Stereoelectroencephalography¶

A patient-specific seizure-network hypothesis is sampled in three dimensions with stereotactically implanted multicontact depth electrodes, and intracerebral recordings, stimulation, anatomy, and clinical signs are integrated to guide epilepsy treatment.

Version: v2 · 2026-09-06 · History
Domain-specific #: 2854
Origin domain: clinical neurophysiology
Subdomain: invasive epilepsy evaluation
Aliases: SEEG, Stereo-EEG, Stereo-electroencephalography

Core Idea¶

Stereoelectroencephalography (SEEG) is an invasive, hypothesis-driven method for investigating focal epilepsy with stereotactically implanted multicontact depth electrodes. It does more than record electrical activity from inside the brain. A multidisciplinary team constructs a patient-specific hypothesis about seizure onset and propagation from semiology, scalp EEG, imaging, neuropsychology, and other data; designs electrode trajectories to sample that hypothesized network while avoiding critical anatomy; records habitual seizures and interictal activity; may stimulate contacts to test function or reproduce electroclinical phenomena; and integrates the results to support a treatment decision.^[1]^[2]

The unit of inference is a sampled anatomo-electro-clinical network. Contact locations provide anatomy, intracerebral signals provide timing and local field activity, and observed clinical signs provide the behavioral sequence. Their correlation supports an interpretation of where a seizure begins, how it propagates, and which sampled structures are functionally important. Modern reviews describe SEEG as sampling superficial and deep structures to build a three-dimensional account of seizure-onset and epileptogenic networks.^[3]

SEEG is not a whole-brain camera. It measures only near implanted contacts, and electrode placement is conditioned on the preimplantation hypothesis. A negative contact does not prove an unsampled region is uninvolved. This hypothesis–sampling dependence is the abstraction's central epistemic constraint.

Structural Signature¶

The recurring relation is:

drug-resistant focal-epilepsy decision problem + patient-specific network hypothesis + stereotactic risk-constrained electrode plan + intracerebral recording/stimulation + anatomo-electro-clinical integration → revised epileptogenic-network model and treatment plan.

Mandatory roles are:

The clinical decision problem: uncertainty remains about a potentially treatable focal epileptic network.
The preimplantation hypothesis: candidate onset, early spread, and alternative networks are stated before trajectories are chosen.
The anatomical model: co-registered imaging and vascular/structural information support targeting and avoidance.
The stereotactic plan: multiple trajectories and contact locations sample the hypothesis in three dimensions under safety constraints.
The implanted depth electrodes: multicontact probes record intracerebral potentials from selected gray-matter structures.
The observation period: spontaneous habitual seizures and interictal patterns are recorded with synchronized clinical data.
The stimulation component: where indicated, electrical stimulation tests functional responses, connectivity, or electroclinical concordance.
The localization record: postoperative imaging and contact labeling establish where each signal was measured.
The integrated inference: anatomy, signal timing, seizure semiology, and stimulation are interpreted together.
The action boundary: resection, ablation, neuromodulation, further evaluation, or no intervention follows only after risk–benefit review.
The sampling limit: claims are restricted to sampled regions and the adequacy of the original hypothesis.

A set of intracranial electrodes qualifies as SEEG only when it participates in this stereotactic, hypothesis-led, three-dimensional investigative method. A single therapeutic depth lead or incidental intracranial signal is not enough.

What It Is Not¶

It is not scalp EEG, which records summed potentials at the scalp with much coarser spatial access and no intracerebral implantation. It is not electrocorticography/subdural-grid monitoring, which samples cortical surfaces through grids or strips and has a different coverage/risk geometry.

It is not stereotaxy alone. Stereotaxy supplies coordinate-based trajectory placement; SEEG adds an epileptological hypothesis, multichannel intracerebral recordings, anatomo-electro-clinical interpretation, and a treatment decision.

It is not the electrode, robot, frame, planning package, or recording amplifier. These are realizations. It is not direct proof of the entire epileptogenic zone, because only selected tissue is sampled. It is not automatically a therapy, although the same contacts can support stimulation mapping or SEEG-guided thermocoagulation in some programs.

Scope of Application¶

SEEG is principally used in selected people with drug-resistant focal epilepsy when noninvasive studies do not adequately resolve a treatment-relevant hypothesis, especially for deep, bilateral, multilobar, discordant, or network-like possibilities. Indications, contraindications, electrode strategy, recording, stimulation, interpretation, and complication management are described in specialist consensus guidance.^[2]

The method can sample mesial temporal, insular, cingulate, opercular, sulcal, and other deep or distributed structures that are difficult to cover with surface grids. Comparative outcome research also emphasizes that SEEG and subdural-grid strategies have different morbidity and coverage profiles rather than being interchangeable implementations.^[4] Electrical stimulation through contacts can contribute to functional mapping and investigation of symptoms, while interpretation must distinguish induced responses from spontaneous seizure organization.^[5]

The data also support research in human cognitive neurophysiology and connectivity, but clinical electrodes are placed for patient benefit rather than uniform research coverage. Any secondary analysis inherits sparse, clinically selected sampling and heterogeneous pathology.

This entry is descriptive, not medical advice. Patient selection and implantation are high-stakes clinical acts requiring an experienced multidisciplinary epilepsy-surgery program and current local standards.

Clarity¶

Three targets are often conflated. The seizure-onset zone is where electrographic seizure activity is observed to begin under a stated recording. The epileptogenic zone/network is the tissue/network whose treatment is hypothesized to be necessary and sufficient for seizure control; it is an inferential construct, not directly measured as one signal. The eloquent or functional network comprises regions whose injury would cause unacceptable deficit. SEEG may inform all three, but they are not aliases.

Contact localization is part of signal meaning. A trace without verified anatomical position and reference scheme cannot support the full inference. Volume conduction, reference choice, artifacts, contact orientation, tissue pathology, anti-seizure medication changes, and limited seizure count affect interpretation.

“Negative SEEG” can mean no habitual seizure recorded, sampled contacts did not show the hypothesized onset, or the sampled network was inadequate. These outcomes have different implications. Absence of a signal at a contact is not absence throughout the brain.

Manages Complexity¶

SEEG converts a high-dimensional presurgical problem into a bounded experiment. The hypothesis chooses a finite set of trajectories; each contact becomes an anatomically located sensor; temporal relationships among contacts and symptoms test competing network accounts. This can resolve ambiguity that remains when scalp signals, lesions, and clinical signs disagree.

The method also separates access from interpretation. A robot or frame can place an electrode precisely, but precision does not guarantee that the right hypothesis was sampled. More contacts do not automatically improve inference because trajectories add procedural risk and redundant sampling. Complexity is managed by maximizing discrimination among plausible hypotheses per safe trajectory.

Its compression can mislead if sparse measurements are rendered as continuous whole-brain maps without uncertainty. Network diagrams and source estimates should preserve which claims are directly sampled, interpolated, or inferred.

Abstract Reasoning¶

A disciplined SEEG logic is:

Build competing anatomo-electro-clinical hypotheses from noninvasive evidence.
Identify observations that would distinguish them.
Design trajectories that sample decisive nodes and early propagation routes while respecting safety.
Verify actual contact locations after implantation.
Record representative habitual events and characterize interictal activity.
Compare onset timing, morphology, recruitment, and clinical sequence across contacts.
Use stimulation selectively and interpret concordance within its limitations.
Revise the network model and state unsampled alternatives.
Translate findings into treatment options under functional and procedural risk.

This is an experiment-design abstraction as much as a recording technique. A plan that samples only the favored hypothesis cannot strongly reject plausible alternatives. A good plan includes contacts whose outcomes would change the clinical decision.

Knowledge Transfer¶

The hypothesis-led sensor-placement pattern transfers to other sparse inverse problems: boreholes in geology, biopsies in oncology, probes in a reactor, or telemetry in a network. Existing evidence defines competing models; scarce invasive measurements are placed where they most discriminate; observations update the model.

Transfer stops at clinical semantics. Brain tissue, seizure dynamics, functional mapping, vascular avoidance, informed consent, and neurological risk cannot be replaced by generic sensor language. Likewise, a dense noninvasive sensor array does not become SEEG merely because it supports network inference.

Examples¶

Suspected insular onset. Semiology and scalp EEG are compatible with temporal or insular onset. Trajectories sample insula, operculum, mesial temporal structures, and plausible spread routes. Earliest sustained change and clinical sequence across verified contacts discriminate the hypotheses.

MRI-negative focal epilepsy. No structural lesion provides an obvious target. A convergent hypothesis from semiology, functional imaging, and scalp EEG guides three-dimensional sampling. The result may support a bounded intervention or reveal that no safe, focal strategy is justified.

Stimulation mapping. Selected contacts are stimulated under a protocol to identify motor, language, sensory, or experiential responses and to compare induced events with habitual seizures. The result constrains treatment near functional networks; stimulation alone is not the whole SEEG method.^[5]

Unsampled false reassurance. No sampled contact shows onset before widespread recruitment, but a plausible region was not implanted. The correct conclusion is unresolved localization under the sampling design, not proof of generalized onset.

Non-example. A depth electrode implanted solely for chronic stimulation, with no hypothesis-led seizure recording and anatomo-electro-clinical analysis, is not an SEEG evaluation.

Structural Tensions¶

Coverage versus procedural risk: more trajectories sample more hypotheses but increase exposure.
Hypothesis focus versus confirmation bias: targeted sampling is necessary and can exclude alternatives.
Spatial precision versus sparse coverage: contacts are local and accurately placed but leave most tissue unmeasured.
Natural seizures versus finite observation: captured events may not represent the full repertoire.
Localization versus network dynamics: a first contact does not by itself define all tissue necessary for seizure generation.
Diagnostic information versus therapeutic action: a clear network may still be unsafe or unsuitable to treat.
Clinical placement versus research reuse: rich data are not population-uniform samples.

Structural–Framed Character¶

SEEG has a structural measurement and inference core: coordinates, contacts, signals, timing, and observations can be documented. Framing enters through hypothesis selection, acceptable risk, definition of treatment success, and interpretation of clinical signs. The method is reproducible only when those decisions are made explicit.

Structural Core vs. Domain Accent¶

The portable core is hypothesis-directed sparse measurement for model discrimination. The domain accent is drug-resistant focal epilepsy, stereotactic intracerebral electrodes, seizure recording, stimulation, contact localization, epileptogenic-network concepts, and neurosurgical risk. Removing these yields generic invasive measurement, so the node is domain-specific.

Measurement is the conservative proposed parent: SEEG maps intracerebral electrophysiological attributes to time-indexed signals through located electrodes and procedures. It also instantiates Hypothesis Testing, Experimental Design, Sampling, and Evidence Integration. Somatotopy and other neural maps can be findings or targets, not parents.

One prospective strict edge to prime:measurement is queued. No live DAG mutation is authorized.

Relationships to Other Abstractions¶

Current abstraction Stereoelectroencephalography Domain-specific

Parents (1) — more general patterns this builds on

Stereoelectroencephalography is a kind of Measurement Prime

Measurement is the conservative proposed parent: SEEG maps intracerebral electrophysiological attributes to time-indexed signals through located electrodes and procedures.

Hierarchy path (1) — routes to 1 parentless root

Stereoelectroencephalography → Measurement

Neighborhood in Abstraction Space¶

Stereoelectroencephalography sits in a sparse region of the domain-specific corpus (99^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With¶

Scalp EEG: noninvasive surface recording.
Electrocorticography/subdural grids: surface intracranial coverage with different geometry.
Depth electrode: a component, not the whole method.
Stereotaxy: placement framework without the SEEG inference cycle.
Seizure-onset zone: an observed onset construct, not the method.
Epileptogenic zone/network: treatment-relevant inference produced in part by the method.
Deep-brain stimulation: therapeutic stimulation with a different objective.
SEEG-guided thermocoagulation: possible therapy using implanted contacts, not identical to diagnostic SEEG.

References¶

[1] Jean Bancaud and Jean Talairach, “Méthodologie de l'exploration SEEG et de l'intervention chirurgicale dans l'épilepsie,” Revue d'oto-neuro-ophtalmologie 45, no. 4 (1973): 315–328, PubMed 4603118. Primary methodological source. registry ↩

[2] Jérôme Isnard et al., “French Guidelines on Stereoelectroencephalography (SEEG),” Neurophysiologie Clinique 48, no. 1 (2018): 5–13, DOI 10.1016/j.neucli.2017.11.005. Specialist consensus covering indications, implantation, recording, stimulation, interpretation, and safety. registry ↩a ↩b

[3] Jorge G. Burneo et al., “Stereoelectroencephalography in Presurgical Epilepsy Evaluation,” Neurology (2022), DOI 10.1212/WNL.0000000000013088. Contemporary clinical review of three-dimensional sampling and epileptogenic-network inference. registry ↩

[4] Nitin Tandon et al., “Analysis of Morbidity and Outcomes Associated With Use of Subdural Grids vs Stereoelectroencephalography in Patients With Intractable Epilepsy,” JAMA Neurology 76, no. 6 (2019): 672–681, DOI 10.1001/jamaneurol.2019.0098. Comparative clinical outcomes source. registry ↩

[5] Patrick House et al., “Stimulation Mapping Using Stereoelectroencephalography: Current and Future Directions,” Frontiers in Neurology 11 (2020), DOI 10.3389/fneur.2020.00320. Review of stimulation mapping roles and limitations. registry ↩a ↩b