sEEG for Non-Epilepsy Mapping

Evidence status. Non-epilepsy sEEG is investigational. Published cohorts are small, indications and workflows vary, and no brief mapping admission can fully establish long-term efficacy of a permanent psychiatric or pain intervention.

1.The gap between imaging and intervention

Resting-state imaging and connectomics can nominate networks but are correlational. sEEG provides local field potentials, cross-region timing, evoked responses, and direct stimulation effects. It can test whether a candidate node participates in affect, motivation, compulsivity, pain, arousal, or cognition in an individual patient.

Its spatial sampling is sparse and hypothesis-driven. Unrecorded regions remain unknown, and a signal on one contact may reflect volume conduction or a passing network state rather than a causal hub.

2.Candidate indications

Recent programs have studied severe treatment-resistant depression, obsessive-compulsive disorder, chronic pain, and mixed transdiagnostic symptoms. Potential goals include identifying an individualized stimulation target, choosing among candidate circuits, defining a biomarker for responsive therapy, or ruling out a proposed permanent implant.

The denominators are small enough to name, and naming them is the honest way to read every claim that follows. A single-center workflow paper reports fourteen participants across three indications: six with major depressive disorder, six with chronic pain, and two with obsessive-compulsive disorder (Saal and colleagues). A hybrid series implanting permanent leads alongside sEEG electrodes reports three patients with treatment-resistant depression (Sheth and colleagues). The most-cited demonstration of the whole pathway, from multi-site mapping through to a chronic responsive device, is a single individual (Scangos and colleagues). The obsessive-compulsive literature at this scale is so far represented by published study designs rather than outcome reports (Seilheimer and colleagues). Read every statement on this page against those numbers.

Small pain studies also use individualized intracranial mapping and blinded stimulation testing. Their results remain specific to the studied syndromes and protocols; see the intracranial pain review for the evidence and its limitations. A positive acute response can nominate a target without establishing durable benefit from a permanent implant.

These are not extensions of routine diagnostic sEEG by analogy. Each indication requires disease expertise, a credible downstream therapy, and a risk-benefit case that survives the possibility of an inconclusive study. It also requires the correct regulatory instrument, and the instruments are not interchangeable. In the United States, implantation and experimental stimulation solely for non-epilepsy research will ordinarily constitute significant-risk device investigation. Significant-risk studies require FDA IDE approval as well as IRB approval before enrollment. Document the study-specific regulatory determination, sponsor and applicable IDE rather than assuming an IRB approval alone is sufficient. FDA distinguishes significant-risk, nonsignificant-risk, and exempt investigations; obtain a formal determination when the pathway is uncertain. Requirements outside the United States differ.

3.Hypothesis and coverage

Translate symptoms into testable domains: sadness, anhedonia, anxiety, urge, relief, cognitive control, pain intensity, pain unpleasantness, or behavioral flexibility. Select regions from convergent clinical, imaging, lesion, and stimulation evidence. Trajectories are bound by the same vascular and mechanical constraints as epilepsy sEEG, and by some that a temporal-predominant epilepsy montage rarely meets: subgenual and orbitofrontal targets bring the frontal sinus, the orbital roof, and the anterior cerebral territory into the planning problem far more often. Scientific desire never overrides safety.

In practice the candidate montages converge on a short list, and a page about mapping should say what gets mapped. Depression protocols sample subgenual cingulate, ventral capsule and ventral striatum, orbitofrontal cortex, amygdala, and hippocampus. This is the montage behind the field's index case, in which gamma power across bilateral amygdala, right orbitofrontal cortex, left subgenual cingulate, and right hippocampus separated high from low symptom states with a mean accuracy of 0.73, and right ventral capsule/ventral striatum was selected for chronic stimulation. Obsessive-compulsive protocols center on ventral capsule and ventral striatum territory, the same anatomy the established psychiatric indications already use. Pain protocols sample orbitofrontal and anterior cingulate cortex for the affective dimension alongside sensory thalamus. Name the montage in the protocol, and be able to say for each electrode which hypothesis it tests and what a null result on that contact would mean.

4.Tasks, symptoms, and stimulation

Use repeated, validated tasks and ecological symptom sampling rather than a single bedside impression. Randomize or blind stimulation trials where feasible, include sham and active controls, record adverse subjective effects, and test whether changes outlast the immediate stimulation period. Pair subjective ratings with behavior and physiology without treating one as ground truth.

Acute euphoria, laughter, anxiety, or autonomic change can prove engagement while predicting little about durable antidepressant or anti-obsessional benefit. Demand specificity: did stimulation change the intended domain, a general arousal state, expectancy, or task performance?

5.Signal analysis

Prespecify frequency bands, evoked responses, network metrics, artifact handling, and multiple-comparison control. Separate discovery from validation epochs. Avoid selecting the most compelling contact and metric after inspecting hundreds of combinations. If a biomarker will control a permanent device, test stability across sleep, medication, movement, and symptom states.

Say in advance how many trials each site will receive. A published statistics-driven framework for this workflow found that roughly ten stimulation trials per site were needed to estimate response variability reliably, at effect sizes of 1.1 or more in paired comparisons. This estimate is specific to the studied tasks, variability, and effect sizes. Plan repetitions, washout, carryover control, and multiplicity for the actual endpoint; ten trials is not a universal sample-size rule, and the number of sites alone does not determine power.

6.What counts as sufficient evidence?

7.Consent and therapeutic misconception

Consent must separate clinical care from research, clarify that the study may not identify a target, and describe hemorrhage, infection, neurologic deficit, seizure, psychiatric destabilization, explantation, privacy, and incidental findings. A participant should know whether a permanent implant is guaranteed, possible, or unavailable.

Quote risks with their denominator and ascertainment method. In a single-center study of 549 epilepsy sEEG implantations with systematic CT review, 19.1% had any radiographic hemorrhage, 2.2% symptomatic hemorrhage, and 0.6% permanent deficit or death (0.4% and 0.2%, respectively). Electrode number was associated with risk, but per-electrode events are not independent and should not be multiplied into a universal patient-level estimate. Use the center's current outcomes, trajectory plan, infection data, and the distinction between radiographic and symptomatic events in consent. These epilepsy data may not transfer directly to non-epilepsy research, where a participant may receive no therapeutic benefit.

Include an independent psychiatrist or disease specialist, neuropsychology, ethics support, and a plan for capacity changes, suicidality, mania, withdrawal of consent, and post-explant care.

8.A defensible program

A program should register protocols, define stopping rules, use independent adverse-event review, preserve negative results, and publish complete denominators. Permanent target selection should integrate the sEEG finding with anatomy, safety, disease biology, and the capabilities of the chronic device.