Creation and Evolution of SEEG

Stereoelectroencephalography (SEEG), a stereotaxic approach for exploration of the pharmaco-resistant epilepsies, was created 65 years ago by J Talairach, MB Dell and J Bancaud at Hospital Sainte-Anne, Paris. For the first time ever, it allowed recording of patients’ seizures in the three-dimensional brain space. The principles of SEEG, based on establishing anatomo-electro-clinical correlations in the patient’s seizures, did not change since 1965. But the technologies for anatomical topography, electrophysiology, clinical observation, and surgery have dramatically evolved. However, it remains a complex investigation and a hard to standardize procedure, since it must always be tailored to the patient’s individual case.

After the publication of the “SEEG in Epilepsy” book in 1965 by Bancaud and Talairach and their collaborators, the 70’s were marked by multiple papers on semiology and localization, and by three major books: (i) Angiography of Cerebral Cortex by Gabor Szikla, (ii) EEG/SEEG Surface/Depth Correlations by Jean Bancaud, (iii) and the Novel Approach of Epilepsy Surgery by Jean Talairach. Due to a tight collaboration between the research labs of the INSERM Unit 97 “Unite de Recherches sur l’Epilepsie” and the Department of Neurosurgery at Hospital Sainte-Anne, the publications on pathophysiology of focal epilepsies emerged during the 70’s and 80’s, with papers on temporal and frontal seizures semiology, ^, the cingulate gyrus, the role of the frontal cortex in generalized seizures, mechanisms of startle epilepsies, as well as monkey models of motor cortex epilepsies, role of cortical monoaminergic terminals, ^, and alteration of amino acids neurotransmitters with iontophoresis and ion-sensitive microelectrodes. In parallel, a new approach in cognitive neurophysiology developed in SEEG on declarative memory and on auditory cortex physiology. The great originality of the clinical/research group structure led to the production of new concepts, like the network hypothesis on the epileptogenic zone, all inspired by the practice of SEEG.

SEEG developed only in Hospital Sainte-Anne, Paris until the mid-70ies. Then Talairach and Bancaud’s pupils exported the method: Heinz-Gregor Wieser in Zurich, Guy Bouvier in Hospital Notre Dame, Montreal, and Alain Rougier in Bordeaux. Many neurologists and neurosurgeons from South America visited Sainte-Anne in the 70’s and 80’s, and some of them stayed longer than a habitual visit as they were political refugees from Chile and Argentina. Osvaldo Betti developed stereotactic radiosurgery with Gabor Szikla then came back to Buenos Aires. Silvia Kochen, a neurologist, after having been a pupil of Bancaud for 2 years, launched an SEEG-based epilepsy surgery program in Buenos Aires. In the 80’s, just after Jean Talairach retired, the pioneers’ team was decimated by the deaths of Gabor Szikla (stereotaxy surgeon) and Alain Bonis (neurologist). Antonino Musolino and Claudio Munari became in charge of stereotaxy, Suzanne Trottier and Patrick Chauvel in charge of neurological care. Jean-Paul Chodkiewicz succeeded Jean Talairach in 1980, and Patrick Chauvel succeeded Jean Bancaud in 1986.

The group split up in 1990. Claudio Munari moved to Grenoble, France and Patrick Chauvel to Rennes, France. Each developed new EMUs with SEEG. Philippe Kahane, Munari’s first resident, was instantly fascinated by SEEG, and Jean-Pierre Vignal, who was in the Sainte-Anne team, accompanied Chauvel. Then, with Munari training Jean Isnard, and Chauvel training Philippe Ryvlin, a new group emerged in Lyon. A few years later, Chauvel moved to Marseille and Munari to Milan. Each of the branches differentiated and enriched the SEEG compendium. Munari worked on new stereotaxic implantation modes like hypothalamic hamartomas and insula oblique trajectories. He actively promoted SEEG-guided epilepsy surgery in children, and carefully studied the relations between lesions and epilepsies. After his premature death, his Niguarda team in Milan led by Giorgio Lo Russo, Laura Tassi and Stefano Francione performed refined correlations with histopathology in malformations of cortical development with Roberto Spreafico and studied the sleep-related frontal seizures with Lino Nobili. ^, Chauvel had been interested since the early 80’s by semiology and physiology of motor seizures then of frontal seizures in general. With his American colleague Antonio Delgado-Escueta who imported the SEEG method to the US, he organized the jubilee celebration for Jean Bancaud in 1987. Gathering the most renowned specialists at that time, they published the first book on frontal lobe seizures and epilepsies in 1992. This book was the first initiative to disentangle the complexity of frontal seizures’ semiology and electrophysiology. Then the Chauvel branch initiated a multidisciplinary strategy based on SEEG. It assembled neurologists, neurophysiologists, neuropsychologists, cognitive neuroscientists, and biomedical engineers working as one team. Eric Halgren and Catherine Liegeois-Chauvel were the first to study cognition from intracerebral recording data. ^, Jean-Louis Coatrieux and Fabrice Wendling adapted signal processing methods to the SEEG signal and were the first to design computational models to interpret it. This innovative approach was born in Rennes. The same philosophy was applied after Chauvel moved to Marseille (in 1997), where Jean Regis was elaborating a functional surgery program based on stereotaxy. The Institut de Neurosciences des Systemes (INS) reproduced at a bigger scale what had previously been validated in assembling epileptology, neuropsychology, cognitive neurophysiology with basic neurophysiology and applied mathematics. The link between all these disciplines was the SEEG method.

Meanwhile, Ryvlin and Kahane animated a Lyon-Grenoble axis and developed epilepsy imaging and engineering approaches. Jean-Pierre Vignal and Louis Maillard, both pupils of the Chauvels at different times, created a new center in Nancy, France with Herve Vespignani. Luc Valton, who had been trained in Marseille, opened a new center in Toulouse, France, and pursued the same philosophy with Emmanuel Barbeau who collaborated with Liegeois-Chauvel when he was in the same Institute. The way this web grew over time in 10 years is interesting, as it is a team story and a human adventure.

Apart from a few exceptions mentioned above, the School of SEEG remained in Paris for 30 years (1960–90). The initiative of starting a new program and taking responsibility for preparing an electrode implantation map away from the shrine was experienced as scary by first-generation pupils even after they had spent some 20 years with the Master. Jean Bancaud died in 1993. The few French Sainte-Anne alumni felt the need to initiate an annual French master class of epilepsy that they named “Ecole Pratique Jean Bancaud”. Their objectives were to keep on exploring together the complexity of the method and to teach the young neurologists on semiology as study of its mechanisms became accessible from analysis of many cases investigated with SEEG. This Ecole is still active today and takes place yearly in France; it took the name of “Ecole Pratique Bancaud-Talairach” after the death of Jean Talairach in 2007.

Hans Luders had got to know the SEEG pioneers in 1987 when he attended Bancaud’s jubilee symposium on frontal lobe epilepsies in France. In the early 2000s, there were debates between Cleveland Clinic subdural grids (SDG) defenders and the French SEEG protagonists on the respective accuracies of the two methods for epileptogenic zone localization. Then Luders and his colleagues published data on SDG morbidity and then decided to evaluate the capabilities of SEEG. Invited by Alim Benabid in Grenoble and in Milan after Claudio Munari’s death, they got convinced and marked an historic turning point in adopting the method. A few years later, Jorge Gonzalez-Martinez was sent on mission in France and came back after being trained in the stereotaxic method.

Thirty years after Antonio Delgado-Escueta imported Bancaud’s concepts at UCLA, Imad Najm invited Patrick Chauvel to join the Cleveland Clinic team. Chauvel retired from Aix-Marseille University Department of Clinical Neurophysiology and Institut de Neurosciences des Systemes and was succeeded by Fabrice Bartolomei and Viktor Jirsa. Catalyzed by the Cleveland Clinic international influence, the last 10 years have seen a worldwide development of SEEG-guided epilepsy surgery. As a side effect of a new French-American functional connectivity, SEEG secondarily spread in Europe outside France and Italy, and around the world (especially China, Taiwan, India and Australia, as well as South America and the Middle East).

The real benefit of such fast expansion of SEEG for epilepsy surgery is difficult to estimate. The method is not straightforward and cannot be standardized. SEEG is not depth EEG (Local Field Potential -LFP- is the signal, connectivity is the framework). A multidisciplinary and close-knit teamwork is required. The body of knowledge that is relevant to perform SEEG is not theoretical and bookish but based on a real apprenticeship and training through mentoring. SEEG requires to think outside of the box, so that the learning curve is steep. Indications for SEEG should critically depend on the team’s SEEG experience. This is currently not the case.

The next future will tell whether we are now attending an apotheosis or contemplating the apoptosis of SEEG.

The current difficulties faced by SEEG are due to multiple conceptual and technical/practical pitfalls. They could be summarized in saying that (i) most of the apprentices do not take SEEG as a comprehensive method; (ii) neurologists and neurosurgeons often consider SEEG as a depth version of grids, so that their first preoccupation is “to cover” a maximal volume of brain or to “target deep structures”; (iii) a crucial conceptual leap remains to be made for a correct analysis of the SEEG data.

SEEG as a comprehensive method . SEEG doesn’t start with electrode implantation. As implantation relies upon precise hypotheses of epileptogenic networks involved, the data for their generation must be collected in phase I. As a team is gaining experience in SEEG, the content of phase I becomes more and more elaborated and the hypotheses more accurate. The master word here is semiology. Semiology is the study of signs. Semiology is not limited to clinical signs but also encompasses electrical and imaging signs. There is a clinical, an electrical and an imaging semiology in epilepsy. A sign is descriptive in nature: it has no absolute value; it always must be “interpreted”. In the SEEG method, signs are interpreted through a correlative process: the goal of video-EEG recording of seizures is to establish electrical-clinical correlations; imaging semiology describes the radiological signs of a putative epileptogenic “lesion” and as such it must be anatomically compared to the results of electrical-clinical correlations. Phase I is definitely the most important step in the SEEG method. If you do not get from video-EEG clearly explicit and falsifiable hypotheses in terms of epileptogenic networks, an implantation plan cannot be designed. Unlike SDG where indecision on localization leads to lateralize and to “cover”, SEEG can be properly performed only when prior estimate of seizure onset and propagation networks is available. Reaching this level of competency is not immediate. Therefore, there is a learning curve for a team to get expertise in electrical-clinical correlations. An apprentice team should not attempt to do SEEG in non-straightforward cases.

Unlike the two-dimensional SDG, SEEG is a three-dimensional brain exploration . Electrode labeling of an electrode as its deep target is a proof of incomprehension of the method. Adjusting electrode trajectory is the essential task to be performed before SEEG. A correct estimation of the epileptogenic zone localization can only be achieved by comparison of seizure onset with its early and late propagation: this is the process that allows to define the primary organization of the ictal discharge. If too many electrode contacts are recording white matter or are not optimally placed in propagation areas of the network, SEEG interpretation becomes arbitrary. Electrode implantation plan must take account of three factors: anatomical sampling, volume sampling and eventual surgical landmarks. There are always one or several missing electrode(s). The systematic use of anatomo-electro-clinical correlations allied with direct electrical stimulation is a means of compensating for a suboptimal implantation.

Getting familiar with SEEG profoundly changes the way that we apprehend epilepsy . The simplistic notion of focus can no longer be envisaged, as the existence of multi-structural (i.e., amygdala, hippocampus, and entorhinal cortex; or orbitofrontal and anterior cingulate cortex; or insula and cingulate gyrus) epileptogenic zone is a prominent feature. A multi-structural seizure onset generating multi-directional cortico-cortical propagation simultaneously and bilaterally in different lobes is a common observation especially in supra-sylvian and parietal-occipital epilepsies. To identify an epileptogenic zone (i.e., the minimal network capable to synchronize the areas involved at seizure onset thus triggering the patient’s seizures) among multiple areas involved in a seizure dynamical state remains a challenge every time. To use too restricted criteria like seizure onset time and relative latencies (a usual practice in SDG) is inadequate in SEEG where frequency patterns prove to be the most accurate marker. However, a still undetermined “epileptic cutoff frequency” is inadequate as a concept, and a sampling issue is always present. The diagnostic situation here can be likened to an inverse problem, where the use of a priori information as a constraint allows to minimize the number of solutions. Given that the available data are multimodal (anatomy, electrical spontaneous or stimulation-triggered activities, current and anamnestic clinical semiology), a clinical reasoning process must be followed to make decisions on surgical strategy.

While SEEG is resetting your conceptualization of epilepsy, it will change your interpretation of EEG. The pioneers early understood that several conditions needed to be met for an epileptiform or an epileptic discharge to be recorded on surface EEG. For instance, the activities from depth sources are not recorded in routine scalp EEG. As a rule, hippocampal or amygdalar activities are not detected if they have not spread to lateral temporal cortex; the same is true for orbitofrontal discharges. However, interictal spikes can be detected from the scalp temporal electrodes if hippocampal and lateral cortex spikes are synchronous, or if the lateral cortex alone is discharging. Ictal fast activities are exceptionally recorded in scalp EEG; this also depends on the geometry of the generators. This consideration is to emphasize that most of the time teams starting with an SEEG program do not understand how much they should invest in the quality of video-EEG. The number of scalp electrodes is insufficient, the montages are inadequate. MEG will never replace EEG, but its intrusion in the presurgical investigations brought new insights on source localization in a world only paying attention to the waveforms.

Current evolution of SEEG is full of paradoxes. It has swiftly proliferated outside of France since the last decade. Epileptologists and neurosurgeons are satisfied with its clinical tolerability by patients. Advances in robotics have simplified the stereotaxic procedure. However, it is too often used as a technique for targeting deep structures rather than as a comprehensive anatomical method to design trajectories of multi-lead electrodes allowing cortical networks exploration. SEEG being presumed less invasive than SDG, neurosurgery is attempting to promote less invasive techniques like laser interstitial thermal therapy or thermocoagulation, as if the size of the epileptogenic zones had decreased with technical evolution. In parallel, video-EEG amplifiers technology has steadily progressed so that hundreds of channels can be managed in smaller size and weight boxes. Higher sampling rates and better digital filters allow good quality recording of infra-slow to high-gamma activity. However, Epilepsy Monitoring Unit architecture and technicians/nurses training are far from being optimal in many centers. Per- and post-ictal semiology assessment is often insufficient. This nonoptimal phase I increases the risk of inadequate electrode implantation plan with too many electrodes in an attempt “to cover” a maximal brain volume like in SDG approach. “Depth grids” are replacing cortical surface grids. Obviously, this situation is a direct consequence of time/productivity increased pressure on hospital personnel by health system economic constraints. From this perspective, SEEG is a complicated and expensive method.

Will technological advances compensate for the workforce problem? Recent advances in signal processing allow a rapid quantification of interictal activities and guided identification of epileptogenic zone. But the medical/scientific training requirements might become an issue in the future. SEEG is not ECoG. A neuroscience/neurophysiology background is mandatory, a specific academic environment is required, so that research programs can be developed in tight relation with the epilepsy team. Its electrophysiology competency will directly benefit from a multidisciplinary environment. Improvement in clinical semiology assessment depends not only on the material conditions above mentioned, but also on the team interest in neuropsychology, behavioral neurology, and cognitive neuroscience in general. In an SEEG-based epilepsy surgery program, the specialized team includes nurses and electrophysiology technicians. Specific teaching must be organized on a permanent basis, with video-EEG and SEEG interpreted together. A solution for improving good practices in the SEEG method would be to reconfigure EMU architecture, for instance from a rectangular to a circular shape with nurse/tech team in the center. This would facilitate rapid and efficacious access to the patient. With the reading rooms in the same space, the neurologists/nurses/technicians interaction would be greatly facilitated. Performance in SEEG-guided epilepsy surgery critically depends on the team learning curve, much more than on technological advances.

A global analysis of the evolution of SEEG shows that its principles and its mode of clinical reasoning have not significantly changed since its creation. Even after major advances in neuroscience, its basic concepts (anatomo-electro-clinical correlations, stimulations procedure, epileptogenic zone, etc.) are not outdated. What its pioneers had early perceived about the nature of epilepsy has now received a scientific basis. “Primary organization” and “multidirectional propagation” of focal seizures are now expressed in terms of neural networks features. Ictal “fast activity” remains a reliable marker of an epileptogenic zone and has now been identified as an abnormal gamma-range oscillation. ^, Cognitive neuroscience and electrophysiology have profoundly changed the landscape around SEEG with extension of the indications of explorations and operations as a corollary. The use of “functional mapping” has now invaded SEEG practice. This new custom is a little awkward as the inherent sampling bias of SEEG prevents from real mapping from taking place. In the classical SEEG, functional mapping was performed through ictal anatomo-electro-clinical correlations with special attention to functional impairment. The seizure was used to map the functions. Utilization of stimulation behavioral data for functional surgical outcome prediction is only licit in primary cortices. They are a few synapses away from sensory receptors or muscle effectors with which they are topically hardwired. From a hodological perspective, the relationship between associative cortices and cognitive functioning is far more complex. This is why functional impairment through stimulation is more easily obtained with white matter than “language areas” in SEEG. As cognitive functioning is underpinned by network activity, a disruptive stimulation effect does not necessarily prognosticate a post-surgery functional deficit following ablation of the area stimulated. Development of other techniques, like transient cortical cooling, would be more appropriate for an accurate anticipation of functional outcome.

SEEG has steadily evolved over the last decades. From the 24 channels ink-writing EEG machine acute recordings (SEEG lasted 12 h) at Hospital Sainte-Anne in the 60’s to the current 256 channels chronic video-SEEG monitoring, the technique has gradually progressed. From the pneumoencephalography and contrast ventriculography practiced in the stereotaxic frame used for the pre-SEEG “reperage” ^, to the structural and functional brain imaging incorporating the Talairach system, the accuracy of electrode localization has tremendously improved. Robotic systems for electrode implantation have also increased anatomical precision and largely contributed to SEEG expansion in simplifying the stereotaxic procedure. SEEG signal processing brings refined markers of the epileptogenic zone while providing valuable insights into the pathophysiology of human focal epilepsies. Video image processing with the help of AI is now able to classify clinical semiology sequences, awaiting for the clinicians to put much effort into it. At last, emphasis on concerted research approaches combining simultaneous surface (EEG-MEG) and depth recording with imaging should eventually open a passage to fully non-invasive presurgical investigations. SEEG is the ideal tool for this transition. A glorious future for SEEG could therefore be its gradual disappearance.