Network Dynamics and Interictal Biomarkers for Seizure Localization

126 concepts91 papers1 proposal

For patients with drug-resistant epilepsy, neurosurgery offers a powerful chance for a cure. However, the success of the procedure depends entirely on a high-stakes mapping problem: precisely identifying and resecting the brain tissue responsible for generating seizures.

Historically, the clinical standard has relied on capturing spontaneous seizures using prolonged, invasive intracranial electroencephalography (iEEG) monitoring. As Erin C. Conrad et al. point out in Interictal intracranial EEG asymmetry lateralizes temporal lobe epilepsy, this requirement forces patients to endure weeks of hospitalization and distress while waiting for a seizure to occur. Furthermore, visible structural abnormalities on an MRI do not always map perfectly to the functional source of a seizure. To overcome these limitations, a major paradigm shift is underway: leveraging computational models, network science, and advanced imaging to identify the seizure source using "quiet" brain activity recorded between seizures.

Decoding Interictal Micro-Dynamics

Rather than waiting for a full-scale seizure to disrupt the brain, researchers are looking at subtle, high-frequency anomalies that occur during normal interictal periods. High-Frequency Oscillations (HFOs)—brief, intense bursts of electrical activity—have emerged as a primary candidate.

However, simply detecting HFOs is insufficient because normal physiological processes also produce high-frequency waves. To differentiate pathological tissue from healthy tissue, we must look at how these signals behave mathematically across different channels. Stephen V. Gliske et al. demonstrated in The intrinsic value of HFO features as a biomarker of epileptic activity that while HFO features lie on low-dimensional, approximately linear manifolds, these structures are highly electrode-specific. This means automated detection algorithms cannot treat the brain uniformly; they must account for channel-specific baseline signatures to isolate true pathology.

[Slow Homeostatic Wave]  ---> Modulates Phase of ---> [Pathological HFOs]
                                                      (Quantified via Modulation Index)

A deeper layer of intuition comes from looking at how these fast oscillations interact with slower brain rhythms. In an epileptic zone, the timing of pathological HFOs is often tightly coupled to the phase of underlying slow waves. Mojtaba Bandarabadi et al. evaluated this relationship in Assessing Epileptogenicity Using Phase-Locked High Frequency Oscillations: A Systematic Comparison of Methods, demonstrating that the Modulation Index (MI)—a metric quantifying this cross-frequency phase-amplitude coupling—serves as a highly reliable predictor of surgical success. The strength of this coupling reveals how deeply the seizure-generating tissue has hijacked the local homeostatic rhythms of the brain.

The Brain as a Dynamical Network

Seizures are rarely generated by an isolated, independent group of neurons; instead, they emerge from pathological interactions across a broader network. By modeling the brain as a directed graph, researchers can identify the seizure source based on how information flows through the system during periods of rest.

An elegant manifestation of this approach is the Source-Sink Metric (SSM) framework developed by Adam Li et al. in Source-sink connectivity: A novel interictal EEG marker for seizure localization. The underlying intuition relies on a balance of excitation and inhibition: during interictal periods, the hyper-excitable tissue responsible for seizures is actively suppressed by surrounding healthy brain structures. As a result, this pathological zone acts as a net "sink," drawing in inhibitory signals from surrounding "sources." By mapping these directional network dynamics from interictal iEEG data, the authors achieved high accuracy in predicting surgical outcomes, proving that the network's baseline topology betrays the hidden source of the disease.

[Healthy Source] --Inhibitory Flow--> [Pathological Sink] <--Inhibitory Flow-- [Healthy Source]
                                      (Seizure Onset Zone)

Taking this network perspective to its logical conclusion, Marc Goodfellow et al. introduced a computational framework in Estimation of brain network ictogenicity predicts outcome from epilepsy surgery to simulate "virtual surgeries." By constructing patient-specific mathematical models of neural mass dynamics, they could systematically remove nodes from the virtual network and calculate the resulting drop in "ictogenicity" (the network's propensity to generate seizures). This shifts the clinical paradigm from a retrospective guessing game to a prospective, quantitative engineering discipline.

Enhancing Contrast via Multimodal Imaging

When electrical recordings are ambiguous or non-invasive precision is required, functional and metabolic imaging can expose the seizure onset zone by highlighting localized physiological stress.

For instance, resting-state functional MRI (rs-fMRI) captures low-frequency blood oxygenation fluctuations to map functional connectivity. Payal Kamboj et al. leveraged this in Automated seizure onset zone locator from resting-state functional MRI in drug-resistant epilepsy, developing an unsupervised algorithm called EPIK. By codifying expert neurological rules into an automated pipeline, EPIK isolates anomalous localized blood-oxygen-level-dependent (BOLD) signals, providing a highly reliable, non-invasive roadmap for pediatric populations.

Standard PET (40 min):  [ Low Contrast: Focus blends with healthy tissue ]
Delayed PET (2-3 hours): [ High Contrast: Focus clearly demarcated due to differential clearance ]

Metabolic imaging via Positron Emission Tomography (PET) offers another window into the seizure zone, though standard protocols often suffer from low signal-to-noise ratios. Yang Hong et al. addressed this in Delayed 18F-FDG PET imaging provides better metabolic asymmetry in potential epileptogenic zone in temporal lobe epilepsy by extending the window between tracer injection and scanning to 2–3 hours. This delay allows the differential rates of metabolic clearance and accumulation between pathological and healthy tissue to widen, significantly boosting metabolic asymmetry and clearly demarcating the focus.

Where metabolism fails to show a clear boundary, cellular inflammation can fill the gap. Kuriko Kagitani‐Shimono et al. demonstrated in Clinical evaluation of neuroinflammation in child-onset focal epilepsy: a translocator protein PET study that targeting the translocator protein (TSPO)—which is overexpressed by activated microglia during neuroinflammation—can successfully localize the seizure zone even in patients whose structural MRIs appear completely normal.

Defining the True Surgical Boundaries

A recurring challenge in epilepsy surgery is that the primary source of electrical hyper-excitability frequently extends beyond the visible boundaries of structural lesions, such as vascular malformations or cortical tumors.

In a retrospective study of cerebral cavernous malformations, Qihang Zou et al. (Predictors of postoperative epileptic seizures after microsurgical treatment in supratentorial single cerebral cavernous malformations: a retrospective study) showed that simply removing the visible lesion is often insufficient for long-term seizure freedom. The surrounding tissue, frequently altered by iron staining and glial scarring, harbors the true electrical footprints of the seizure network. Utilizing real-time intraoperative electrocorticography (ECoG) allows surgeons to map and resect these surrounding irritative zones.

When pre-operative diagnostic tools reveal that these networks span multiple lobes, a highly localized resection will inevitably result in under-treatment. As detailed by Haruhiko Kishima et al. in Which is the Most Appropriate Disconnection Surgery for Refractory Epilepsy in Childhood?, recognizing when a localized zone is embedded within a broader multi-lobar network allows clinicians to opt for wider surgical disconnections early on, preventing secondary operations and significantly improving long-term developmental outcomes.

Key Foundational Work

EEG Source Imaging: Correlating Source Locations and Extents With Electrocorticography and Surgical Resections in Epilepsy Patients (2007) — contributed 3 concept elements to this direction.
Multimodality Data Integration in Epilepsy (2007) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_epileptogenic_zone_localization_c64869b1,
  title = {epileptogenic zone localization},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/epileptogenic-zone-localization/rdr_c64869b1f2a3413cb047d86942b38df4},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-21},
}

This exact version (version 1, frozen text):

@misc{lacuna_epileptogenic_zone_localization_c64869b1_v1,
  title = {epileptogenic zone localization},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/epileptogenic-zone-localization/rdr_c64869b1f2a3413cb047d86942b38df4/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-21},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake