The main goal of the first programming visit is to determine the TW for each of the electrode contacts . Directional leads are often first tested in ring mode, followed by directional stimulation on individual segments if efficacy in ring mode is limited by stimulation-related side effects, or to identify programming strategies that allow a larger TW with lower energy consumption . Supporting programming software allows for visualizations of the anatomical context along with the predicted volume of tissue activated (VTA) and for functional mapping of the thresholds for clinical efficacy and side effects. Some of these tools also allow for an integrated analysis of power consumption adjusted to each contact impedance.
Visualization of Volume of Tissue Activated
The volume of neural activation can be estimated using finite element techniques, wherein a computer system subdivides a volumetric region of interest and solves for a scenario by applying appropriate mathematical relationships, initial conditions, and boundary conditions and iterating until a stable solution results. In this domain, this is applied by positioning an appropriate model of a stimulating electrode, configuring it with appropriate stimulation parameters (polarities, stimulus amplitude), and determining the resulting current flows and voltage gradients in the surrounding tissue. These electromagnetic gradients can then be applied, in a second step, to a population of simulated neurons positioned and oriented preferentially around the stimulating electrode so as to determine points at which neurons may be activated by a given stimulation regime . These neurons may be modeled such that they respond both to the intensity of the stimulation (current amplitude) and the time-dependent parameters such as the width of a stimulation pulse . As an example, various neuron fiber diameters may be included, such that the response of smaller fibers to a given stimulus pulse is different than that of larger fibers. The resulting neuron population that is predicted to be activated can then be represented visually, as a volume enclosing the points of activation.
Model-based approaches have been shown to have some predictive value when used in visual programming systems . However, validating these models can be challenging, as it is difficult to know with certainty which neural tissue volumes are indeed activated by stimulation. One thing that has been established is that fully featured models incorporating voltage drop and capacitance of the electrode–electrolyte interface, tissue encapsulation of the electrode, and diffusion tensor-based 3D tissue anisotropy and inhomogeneity produce more realistic predictions than simpler models that do not account for these factors .
As new stimulation modalities are introduced, including the new options afforded by directional leads, these VTA models may be used to compare the different stimulation options available. As examples, in bipolar stimulation, they can be used to understand the differences in volume between anodal and cathodal activation regions. For interleaved stimulation, they can be used to understand areas of overlap in separate pulse trains, wherein tissue receives stimulation at double the underlying frequency . In directional stimulation, the ability to push stimulation off axis, commonly referred to as displacement of the VTA centroid, can be explored in comparison to traditional ring mode stimulation. This may be of specific interest, as the thresholds to activation in directional stimulation may be different than in traditional rings. For fixed stimulation amplitudes, the smaller surface area of a directional segment creates relatively higher gradients around the electrode which may increase the probability of neural activation .
Visualization Strategies: Opportunities
A preoperative magnetic resonance imaging (MRI) may be used to establish the location and boundaries of anatomical structures, either through adaptation of an anatomical atlas or direct segmentation of structure boundaries from intensity information in the images themselves. Further, fiber tracts may additionally be represented if appropriate imaging sequences (diffusion tensor imaging) or atlases are available.
Final Lead Location and Orientation
A postoperative, or in some cases an intraoperative, image may be used to establish the final lead location and orientation. For this purpose, images that clearly delineate the electrode positions and the orientation of fiducial markers included in the lead body via a hyper- or hypo-intense image artifact can be used.
Visualization of Predicted Stimulation Extent
A VTA or similar model may be applied to the representation of the stimulating electrodes. Models may be created for one or more potential sets of stimulating parameters (i.e., different active contacts, different current or voltage amplitudes, monopolar vs. bipolar stimulation, directional vs. conventional ring stimulation, etc.) such that the overlap of the resulting stimulation field to patient anatomy can be assessed.
Integration of Ancillary Information
In addition to anatomy, lead location, and stimulation extent, visualization solutions may additionally allow for the addition of other information of potential use in identifying optimal stimulation settings. For example, intra- or postoperative electrophysiological measures may be visualized in the patient-specific anatomical context. In addition, aggregated historical information about stimulation outcome in the form of a statistical outcome or side effect map may be visualized in the patient-specific anatomical context to further inform programming decisions.
Visualization Strategies: Challenges
Limitations of Anatomical Models
Currently used anatomical models on which the lead is visualized are usually not patient-specific, and even when patient-specific, they are based on presurgical MRI images that do not account for procedural brain shift and postsurgical anatomical changes reported at up to 4 mm in the deep brain [42,43,44,45,46].
Deviation of DBS Lead
DBS leads show large deviations from their intended implanted orientation: more than 30° rotation in 42% of the leads and more than 60° rotation in 11% of the leads . Thus, the orientation of the individual segmented contacts might be no longer valid relative to the underlying anatomy in presurgical MRI presented during programming. Furthermore, significant lead migration (greater than 3 mm) along the ventrodorsal axis or upward displacement from immediate to delayed CT has been reported in over 12% of leads placed at an expert center [48, 49].
Variability in Tissue Impedance and Anisotropy
Validating VTA models is challenging as it is difficult to know with certainty which neural tissue volumes are stimulated . VTA models made available for programming are based on generic homogeneous models that do not account for tissue inhomogeneities, for instance permittivity and conductivity of brain structures, which may alter VTA predictions from − 44% to 174% . Also, VTA models available in programming platforms do not account for patient-specific tissue anisotropy that can at best only be modeled using tractography from high-resolution patient-specific diffusion tensor imaging (DTI) [52, 53]. Other real-world factors, not usually modeled, that influence the accuracy, shape, and extent of the VTA include physiology and pathophysiology , brain pulsation and patient hydration , glial scar formation around the electrode , and local fluid retention .
Software Platforms to Assist DBS Programming
SureTune™ (Supplementary Fig. S1) is a programming visualization tool available for clinical use in many regions and close to being released also in the USA. It incorporates the Bardinet–Yelnick anatomic atlas and registration algorithm , and further allows a user to manually adjust structure size, shape, and location to allow for valid representation of a patient’s specific anatomical variation. It also supports intensity-based segmentation of MRI visible structures. Within the existing software, leads can be placed according to stereotactic coordinates or aligned via postoperative images. Anatomic representations of structures can be validated by co-registration of microelectrode recordings to confirm anatomical boundaries if additional confidence is desired. Stimulation modeling is available for common stimulation configurations, and stimulation plans can be created for use in the clinical setting. Finally, via a service offering, statistical maps of outcome or side effects can be created which can then be prospectively visualized to inform future clinical decision-making or research applications: a recent study  showed that such statistical maps correlated well with best clinical programming, demonstrating predictive utility in prospective cohorts for GPi stimulation in dystonia.
While potentially useful for optimizing clinical care, SureTune™ has several limitations. Its atlas is a single brain histologically based example that may not account for subject to subject variation in anatomy. While its approach of algorithmic fitting plus human adjustment can be highly accurate, it is also dependent on image quality and human judgment to achieve a good patient-specific fit. Finally, it uses homogeneous and isotropic assumptions in its VTA methodology, which may introduce errors in the visualization due to variation of tissue properties within a specific patient’s brain.
SureTune™ is tightly integrated with other Medtronic surgical tools, allowing import of surgical information from Stealth™ navigation and planning software and the documentation of intraoperative electrophysiology which may further validate a patient-specific visualization. Medtronic’s current programming offering allows visualization of the VTA created by the device and the annotation of clinical observations. Future updates to SureTune™ will extend interoperability, allowing anatomical and electrophysiological information to appear on the programming tablet along with the stimulation volumes.
Boston Scientific GUIDE™ XT
The Boston Scientific neuromodulation stimulators use a system that includes both VTA modeling and functional testing. Visualization of VTAs with the Boston Scientific system is available through a software tool (GUIDE™ XT). GUIDE™ XT is compatible with the BrainLab Elements (Munich, Germany) surgical planning software, available for clinical use in many countries. The software enables 3D modeling and visualization of the VTA relative to the patient’s anatomical structures. To generate a model of lead location in the brain, a simulated DBS lead from a patient’s postoperative CT scan is registered to an anatomical atlas based on segmentation of a patient’s preoperative MRI. Clinical stimulation parameters can then be programmed onto the simulated lead to generate the associated VTA. In cases of suboptimal electrode placement or clinical response, additional options may be needed, and current steering between directional segments may be useful to expand the TW (Supplementary Fig. S2) . Visualization tools may be useful for understanding these nuances of directional DBS and adjusting stimulation appropriately. For example, the VTA created by coactivation of two segments has a different shape than the VTA created by activation of a single segment, with the former having a lesser radial extent of activation than the latter at the same amplitudes. The Vercise™ system also includes the Neural Navigator software, intended to aid in functional testing of different stimulator settings by mapping the resulting clinical effects. Therapeutic effects and side effects associated with monopolar stimulation on both ring and directional electrodes are represented on a 2D clinical effects map (Supplementary Figs. S3 and S4), and effects may be recorded on all leads for all stimulation configurations. VTAs for both standard DBS leads and directional DBS leads are also visualized within the Neural Navigator software, which can import the anatomical structures from the BrainLab platform.
While the Abbott Infinity™ clinician programmer does not have a VTA visualization tool, it includes the Informity™ programming software designed primarily for simplifying functional programming. The software guides the user through directional programming using visual representations of stimulation responses, occurrences of stimulation-induced symptom relief and side effects, to create an action plan of ranked electrode montage alternatives that may be needed over time during therapy. The layout and workflow of the Informity™ programming software enable “event markers” to document the amplitudes that produce symptom relief and side effects (Supplementary Figs. S5, S6, and S7). After monopolar survey of ring and segmented electrodes, the Informity™ software allows clinicians to rank the investigated montages according to various clinically relevant criteria. With traditional omnidirectional programming, amongst montages with comparable outcomes some clinicians prioritize maximizing the TW while others prioritize minimizing the TCS (power consumption). To that end, the decision support tool available within the Informity™ software not only allows ranking the montages on the basis of power (microwatts) and TW (milliamps) but also on the basis of a measure referred to as the therapeutic window percentage (TW%). TW%, which is TW expressed as a percentage of TCS, is a means to balance the trade-off between maximizing TW and minimizing TCS and gives clinicians a means to optimize gains in both TCS and TW simultaneously. A final sorting option, balanced threshold, gives clinicians an additional level of optimization to balance gains in TW% while minimizing power consumption as per clinician preference (Supplementary Fig. S7). The software allows complete customization of symptom relief and side effect lists and enables the export of the entire functional programming session in the form of a PDF report that can be exported to electronic medical records for future review.