Stereo EEG via multiple single-path omnidirectional trajectories within a single platform: institutional experience with a novel technique

Abstract

OBJECTIVE

Stereotactic electroencephalography (SEEG) is being used with increasing frequency to interrogate subcortical, cortical, and multifocal epileptic foci. We describe a novel technique for SEEG in patients with suspected epileptic foci refractory to medical management.

METHODS

Standard epilepsy evaluation and neuroimaging are used to create a hypothesis-driven SEEG plan, which informs the 3D printing of a novel single-path, multi-trajectory, omnidirectional platform. Following skull-anchor platform fixation, electrodes are sequentially inserted according to the preoperative plan. Platform and anatomic variables influencing localization error were evaluated using multivariate linear regression.

RESULTS

Using this novel technology, 137 electrodes were inserted in 15 focal epilepsy patients with favorable recording results and no clinical complications. The mean entry point localization error was 1.42mm (SD 0.98), and the mean target point localization error was 3.36mm (SD 2.68). Platform distance, electrode trajectory angle, and intracranial distance, but not skull thickness, were independently associated with localization error. This SEEG technique is feasible, easy to use, accurate, and affordable relative to competing technologies.

CONCLUSIONS

The multiple, single-path omnidirectional platform described herein offers excellent accuracy and clinical results, while avoiding cumbersome frames and prohibitive costs associated with robotic implantation.

INTRODUCTION

Accurate localization of epileptogenic foci is the fundamental tenant of preoperative evaluation for surgical epilepsy²¹. Traditionally, monitoring with scalp electrodes⁸, magnetic resonance imaging (MRI)^12,14, positron emission tomography (PET)¹⁵, neuropsychological testing⁷, and analysis of seizure semiology have provided the mainstay of non-invasive diagnosis and evaluation. Invasive monitoring with subdural grids and strips offers finer resolution of the epileptogenic zone, particularly in patients with a superficial, cortical focus. However, in patients without a superficial onset, subdural electrodes may provide inadequate localization to permit a safe and effective resection. Stereotactic electroencephalography (SEEG) offers interrogation of deeper structures and epileptic networks.

SEEG depth electrodes were originally inserted using a Talairach frame. Techniques have evolved to using stereotactic frames, frameless stereotaxy, and robotic insertion. Each technique carries its own set of limitations including prolonged operative time, accuracy shortcomings, and prohibitive device costs, respectively. In this report we describe a novel technique for implanting SEEG electrodes utilizing a personalized platform outfitted with a multitude of individual trajectories for complex electrode insertion. This technology provides an alternative method of implanting electrodes accurately, safely, and efficiently, while avoiding cumbersome, traditional frames or expensive infrastructure such as a robot. Our institutional experience with planning, manufacturing, and electrode implantation are described, along with a detailed analysis of device-related and anatomic contributors to electrode accuracy.

METHODS

Patients

All patients, pediatric and adult, at Vanderbilt University Medical Center undergoing SEEG utilizing the FHC microTargeting Multi-Oblique Epilepsy Platform (FHC, inc, ME, USA) from January 2015 to June 2016 were included in this study. Four surgeons performed these procedures. Patients were recommended for invasive monitoring after being diagnosed with refractory epilepsy and undergoing a non-invasive epilepsy evaluation including a combination of continuous video EEG, MRI, PET, MEG, SPECT, neuropsychological testing, fMRI, and/or WADA. After completion of the noninvasive evaluation, patients were discussed at a multidisciplinary surgical epilepsy conference including Neurology, Neurosurgery, Radiology, and Neuropsychology. The presentation resulted in hypotheses of the epileptogenic zone. Patients who were identified as having an epileptogenic zone that was deep, possibly bilateral, a part of an epileptic network, or with a history of failed subdural grid evaluation were considered candidates for SEEG. A hypothesis-driven SEEG implantation plan using a combination of orthogonal and oblique trajectories, as indicated, was formulated.

The workflow is described in detail below. In summary, the SEEG implantation plan is converted into trajectory plans on stereotactic MRIs. This can be performed at any point before surgery after the stereotactic MRI. One week prior to electrode implantation, patients undergo bone fiducial anchor (Waypoint Fiducial Anchors, FHC, inc, ME, USA) placement and stereotactic CT. The stereotactic CT is coregistered with the stereotactic MRI including the planned trajectories. The coregistered plan is then used for 3D printing of the platform. On the date of the implantation, the platform is fixated to the bone fiducial anchors and the electrodes are implanted as described.

Trajectory Planning

Patients undergo a stereotactic MRI to include T1 with and without contrast sequences, FLAIR, T2, and MRA sequences, with variable inclusion of MRV and CTV. MR sequences and MEG or fMRI data are coregistered within WayPoint Navigator software (FHC, inc, ME, USA). Next, patient anatomy is identified to generate the anterior commissure – posterior commissure (AC-PC) line, providing orientation for trajectory planning. Once registered, the previously agreed upon hypothesis-driven SEEG implantation plan is converted into electrode trajectories within the software. Trajectories are formed by choosing an entry point and target point. Depending on surgeon preference, the trajectories are planned in different display orientations including AC-PC view, trajectory view, or the original MRI orientation. To expedite the process for multiple trajectory planning, there are functions that allow trajectory cloning (for ipsilateral parallel trajectories) and mirroring (for contralateral trajectories). Once all trajectories are planned, each is evaluated for risk of vascular injury with a specific emphasis on the entry-site. Plans are adjusted as needed to avoid intersection of an artery or vein at the entry-site.

Once all trajectories are finalized, the software conducted an automated simulation of electrode implantation at each trajectory. Preprogrammed electrode dimensions exist from a multitude of electrode manufacturers. During implantation simulation, the surgeon may visualize the actual coverage that can be expected. The electrode length – or depth of insertion – is easily adjusted to ensure complete and accurate coverage.

Bone Fiducial Anchor Insertion

Approximately one week prior to lead insertion, patients are fitted with bone fiducial anchors. The bone fiducial anchors are screwed into the skull and act as localizers for construction of the platform and as fixation points for the platform at the time of electrode insertion. The number of fiducial bone anchors is variable depending on the planned trajectories (range 4-6 anchors). Anchors are placed through 5mm incisions using local or general anesthesia, depending on surgeon preference and patient factors. Location for bone fiducial anchors depends upon the number and location of trajectories. A stereotactic CT scan is then obtained with the bone fiducial anchors in place, allowing coregistration with the stereotactic MRI. Patients are discharged home the same day.

Platform Manufacturing

After coregistration, the stereotactic plan is then digitally uploaded to the manufacturer where the 3D platforms are created and shipped to the hospital, a process that takes roughly 3-5 business days (microTargeting Multi-Oblique Epilepsy Platform, FHC, ME, USA). Given often bilaterality, and the diversity of sites and trajectories, the platform system may contain up to 3 separate pieces, each with at least 3 anchor pedestals (Figure 1a).

Figure 1.

Three Starfix frames (A) for a single patient requiring bilateral, multi-lobed interrogation. A frame with 9 individual trajectories (B) is affixed to the skull whereupon electrodes are sequentially guided to their intracranial target.

Electrode Insertion

Approximately one week after bone fiducial anchor placement, patients return to the operating suite for electrode insertion. After inducing general anesthesia, the head is fixated in 3-point Mayfield pins if necessary and the scalp is prepped. The incisions overlying the bone fiducial anchors are opened and the anchors exposed. The platform is then fixated to each of the bone fiducial anchors (Figure 1b). A drill guide is inserted into one of the stereotactic platform trajectories. A wrench is then inserted down the guide to mark the skin for a 3mm incision. A high-speed drill bit is fixed with a depth stop for the skull thickness. The burr hole is then drilled through the platform instrument guide. Next, an obturator is advanced down the tool guide to assess the dural integrity. If the dura had not been incidentally opened with the drill, then an endoscopic monopolar was inserted and the dura was opened with cautery.

The workflow from this point forward varies depending on the electrode manufacturer. If the surgeon preferred an anchor bolt as available with Adtech (Adtech, WI, USA) or PMT (PMT Corporation, MN, USA), then the bolt wrench is passed down the tool guide and used to screw in the anchor bolt directly into the hole for stereotactic electrode insertion. An obturator with a depth-stop is passed down the bolt a distance premeasured from the platform plan. After reaching target, the obturator is withdrawn, and in similar fashion, an electrode with a premeasured depth-stop is advanced along the trajectory through the anchor bolt. The electrode is secured with a watertight twist cap, the depth-stop is removed, and the electrode is allowed to hang freely. The same sequence is repeated for all trajectories.

If Integra (Integra Life Sciences, Plainsboro, New Jersey) electrodes are used, a cannula-guided approach is used. The cannula and electrode depths for each trajectory are calculated and marked using depth stops. After the dura has been punctured as described above, reducing tubes are inserted through the drill guides to allow the cannula to be advanced to the desired depth and locked in place. The inner stylet is removed and the electrode advanced into location. A special C-arm attachment to the frame secures the electrode at a fixed distance from the frame. The outer cannula is then pulled back while the electrode is held in place. The electrode inner stylet is removed and extracranial portion of the electrode is pulled through the cannula in reverse; and secured to the skin using a nylon suture.

After successful insertion of all electrodes, they are tested with electrocorticography to confirm successful recording. The platform and bone fiducial anchors are removed, and the bone fiducial anchor incisions are closed with an absorbable suture. Patients are then emerged from anesthesia and transferred to the intensive care or epilepsy monitoring unit, depending on practice pattern of the surgeon. A postoperative stereotactic CT is obtained to establish the location of each electrode. This postoperative CT is coregistered with the preoperative plan in the planning software and electrode accuracy is assessed.

Error measurement and statistical analysis

To assess the accuracy and limitations of this implantation technique, the electrode trajectories and terminal locations displayed by the post-implantation stereotactic CT were compared with the preoperative plan. The Euclidean distance between the two was recorded both for entry point localization error (EPLE) and target point localization error (TPLE), as described in detail elsewhere.⁵ The entry-point was defined as the point at which the trajectory passes through the inner cortical skull surface. The target-point was defined as the planned termination of the electrode trajectory. The angle of lead implantation was defined as the degree off the 90-degree skull orthogonal. The platform to skull distance was defined as the linear measurement between the bottom surface of the platform and the skull surface. The skull thickness was measured along the trajectory of the electrode. The intracranial distance was measured from the inner cortical skull surface to the target-point.

Localization errors were summarized as a mean with standard deviation (SD), while non-normally distributed frame characteristics were described as a median with interquartile range [IQR]. Univariate linear regression was used to compare lead and anatomic characteristics against both localization errors. Measureable factors influencing localization error were sought. Parameters found to be significant to a p-value <.05 were incorporated into a multivariate linear regression model to identify independent factors responsible for localization error. All analysis and graph construction were performed with Stata, version 14 (StataCorp, College Station, TX).

RESULTS

The institutional experience includes 15 patients and 137 intracranial leads (mean 9) (Table 1). An average of 3 lobes were interrogated per patient (range 2-7). Six patients underwent bilateral electrode implants. The average operative time was 207 minutes, or 25 minutes per lead. The mean EPLE was 1.42mm (+/− 0.98) and the mean TPLE was 3.36mm (+/−2.68). There were no major complications and all patients had satisfactory EEG recording trials resulting in subsequent therapeutic procedures. In 4 patients, small extra-axial or subarachnoid hemorrhages were detected on routine post-operative CT. No patients experienced symptomatic implantation sequelae, and there were surgical adverse events requiring lead removal or replacement.

Table 1.

Patient and lead characteristics

ID	Age (sex)	laterality	# lobes	# leads	Operative time (min)	Operative time per lead (min)
1	59 (F)	unilateral	2	5	175	35
2	65 (M)	unilateral	2	6	224	37
3	24 (M)	bilateral	2	8	222	28
4	21 (M)	bilateral	7	8	221	28
5	39 (M)	unilateral	2	8	205	34
6	37 (F)	bilateral	2	11	237	22
7	49 (F)	unilateral	2	8	165	21
8	20 (M)	unilateral	2	8	278	35
9	32 (F)	unilateral	2	7	151	22
10	27 (M)	bilateral	4	9	183	20
11	23 (M)	unilateral	2	9	175	19
12	45 (M)	bilateral	6	10	194	19
13	17 (M)	bilateral	4	13	216	17
14	9 (M)	unilateral	3	12	174	15
15	15 (M)	unilateral	4	15	278	19

The distance from the platform to the skull was variable for each trajectory, averaging 43.2mm (36.4-50.7mm IQR). Similarly, the angle of lead implantation through the bone was routinely not orthogonal. The angle off the orthogonal averaged 25.3 degrees (15.3 – 35.5 IQR). Both increasing distance from the platform to the skull (Figure 2) and the increasing angle of implantation off of the orthogonal (Figure 3) correlated to worsening EPLE (coeff > 0, p<.05) (Table 3). Intracranial electrode distance to target was variable, averaging 41.7mm (34.9 – 50.7mm IQR). Increasing distance from the platform to the skull (Figure 2), increasing angle of implantation off of the orthogonal (Figure 3), and intracranial electrode distance (Figure 4) correlated to worsening TPLE (Table 2). Neither skull thickness nor the use of a bolt fixation device was significantly associated with localization error (p>.05).

Figure 2.

The influence of platform distance on entry point localization error (EPLE) (square) and target point localization error (TPLE) (circle). The shaded area represents the 95% CI.

Figure 3.

The influence of insertion angle on entry point localization error (EPLE) (square) and target point localization error (TPLE) (circle). The shaded area represents 95% the CI.

Table 3.

Relationship between localization error and lead parameters adjusted for covariates

Covariate	Coef.	SE	P-value
EPLE
Platform to skull distance	0.027	0.0084	0.002
Intracranial distance	NA	NA	NA
Angle from orthogonal	0.016	0.007	0.022
TPLE
Platform to skull distance	0.066	0.023	0.005
Intracranial distance	0.08	0.018	<.001
Angle from orthogonal	0.008	0.019	0.68

Figure 4.

The influence of intracranial distance on target point localization error (TPLE) (circle). The shaded area represents the 95% CI.

Table 2.

Univariate relationship between localization error and lead parameters

Covariate		EPLE		TPLE
	Median [IQRI	Coef.	p-value	Coef.	p-value
Platform to skull distance, mm	43.17 [36.4–50.7]	0.0312	<.001	0.066	0.008
Skull thickness, mm	6.7 [5.2–8.9]	0.041	0.196	0.076	0.393
Intracranial distance, mm	41.7 [34.9–50.7]	NA	NA	0.08	<.001
Angle from orthogonal, degrees	25.3 [15.3–35.5]	0.023	0.001	0.043	0.03
Bolt (ref. No bolt)	NA	−0.01	0.964	−0.32	0.551

In the multivariate model, increasing distance from the platform and angle of implantation were independent predictors of worsening EPLE (Table 3). The angle of implantation was found not to contribute significantly for TPLE. The remaining metrics, increasing distance from the platform and intracranial distance were independent predictors of an increased TPLE.

Additionally, there was an interaction effect demonstrated between platform distance and angle of implantation. As the angle from orthogonal increases, and as the distance between the platform and the skull increases, the associated increase in EPLE was observed to be greater than what would be expected from the sum of the individual effects. This interaction term – or synergism – itself was an independent predictor of error (p=.004) and is illustrated via a heat map pictorially describing this relationship (Figure 5).

Figure 5.

The influence of platform distance and insertion angle on entry point localization error (EPLE). The vertical legend on the right represents EPLE in millimeters. EPLE increases synergistically with distance and angle.

DISCUSSION

Herein, we describe our experience with the FHC microTargeting Multi-Oblique Epilepsy Platform outfitted with numerous, single-path trajectories to interrogate epileptic foci. To the authors’ knowledge, this is the first description of a frameless, platform-based device utilizing individual trajectory guides for the insertion of multi-directional depth electrodes in SEEG. We report a mean EPLE of 1.42mm and a mean TPLE of 3.36mm, comparable to other non-robot-based technologies.²⁰ The surgical goal was achieved in each case and no clinical complications were observed. Additionally, we demonstrate the relationship between localization error and modifiable factors including platform distance, insertion angle, and intracranial distance traversed.

For surgical epilepsy, the most important predictor of success is accurate localization of the epileptic zone (EZ)^6,17. While subdural grid electrodes provide the foundation for EEG evaluation of neocortical epilepsy, their utility diminishes with distance from the brain surface and their feasibility wanes with bilateral or noncontiguous targets. SEEG, therefore, is valuable under several circumstances: a deep EZ, such as within the temporal lobe, cingulate gyrus, or insula; previously failed or non-diagnostic subdural grid electrodes; and bilateral or otherwise topographically disparate candidate foci^9,11.

Accurate placement of depth electrodes demands detailed anatomic mapping both of the lesion and of the relevant structures to be avoided, and begins with precise coregistration of preoperative cranial imaging with fixed surgical anatomy. Generally, the SEEG platforms can be categorized into frame-based and frameless systems, each carrying its own advantages and disadvantages⁶. Some frames support only multiple parallel and orthogonal trajectories, while others allow oblique trajectories capable of maneuvering about important neurovascular structures¹. Though reliable and accurate, frame-based methods may generate patient discomfort (especially in the pediatric population), consume valuable operative time, limit the operative field, and prohibit refinement or addition of trajectories intraoperatively^16,19.

These drawbacks have provided the motivation for depth electrode insertion using frameless technology. Robot-assisted SEEG is one such method that allows implantation of multiple electrodes in virtually any direction, providing the surgeon with unique flexibility⁵. Surgical times can be truncated by eliminating the need for repeated frame coordinate adjustments¹⁸. Additionally, intraoperative modifications to trajectories is permissible¹⁰. However, a price tag ranging from hundreds of thousands of dollars to $2.5 million for the device can deter adoption of this technology, particularly for modest-volume epilepsy centers³. Maintenance of the technology and additional training required for some adaptable parts may also serve as barriers to a smooth transition to robot-assisted implantation.

While SEEG has undergone many refinements since its inception a half-century ago², the approach described in this report offers several important advantages relative to its predecessors. Its comparatively simple design combines speed and versatility to deliver a surgeon-friendly experience without compromising patient comfort. Because all lead trajectory planning is conducted prior to the date of surgical implantation, neither immediate preoperative nor intraoperative imaging is necessary. In a similar frameless design, Balanescu and authors report a 50% reduction in the time to place a single electrode relative to an analogous frame-based design¹. Time is also saved in the application of the platform, which typically can be done in less than a minute by securing the platform legs to corresponding bone anchors. As with other frameless systems, the platforms have been specifically manufactured to fit precisely on the bone anchors. Accordingly, registration of the preoperative imaging to the stationary bone markers obviates the need for a cumbersome frame placement, also serving to reduce operative time. Furthermore, the frameless designs means the head need not remain in complete fixation throughout the case. The FHC platform is particularly beneficial in pediatric patients (3 of our 15) who may be less likely to tolerate rigid fixation around a frame.

As illustrated by this case, the FHC individual-trajectory platform can accommodate complex, multiple, and bilateral targets in the same patient, while avoiding important neural and vascular structure. This cohort also illustrates how omnidirectional electrode insertion can be achieved without sacrificing accuracy. With a mean EPLE of less than 1.5mm, our results fall within the acceptable ranges. Our mean TPLE is more than twice this Euclidean distance at 3.36mm, but the reader is cautioned against concluding a discouraging result. Unlike other intracranial stereotactic procedures (deep brain stimulation, biopsy), EPLE – not TPLE – is the crucial measurement. It is at the entry point wherein most hemorrhagic complications occur, and placement errors can result in symptomatic hemorrhage. In SEEG, the trajectory, not the TP, determines recording success. In a recent editorial on the subject, Professor Cardinale notes “the TP represents not a real target, but just the deepest point of the trajectory.” 4 Validation in a larger patient cohort is warranted, but these results indicate an omnidirectional frameless platform can offer excellent stereotactic lead placement capabilities.

Finally, this technology can be easily adopted with relatively little capital investment by hospitals. Wherein a robot can cost millions, the Waypoint Navigator software and FHC platform are far more affordable, and compatible with most modern imaging technology. Moreover, similar frameless technology is already being utilized by neurosurgeons for implantation of deep brain stimulation (DBS) electrodes.¹³ Thus, familiarity with the technology by some functional neurosurgeons may ease the learning curve when applied to epilepsy patients requiring SEEG.

These advantages notwithstanding, there exist important considerations that must be made before adopting the technique. Three limitations warrant specific elaboration: first, two separate procedures (bone anchor placement & electrode insertion) are required, while most frame-based methods can be completed in just one. Naturally, the additional OR time and inconvenience to the patient must be weighed. In our institutional experience with this technology (primarily for DBS), the bone anchor placement is quick and uniformly well tolerated¹³. Second, there is little forgiveness for manufacturing mishaps or interruptions. A defective platform design or shipping delay, for example, could conceivably cause surgery postponement or cancellation. Essentially, an extraneous entity beyond the immediate control of the surgeon is introduced leaving open a small, but plausible window for error. Among more than 1000 cases performed at our institution using this technology for DBS, only 2 such instances have arisen. Third, intraoperative adjustment of a given electrode trajectory is not possible. Each path is itself a rigid column through which passes the sheath, obturator, drill, and electrode in sequence. As could be done with robot-technology, fine-tuning the insertion point and angle with this technology is not permitted. Minor depth adjustments, meanwhile, are permissible and can be achieved without compromising the length-dependent recording feature of SEEG electrodes.

Unfortunately, operative times are difficult to interpret, as there was significant variability depending on surgeon and experience with these cases. Limited case number per surgeon precluded a quantitative comparative analysis, or reasonable detection of a learning curve. Though subjectively, in our experience subsequent surgeries were met with smoother and swifter operative flow.

Our results also suggest that relatively simple adjustments to the existing planning algorithms can improve accuracy. Based on initial experience, modifications are being made to account for the affects of platform to skull distance and angle from orthogonal. Efforts are made to reduce the platform-to-skull distance, although a minimum distance is required depending on the electrode manufacturer and technique of insertion (bolt vs. no bolt). Longer drill guides are also being used that provide less distance below the frame for drill flexing and scything. Additionally, although non-orthogonal angles from the bone surface are a necessity in SEEG, efforts are made in planning to avoid the extremes of oblique angles to prevent scything. We expect the manufacturing alterations made based upon these results will minimize localization error in our future practice.

Despite these limitations, the advantages offered by the omnidirectional platform make the technology an attractive alternative to more traditional SEEG methods. comparison between patients undergoing frame-based and frameless SEEG (with and without robot-assistance), may help determine which technology offers superior targeting and recording results, economic output, and patient satisfaction. With the ultimate goal of seizure freedom, further efforts to create an instrument which maximizes efficiency, while reducing costs and discomfort should be pursued.

CONCLUSION

We report an institutional cohort of 15 patients with complex medically refractory epilepsy who undergo SEEG via a novel frameless technology. The localization accuracy, ease of use, and favorable cost profile depict the omnidirectional platform as a feasible and promising technology. Engineering efforts to lower the platform height and avoid the use of shallow entry angles, may further minimize localization errors. Larger cohorts and the experience of other centers are needed to validate this technology before it is considered among the primary tools in the SEEG armamentarium.