4. Maintain blinding throughout the study so that both patients and treating/assessing researchers remain unaware of treatment assignment.
   5. Instruct patients not to discuss their treatment with other participants.

2 Acquisition of magnetic resonance images

NOTE: After enrollment, acquire structural and functional MRI to provide the imaging data required for individualized target selection in the subsequent steps. Imaging was performed at the First Affiliated Hospital of Zhejiang University School of Medicine using a 3.0 Telsa scanner with a standard full head coil.

1. Ask participants to lie supine with eyes closed.Position foam pads to minimize motion and provide earplugs to reduce noise.
2. Acquire high-resolution 3D anatomical images using a magnetization-prepared rapid acquisition gradient echo (MPRAGE) sequence with T1-weighted contrast.
3. Set the acquisition parameters as follows: TR = 7.1 ms; TE = 2.9 ms; FOV = 260 x 260 mm²; matrix = 256 × 256; number of slices = 146; slice thickness = 1 mm; flip angle = 8°
4. Obtain resting-state echo planar imaging (EPI) with: TR 1800 ms; TE 30 ms; FOV 240 x 240 mm²; gap 0.8 mm; matrix 64 × 64; 28 slices; 4 mm thickness; specify run length (time points).

3 Individualized target selection based on MRI

NOTE: Using the acquired MRI data, preprocess fMRI and derive seed-based functional connectivity to determine the stimulation target coordinate, then generate the individualized structural model for neuronavigation.

1. fMRI data preprocessing:
   1. Preprocess fMRI data using the DPABI platform (Version8.2).
   2. Perform DICOM to NIfTI conversion.
   3. Discard the initial 10 time points and realign head motion to exclude participants with displacement > 1.5 mm or rotation > 1.5° in any direction (x, y, or z).
   4. Reorient the T1-weighted and functional images to the anterior commissure. Coregister the T1 image to the functional images.
   5. Segment the T1 image and normalize the segmented images to the Montreal Neurological Institute (MNI) standard space using DARTEL.
   6. Apply the transformation parameters to the functional images and regress 24 head motion covariates (Friston-24).
   7. Resample the functional images to a voxel size of 3 mm x 3 mm x 3 mm.
   8. Smooth the resampled images spatially using a Gaussian kernel with a full width at half maximum (FWHM) of 4 mm, which was chosen to preserve spatial specificity of relatively small ROIs such as the ACC and V1 while still improving the signal-to-noise ratio.
   9. Remove cerebrospinal fluid and white matter signals by regression analysis. Band-pass filter the time series between 0.01 and 0.1 Hz.
   10. Remove time points with excessive motion according to the predefined thresholds.
2. Determine the coordinates of the target point according to the functional connection:
   1. Define the anterior cingulate cortex (ACC) region of interest (ROI) as a 9 mm radius sphere centered at MNI (−4, 50, 4), based on previously reported coordinates from an earlier study of BD¹⁹.
   2. Extract the time series from the ACC ROI and construct seedbased functional connectivity (FC) maps (voxelwise) using the ACC time course.
   3. Use the computed seedbased FC to identify optimized TMS targeting coordinates within the primary visual cortex (V1). In parallel, define the same ACC ROI a priori in a reference cohort of 30 healthy subjects.
   4. Identify V1 voxels showing significant FC with the ACC (voxelwise p < 0.001; clustered FWE-corrected p < 0.05) based on the ACC ROI defined above.
   5. Select the peak V1 coordinate at MNI (−3, −66, 18), which shows significant connectivity with the ACC (r = 0.269), as the rTMS target for the activestimulation group (Figure 1).
3. Structural MRI preprocessing and individual model generation before treatment:
   1. Anonymize the patient's structural image data to remove all personal identifiers and enable secure data transfer.
   2. Standardize the raw images by adjusting resolution, voxel spacing, orientation, and signal intensity.
   3. Segment the standardized images into distinct tissue classes, including gray matter, white matter, cerebrospinal fluid, skull, and internal nuclei.
   4. Register the segmented images to the standard image coordinate system (MNI152) according to the stimulation protocol requirements, and extract the corresponding brain regions, standardized coordinates, and other relevant parameters.
   5. Reconstruct 3D models of the brain regions and skull separately, and determine the stimulation target coordinates within the model space.

4 Measurement of resting motor threshold (RMT)

1. With the individualized target determined and the head model prepared, determine each participant's resting motor threshold over M1 to scale stimulation intensity for the upcoming intervention.
   1. Seat the patient comfortably in the treatment chair.
   2. Fit the patient with a sensor-equipped headband. Ensure that it is securely fastened and that no object obstructs the line of sight between the sensors and the tracking cameras.
   3. Enter the patient's identification number into the neuronavigation system, verify the information, and access the corresponding interface.
   4. Display the patient's head model on the screen, showing markers at the forehead center, bilateral ear tips, nasal tip, and vertex. Adjust the marker positions using the mouse to align them accurately.
   5. Use the neuronavigation pointer to mark the anatomical landmarks on the patient's head (glabella, bilateral ear tips, nasal tip, and vertex). Press the foot pedal to confirm each location. Verify that each successfully marked point appears as a green dot on the screen.
   6. Move the neuronavigation pointer across the patient's scalp. Observe the movement trajectory as green dots projected on the screen to confirm the alignment between the patient's head position and the uploaded MRI model. Complete the spatial registration.
   7. Use the robot-assisted neuronavigation system with a figure-of-eight coil to position the coil over the left M1 target location recommended by the system.
   8. Deliver single TMS pulses to the left M1. Record motor evoked potentials (MEPs) from the fully relaxed right first dorsal interosseous (FDI) muscle using surface electromyography.
   9. Determine the RMT as the lowest stimulation intensity that produces MEPs of at least 50 µV in amplitude in at least 5 out of 10 consecutive trials²⁰.

5 Robotic neuro-navigated rTMS intervention

1. With stimulation intensity set by the measured RMT and the V1 target registered in the neuronavigation system, deliver the robot-assisted rTMS according to the scheduled treatment sessions.
   1. Administer robot-assisted neuro-navigated rTMS once daily for 10 consecutive days to all participants.
   2. Deliver each session as 60 training blocks of 5s 10 Hz stimulation at 110% of RMT, with 20 s inter-training intervals, for a total of 3000 pulses per session and a total duration of 25 min. For sham stimulation, apply the same protocol but use a sham TMS coil, producing a similar sound, with no magnetic field applied.
   3. Enter the MNI coordinates corresponding to the stimulation target point in the patient's V1 region.
   4. Adjust the angle and height of the seat to ensure that the position of the patient's head is within the green box on the screen, and allow the robot to bring the coil to the patient's corresponding target position according to the neuronavigation system. Use robot navigation to adjust the target point of the actual stimulus (shown as a red ball) to coincide with the location on the patient's head model on the screen (shown as a green ball). Display the distance between the two points in the lower-left corner of the screen. When the distance is less than 1 mm, begin rTMS stimulation.
   5. Display the distance between the coil and the target point on the screen throughout the treatment, and calibrate when the patient's head shifts to ensure that the scalp remains close to the coil (Figure 2).
   6. After each daily treatment session, instruct the patient to complete a self-report adverse effects questionnaire, which includes items on dizziness, headaches, and other possible symptoms.

6 Clinical data collection

1. Assess participants' cognitive function at baseline and 2 weeks post-treatment using the THINC Integration Instrument (THINC-it).
2. Administer the Depression Perceived Deficits Inventory, Identification Task (IDN), One-Back Task (1-BACK), Digit Symbol Substitution Test (DSST), and Trail Making Test-Part B (TMT-B) to evaluate self-reported cognitive deficits, attention, processing speed, working memory, and executive function, respectively.

Results

In this study, the THINC-it tool was used to assess cognitive changes in patients before and after treatment. Table 1 shows the baseline situation of the two groups of patients, in which there was no difference between the two groups at baseline in any of the six indicators related to cognitive function. Whereas a significant interaction was found in the indicator Symbol Check (Accuracy), no significant interaction was found for several other indicators (Table 2). Although there was no s...

Discussion

This protocol proposes to identify targets based on V1-ACC functional connectivity and to improve cognitive impairment in bipolar disorder using rTMS at 10 Hz under robotic neural navigation. Assessment of patients' cognitive function before and after the intervention by the THINC-it tool revealed that the trend of cognitive improvement was more pronounced in patients who underwent active TMS stimulation.

The key steps in the protocol mainly include target customization and rTMS interventi...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
Black Dolphin Navigation TMS Robot	Xi’an Solide Brain Modulation	SmarPhin S-50	Integrated system with active/sham coils, TMS robot, optical tracking, tracking headband, neuronavigation, and earplugs; enables MRI-guided targeting and real-time coil–target distance monitoring.
GE Signa Premier 3.0T MRI Scanner	GE Healthcare	https://www.gehealthcare.com/products/magnetic-resonance-imaging/3t-mri-scanners
THINC-it Cognitive Assessment Tool	Lundbeck	1.26 1	Includes four cognitive tasks and one self-report scale, assessing attention/concentration/alertness, working memory, and executive function.