Comparison of magnetic resonance imaging sequences for depicting the subthalamic nucleus for deep brain stimulation

A literature review of magnetic resonance imaging sequence advancements in visualizing functional neurosurgery targets

OBJECTIVE

Historically, preoperative planning for functional neurosurgery has depended on the indirect localization of target brain structures using visible anatomical landmarks. However, recent technological advances in neuroimaging have permitted marked improvements in MRI-based direct target visualization, allowing for refinement of "first-pass" targeting. The authors reviewed studies relating to direct MRI visualization of the most common functional neurosurgery targets (subthalamic nucleus, globus pallidus, and thalamus) and summarize sequence specifications for the various approaches described in this literature.

METHODS

The peer-reviewed literature on MRI visualization of the subthalamic nucleus, globus pallidus, and thalamus was obtained by searching MEDLINE. Publications examining direct MRI visualization of these deep brain stimulation targets were included for review.

RESULTS

A variety of specialized sequences and postprocessing methods for enhanced MRI visualization are in current use. These include susceptibility-based techniques such as quantitative susceptibility mapping, which exploit the amount of tissue iron in target structures, and white matter attenuated inversion recovery, which suppresses the signal from white matter to improve the distinction between gray matter nuclei. However, evidence confirming the superiority of these sequences over indirect targeting with respect to clinical outcome is sparse. Future targeting may utilize information about functional and structural networks, necessitating the use of resting-state functional MRI and diffusion-weighted imaging.

CONCLUSIONS

Specialized MRI sequences have enabled considerable improvement in the visualization of common deep brain stimulation targets. With further validation of their ability to improve clinical outcomes and advances in imaging techniques, direct visualization of targets may play an increasingly important role in preoperative planning.

Defining the Border of the Subthalamic Nucleus for Deep Brain Stimulation: A Proposed Model Using the Symmetrical Sigmoid Curve Function

BACKGROUND

The subthalamic nucleus (STN) is an important target during deep brain stimulation (DBS). Accurate lead placement is integral to achieving satisfactory clinical outcomes; however, the STN remains a structure whose visualization is highly variable with borders often difficult to define. We aimed to develop an objective method of evaluating the visibility of the STN on preoperative magnetic resonance imaging (MRI) to standardize future comparative assessments between imaging protocols and patient-specific parameters.

METHODS

An imaging study of 64 prospectively collected patients undergoing bilateral DBS of the STN for various movement disorders was performed with institutional approval. MRI scans were acquired using a uniform protocol involving general anesthesia, cranial fixation in a Leksell stereotactic frame, and long acquisition times using a 3T MRI scanner. The images were analyzed using the iPlan Stereotaxy, version 2.6, workstation. High-resolution T2-weighted axial sections were evaluated, and the voxel values in the region of the presumed posterior border of the STN (as defined by the operating neurosurgeon) were obtained. A 4-parameter logistic symmetrical sigmoid curve was used to map the voxel values as they progressed from within to outside the region of the STN border. The inflection point and Hill coefficient of this symmetrical curve was calculated to provide objective information on the location and clarity of the STN border, respectively. These findings were compared with the surgeon's judgment of the STN border. To demonstrate the use of the sigmoid curve, the patients' head volumes were also calculated and evaluated to assess whether larger head volumes adversely affected STN visibility.

RESULTS

The symmetrical sigmoid curve model provided objective information on the visibility of the STN on T2-weighted MRI scans and could be generated in 86% of the patients. The other 14% of patients had MRI scans that generated linear graphs, indicating the poorest scoring for STN image quality. No correlation between head volume and STN visibility was identified.

CONCLUSIONS

Our proposed statistical model allows for standardized examination of the visibility of the STN border for DBS and has potential for both clinical and academic applications.

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.

Subthalamic Nucleus Visualization on Routine Clinical Preoperative MRI Scans: A Retrospective Study of Clinical and Image Characteristics Predicting Its Visualization

BACKGROUND

The visualization of the subthalamic nucleus (STN) on magnetic resonance imaging (MRI) is variable. Studies of the contribution of patient-related factors and intrinsic brain volumetrics to STN visualization have not been reported previously.

OBJECTIVE

To assess the visualization of the STN during deep brain stimulation (DBS) surgery in a clinical setting.

METHODS

Eighty-two patients undergoing pre-operative MRI to plan for STN DBS for Parkinson disease were retrospectively studied. The visualization of the STN and its borders was assessed and scored by 3 independent observers using a 4-point ordinal scale (from 0 = not seen to 3 = excellent visualization). This measure was then correlated with the patients' clinical information and brain volumes.

RESULTS

The mean STN visualization scores were 1.68 and 1.63 for the right and left STN, respectively, with a good interobserver reliability (intraclass correlation coefficient: 0.744). Older age and decreased white matter volume were negatively correlated with STN visualization (p < 0.05).

CONCLUSION

STN visualization is only fair to good on routine MRI with good concordance of interindividual rating. Advancing age and decreased white matter are associated with poor visualization of the STN. Knowledge about factors contributing to poor visualization of the STN could alert a surgeon to modify the imaging strategy to optimize surgical targeting.

T2 Star-weighted Angiography (SWAN) Allows to Concomitantly Assess the Prostate Contour While Detecting Fiducials Before MR-based Intensity-modulated Radiation Therapy in Prostate Carcinoma

PURPOSE

To evaluate the performance of T2 star-weighted angiography (SWAN) to concomitantly assess the prostate contour while detecting fiducials before magnetic resonance (MR)-based intensity-modulated radiation therapy (IMRT) in prostate carcinoma.

MATERIALS AND METHODS

Forty patients (mean age: 73.1 ± 7.5 years; average Gleason score: 7 ± 1; average prostate-specific antigen: 14.7 ± 11.6 ng/mL) underwent MR and computed tomography imaging before fiducial-based IMRT. MR protocol included SWAN, T2-weighted (T2w) and diffusion-weighted imaging in a first group (n = 20) and SWAN, T2w and T2-star weighted imaging in a second group (n = 20). In group 1, the depiction of fiducials, image sharpness and visibility of prostate boundaries were independently evaluated by 2 readers on SWAN, T2w or diffusion-weighted images. In group 2, a similar evaluation was performed by 2 other readers on SWAN and T2-star images only. Depiction of fiducials was compared to computed tomography findings.

RESULTS

The median scores of visibility of prostate boundaries, image sharpness and depiction of fiducials by SWAN were above average to excellent for all readers. In group 1, readers correctly located 56 of 57 (98.2%) and 47 of 57 (82.5%) fiducials, respectively; and 50 of 51 (98%), and 48 of 51 (88.2%) fiducials in group 2, respectively.

CONCLUSION

By allowing adequate visualization of the prostate boundaries and high depiction of fiducial markers concomitantly, SWAN might be used for treatment planning of IMRT. The use of this sequence might simplify the registration process and limit any errors associated with image fusion.

Post-operative imaging in deep brain stimulation: A controversial issue

In deep brain stimulation (DBS), post-operative imaging has been used on the one hand to assess complications, such as haemorrhage; and on the other hand, to detect misplaced contacts. The post-operative determination of the accurate location of the final electrode plays a critical role in evaluating the precise area of effective stimulation and for predicting the potential clinical outcome; however, safety remains a priority in postoperative DBS imaging. A plethora of diverse post-operative imaging methods have been applied at different centres. There is neither a consensus on the most efficient post-operative imaging methodology, nor is there any standardisation for the automatic or manual analysis of the images within the different imaging modalities. In this article, we give an overview of currently applied post-operative imaging modalities and discuss the current challenges in post-operative imaging in DBS.

Intra-operative micro-electrode recording in functional neurosurgery: Past, present, future

The field of functional neurosurgery has experienced a rise, fall and lastly a renaissance over the past 75years. Micro-electrode recording (MER) played a key role during this eventful journey. However, as the intra-operative MRI continues to evolve, a pertinent question about the utility of MER has been raised in recent years. In this article, we critically review these current controversies. The English literature is reviewed and the complex technique of MER is discussed in a simplified manner. The improvement of neuroimaging and its application in functional neurosurgery, especially in deep brain stimulation, is discussed. Finally, the current controversies and technical advances which can direct the future are reviewed. The results of existing meta-analyses addressing the controversies are summarized. Wide variations of pre-operative and intra-operative targeting methods have been described in the literature. Though functional neurosurgery is generally safe, complications do occur and multiple passes during MER can certainly add to the risk of inadvertent hemorrhage and infection. Additionally, the recent introduction of newer MRI modalities has ensured better delineation of the target. However, MER is still useful to address brain shift, for mapping of newer targets, for ablative surgeries and in centers without an intra-operative imaging facility. In the current scenario, it is nearly impossible to conduct a prospective study to decide the utility of MER. The importance of MER may further diminish in the future as a routine procedure, but its role as a gold standard procedure may still persist.