Abstract
Dessert GE, Thio BJ, Grill WM. Brain Commun. 2023;5(6):fcad304. doi:10.1093/braincomms/fcad304. PMID: 38025277; PMCID: PMC10655844 Stereo-EEG is a minimally invasive technique used to localize the origin of epileptic activity (the epileptogenic zone) in patients with drug-resistant epilepsy. However, current stereo-EEG trajectory planning methods are agnostic to the spatial recording sensitivity of implanted electrodes. In this study, we used image-based patient-specific computational models to design optimized stereo-EEG electrode configurations. Patient-specific optimized electrode configurations exhibited substantially higher recording sensitivity than clinically implanted configurations, and this may lead to a more accurate delineation of the epileptogenic zone. The optimized configurations also achieved equally good or better recording sensitivity with fewer electrodes compared with clinically implanted configurations, and this may reduce the risk for complications, including intracranial hemorrhage. This approach improves localization of the epileptogenic zone by transforming the clinical use of stereo-EEG from a discrete ad hoc sampling to an intelligent mapping of the regions of interest.
Commentary
During the mid-1900s, Talairach et al. (1962) pioneered stereo electroencephalography (sEEG), aiming to develop a 3-dimensional system for optimal brain region recording. 1 Their objectives extended beyond accuracy in targeting brain structures; they sought to significantly improve recording efficiency. 2 In contemporary clinical practice, sEEG workflows typically involve drafting trajectory plans, considering entry and target points, followed by vascular imaging to refine plans and prioritize safety. However, it is not routine practice to use modeling techniques, which account for intrinsic properties of recording electrodes and surrounding tissue, to inform comprehensive sampling of all intended areas.
In this study, the authors devised algorithms that calculated the spatial recording sensitivity of sEEG electrodes and optimized sEEG montages, by leveraging patient-specific imaging and biophysical modeling with the goal of better delineating the epileptogenic zone. 3 This novel study aimed to maximize the recording sensitivity of sEEG configurations and reduce the number of implanted electrodes while ensuring comprehensive recording coverage. Using patient-specific imaging, the authors created patient-specific head models that accounted for the complex electrical properties of tissue within the head. Using the models, they determined the cortical tissue that was recordable by a sEEG electrode and by extension a sEEG montage. Sources near the electrode contact may generate smaller signals compared with those further away, based on the orientation of the sources within the active tissue and the location of the contact. Given the complex relationship between contact location and recording sensitivity, patient-specific extended source modeling is needed to determine the recording sensitivity of each electrode contact. Next, the authors optimized clinically relevant sEEG montage parameters including electrode insertion angle, maximum length, and distance to neighboring structures to map 3 regions of interest: the left temporal lobe, a clinician-defined region of interest, and the left hemisphere. Through this approach, the authors demonstrated that their algorithm could maximize the recording sensitivity within a region of interest using the same number or reduced number of implanted electrodes without sacrificing recording sensitivity.
The initial plan for sEEG targeting to pinpoint the epileptogenic network is primarily driven by a clinical hypothesis, combined with EEG and imaging with little consideration for recording sensitivities. These algorithms have the potential to save numerous hours during the initial templating and refinement stages, all while preserving and enhancing sampling quality. The authors conclude that optimized sEEG electrode configurations, based on patient-specific models and simulation of voltages generated by cortical dipole sources, yield sEEG montages with substantial increases in recording sensitivity and fewer implanted electrodes. These findings address real clinical dilemmas, as neurosurgeons may need to revisit sEEG placement when seizure recordings reveal that certain areas within the seizure network are either inadequately targeted or overlooked due to sparse electrode coverage. This situation contributes to patient discomfort, requires additional resource allocation, longer lengths of stay, and entails the risks associated with an additional surgical procedure. Additionally, the most severe risk associated with sEEG is intracranial hemorrhage, and reducing the number of electrodes may be a factor in mitigating this risk. The authors showcased the importance of accounting for spatial recording sensitivity by highlighting a paradox in clinically implanted sEEG montages where a seemingly dense electrode montage might overlook crucial areas and result in patchy sampling or redundant sampling. However, to enhance the impact of the study, the authors could have identified specific anatomical regions that were overlooked by overlaying atlases, which would have further elevated its significance to clinicians.
The authors propose that in clinical settings, the identification of interictal spikes usually depends on amplitudes surpassing 1 mV. While this criterion applies in numerous cases, conditions like periventricular nodular heterotopia may not generate such amplitudes, necessitating the adjustment of recording sensitivity by the sEEG reader for both interictal and ictal activity. However, the authors acknowledge that forthcoming source localization algorithms could potentially leverage lower-amplitude signals to pinpoint neural sources that were previously imperceptible to clinicians. In practice, there may be variability in spacing between contacts in the implanted leads which is not considered in the study. A final limitation of this study, as acknowledged by the authors, is that although dipoles serve as suitable source models for sEEG, discrepancies arise in voltage outputs between dipole models and realistic neurons within 1 mm of sources. 4 In these instances, the dominant higher-order components of the current multipole contribute to the peak of voltage spikes, while dipole sources may fail to represent high frequency oscillations, which are clinically important. Nevertheless, this serves as an additional avenue for future research on sEEG modeling tools.
While recording sensitivity is clearly significant, it is crucial to factor in the clinical perspective when planning sEEG montages. Take, for instance, a case involving temporal encephalocele, where sampling may be restricted to a few electrodes covering the area around the encephalocele/temporal pole, along with the mesial temporal region. In such situations, while recording sensitivity remains a consideration, the team frequently finds themselves deliberating between a limited resection and a standard anterior temporal lobectomy, thereby diminishing the relevance of recording sensitivity for the entire temporal lobe. Nevertheless, the realm of sEEG calls for automation. Despite the development of several algorithms to plan sEEG montages automatically, these optimize different aspects of sEEG montages without comprehensive integration into standard clinical practice. It is crucial to integrate features from diverse algorithms, incorporating methods for robustness in terms of recording sensitivity, gray matter coverage, avoidance of vasculature and minimizing implanted leads. 5 Additionally, these algorithms need to be integrated with functional connectivity and source imaging data obtained from scalp EEG recordings. Finally, these algorithms should also address supplementary objectives of sEEG implantation, such as testing of eloquent cortex, delineating resection boundaries, planning radiofrequency ablation, and planning for permanent neuromodulation devices. Therefore, there remains a need for clinical expertise in planning that must not be overlooked, which underscores the importance of clinical supervision and flexibility within these automated systems.
Certainly, there is an art to sEEG planning, akin to many aspects of medicine, but it is time to merge the artistry with scientific principles of sEEG when it improves clinical practice. Therefore, it is essential for clinicians to utilize sEEG in conjunction with modeling tools to sample accurately the epileptogenic zone, rather than in isolation, to realize the vision upon which sEEG was originally created.
