Comparison of the Stereotactic Accuracies of Function-Guided Deep Brain Stimulation, Calculated Using Multitrack Target Locations Geometrically Inferred from Three-Dimensional Trajectory Rotations, and of Magnetic Resonance Imaging-Guided Deep Brain Stimulation and Outcomes

STN-DBS electrode placement accuracy and motor improvement in Parkinson's disease: systematic review and individual patient meta-analysis

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is an effective neurosurgical treatment for Parkinson's disease. Surgical accuracy is a critical determinant to achieve an adequate DBS effect on motor performance. A two-millimetre surgical accuracy is commonly accepted, but scientific evidence is lacking. A systematic review and meta-analysis of study-level and individual patient data (IPD) was performed by a comprehensive search in MEDLINE, EMBASE and Cochrane Library. Primary outcome measures were (1) radial error between the implanted electrode and target; (2) DBS motor improvement on the Unified Parkinson's Disease Rating Scale part III (motor examination). On a study level, meta-regression analysis was performed. Also, publication bias was assessed. For IPD meta-analysis, a linear mixed effects model was used. Forty studies (1391 patients) were included, reporting radial errors of 0.45-1.86 mm. Errors within this range did not significantly influence the DBS effect on motor improvement. Additional IPD analysis (206 patients) revealed that a mean radial error of 1.13±0.75 mm did not significantly change the extent of DBS motor improvement. Our meta-analysis showed a huge publication bias on accuracy data in DBS. Therefore, the current literature does not provide an unequivocal upper threshold for acceptable accuracy of STN-DBS surgery. Based on the current literature, DBS-electrodes placed within a 2 mm range of the intended target do not have to be repositioned to enhance motor improvement after STN-DBS for Parkinson's disease. However, an indisputable upper cut-off value for surgical accuracy remains to be established. PROSPERO registration number is CRD42018089539.

Robot-Assisted Stereotaxy Reduces Target Error: A Meta-Analysis and Meta-Regression of 6056 Trajectories

BACKGROUND

The pursuit of improved accuracy for localization and electrode implantation in deep brain stimulation (DBS) and stereoelectroencephalography (sEEG) has fostered an abundance of disparate surgical/stereotactic practices. Specific practices/technologies directly modify implantation accuracy; however, no study has described their respective influence in multivariable context.

OBJECTIVE

To synthesize the known literature to statistically quantify factors affecting implantation accuracy.

METHODS

A systematic review and meta-analysis was conducted to determine the inverse-variance weighted pooled mean target error (MTE) of implanted electrodes among patients undergoing DBS or sEEG. MTE was defined as Euclidean distance between planned and final electrode tip. Meta-regression identified moderators of MTE in a multivariable-adjusted model.

RESULTS

A total of 37 eligible studies were identified from a search return of 2,901 potential articles (2002-2018) - 27 DBS and 10 sEEG. Random-effects pooled MTE = 1.91 mm (95% CI: 1.7-2.1) for DBS and 2.34 mm (95% CI: 2.1-2.6) for sEEG. Meta-regression identified study year, robot use, frame/frameless technique, and intraoperative electrophysiologic testing (iEPT) as significant multivariable-adjusted moderators of MTE (P < .0001, R2 = 0.63). Study year was associated with a 0.92-mm MTE reduction over the 16-yr study period (P = .0035), and robot use with a 0.79-mm decrease (P = .0019). Frameless technique was associated with a mean 0.50-mm (95% CI: 0.17-0.84) increase, and iEPT use with a 0.45-mm (95% CI: 0.10-0.80) increase in MTE. Registration method, imaging type, intraoperative imaging, target, and demographics were not significantly associated with MTE on multivariable analysis.

CONCLUSION

Robot assistance for stereotactic electrode implantation is independently associated with improved accuracy and reduced target error. This remains true regardless of other procedural factors, including frame-based vs frameless technique.

The Accuracy of Direct Targeting Using Fusion of MR and CT Imaging for Deep Brain Stimulation of the Subthalamic Nucleus in Patients with Parkinson's Disease

INTRODUCTION

 In 33 consecutive patients with Parkinson's disease (PD) undergoing awake deep brain stimulation (DBS) without microelectrode recording (MER), we assessed and validated the precision and accuracy of direct targeting of the subthalamic nucleus (STN) using preoperative magnetic resonance imaging (MRI) and stereotactic computed tomography (CT) image fusion combined with immediate postoperative stereotactic CT and postoperative MRI, and we report on the side effects and clinical results up to 6 months' follow-up.

MATERIALS AND METHODS

 Preoperative nonstereotactic MRI and stereotactic CT images were merged and used for planning the trajectory and final lead position. Immediate postoperative stereotactic CT and postoperative nonstereotactic MRI provided the validation of the final electrode position. Changes in the Unified Parkinson's Disease Rating Scale III (UPDRS III) scores and the levodopa equivalent daily doses (LEDD) and appearance of adverse side effects were assessed.

RESULTS

 The mean Euclidian distance (ED) error between the planned position and the final position of the lead in the left STN was 1.69 ± 0.82 mm and that in the right STN was 2.12 ± 1.00. The individual differences between planned and final position in each of the three coordinates were less than 2 mm. The UPDRS III scores improved by 75% and LEDD decreased by 45%. Few patients experienced complications, such as postoperative infection (n = 1), or unwanted side effects, such as emotional instability (n = 1).

CONCLUSION

 Our results confirm that direct targeting of an STN on stereotactic CT merged with MRI could be a valid method for placement the DBS electrode. The magnitude of our targeting error is comparable with the reported errors when using MER and other direct targeting approaches.

Pretreatment brain volumes can affect the effectiveness of deep brain stimulation in Parkinson's disease patients

We aimed to assess whether brain volumes may affect the results of deep brain stimulation (DBS) in patients with Parkinson’s disease (PD). Eighty-one consecutive patients with PD (male:female 40:41), treated with DBS between June 2012 and December 2017, were enrolled. Total and regional brain volumes were measured using automated brain volumetry (NeuroQuant). The Unified Parkinson Disease Rating Scale motor score quotient was used to assess changes in clinical outcome and compare the preoperative regional brain volume in patients categorized into the higher motor improvement and lower motor improvement groups based on changes in the postoperative scores. The study groups showed significant volume differences in multiple brain areas. In the higher motor improvement group, the anterior cingulate and right thalamus showed high volumes after false discovery rate (FDR) correction. In the lower motor improvement group, the left caudate, paracentral, right primary sensory and left primary motor cortex showed high volume, but no area showed high volumes after FDR correction. Our data suggest that the effectiveness of DBS in patients with PD may be affected by decreased brain volume in different areas, including the cingulate gyrus and thalamus. Preoperative volumetry could help predict outcomes in patients with PD undergoing DBS.

Deep Learning-Based Deep Brain Stimulation Targeting and Clinical Applications

Background

The purpose of the present study was to evaluate deep learning-based image-guided surgical planning for deep brain stimulation (DBS). We developed deep learning semantic segmentation-based DBS targeting and prospectively applied the method clinically.

Methods

T2^∗ fast gradient-echo images from 102 patients were used for training and validation. Manually drawn ground truth information was prepared for the subthalamic and red nuclei with an axial cut ∼4 mm below the anterior–posterior commissure line. A fully convolutional neural network (FCN-VGG-16) was used to ensure margin identification by semantic segmentation. Image contrast augmentation was performed nine times. Up to 102 original images and 918 augmented images were used for training and validation. The accuracy of semantic segmentation was measured in terms of mean accuracy and mean intersection over the union. Targets were calculated based on their relative distance from these segmented anatomical structures considering the Bejjani target.

Results

Mean accuracies and mean intersection over the union values were high: 0.904 and 0.813, respectively, for the 62 training images, and 0.911 and 0.821, respectively, for the 558 augmented training images when 360 augmented validation images were used. The Dice coefficient converted from the intersection over the union was 0.902 when 720 training and 198 validation images were used. Semantic segmentation was adaptive to high anatomical variations in size, shape, and asymmetry. For clinical application, two patients were assessed: one with essential tremor and another with bradykinesia and gait disturbance due to Parkinson’s disease. Both improved without complications after surgery, and microelectrode recordings showed subthalamic nuclei signals in the latter patient.

Conclusion

The accuracy of deep learning-based semantic segmentation may surpass that of previous methods. DBS targeting and its clinical application were made possible using accurate deep learning-based semantic segmentation, which is adaptive to anatomical variations.

Morphometric Study of the Internal Globus Pallidus Using the Robert, Barnard, and Brown Staining Method

BACKGROUND

The globus pallidus internus (Gpi) is a major target in functional neurosurgery. Anatomical studies are crucial for correct planning and good surgical outcomes in this region. The present study described the anatomical coordinates of the Gpi and its relationship with other brain structures and compared the findings with those from previous anatomical studies.

METHODS

We obtained 35 coronal and 5 horizontal brain specimens from the Department of Anatomy and stained them using the Robert, Barnard, and Brown technique. After excluding defective samples, 60 nuclei were analyzed by assessing their distances to the anatomical references and the trajectories to these nuclei.

RESULTS

The barycenter of the Gpi was identified at the level of the mammillary bodies and 1 cm above the intercommissural plane. Thereafter, the distances to other structures were found. The mean ± standard deviation distance was 15.62 ± 2.66 mm to the wall of the third ventricle and 17.02 ± 2.69 mm to its midline, 4.74 ± 1.12 mm to the optic tract, 2.51 ± 0.8 mm and 13.56 ± 2 mm to the internal and external capsule, and 21.3 ± 2.44 mm to the insular cortex. The cortical point of entry should be located 22.03 ± 4.34 mm to 48.74 ± 4.44 mm from the midline.

CONCLUSION

The Gpi has less variability in distance to closer anatomical references, such as the optic tract and internal capsule. Distant locations showed a more inhomogeneous pattern. Anatomical studies such as ours are important for the development of new therapeutic approaches and can be used as a basis for new research involving volumetric and specific group analyses.

MR Safe Robot Assisted Needle Access of the Brain: Preclinical Study

We report the results of preclinical experiments for direct MRI-guided needle interventions in the brain. An MR Safe robot was incorporated into an intraoperative MRI system. Deep regions of the brain simulated in a cranial mockup were targeted with a needle under robotic assistance. The 3D accuracy of in-scanner targeting at an average depth of 95[Formula: see text]mm was 1.55[Formula: see text]mm, with no manual corrections.