Temporally Dependent Iterative Expansion (TEDIE), a novel algorithm that accurately identifies seizure-causing epileptic zones in the brain

Unmet Need

Epilepsy is one of the most common neurological conditions, affecting approximately 50 million people worldwide as of 2023. Patients with epilepsy face a risk of premature death up to three times higher than the general population. While anti-seizure medications are available, 30% of patients have drug-resistant epilepsy. For these individuals, brain resection or ablation surgery is a proven treatment, but 40-60% remain symptomatic after surgery.

This outcome is largely due to challenges in accurately identifying the epileptic zones (EZ) in the brain that cause seizures. Diagnosing EZs typically begins with inpatient electroencephalography (EEG) monitoring to record brain activity during a seizure. Noninvasive EEG, which involves attaching electrodes to the scalp, may be insufficient, leading patients to undergo invasive stereo-EEG (sEEG), where electrodes are implanted into suspected brain regions via small holes drilled into the skull for improved signal resolution.

The placement of sEEG electrodes often relies on the epileptologist’s intuition, as recording sensitivity varies across individual patients. Additionally, epileptologists must manually inspect EEG and sEEG recordings to identify the EZ. This process is time-intensive, prone to error, and limited to observing brain activity at discrete timepoints rather than tracking the full seizure path. Signals originating from the true EZ may be imperceptible, while zones with larger signals at specific timepoints may represent secondary aggregations rather than the seizure source. A tool that reconstructs the full seizure path over time is critically needed to improve the accuracy of EZ identification.

Technology

Duke researchers have developed Temporally Dependent Iterative Expansion (TEDIE), an innovative tool designed to reconstruct the full path of a seizure as it spreads through the brain. TEDIE helps epileptologists more accurately identify the EZ, including the seizure source location and its size, enabling better evaluation of surgical candidacy and selection of surgical targets.

TEDIE operates by combining patient-specific brain imaging with sEEG recordings. Its novel algorithm dynamically models changes in neural activity during a seizure, iteratively optimizing the location and magnitude of activity over time. For each timepoint, TEDIE generates a seizure source likelihood map tailored to the patient’s geometry, which, when combined, creates a dynamic “movie” of seizure propagation. TEDIE also includes an automated sEEG planning feature that calculates and visualizes the sensitivity of sEEG electrodes on brain tissue, and optimizes electrode trajectories to minimize invasive implants while maintaining recording accuracy.

In clinical applications with data from 46 epilepsy patients at Duke University Medical Center and the University of Pennsylvania, TEDIE demonstrated:

• Accurate identification of EZs in patients who became seizure-free post-surgery.
• The ability to predict whether a patient would benefit from surgery and the type of surgery required.
• Identification of new surgical targets for patients previously deemed non-surgical candidates.

Compared to existing algorithms like sLORETA and IRES, TEDIE produced more accurate, focal, and interpretable seizure reconstructions. For sEEG planning, TEDIE optimized electrode placement to map more brain cortex with the same number of electrodes or the same amount of cortex with fewer electrodes than manual configurations.

Other Applications

• Extending TEDIE’s capabilities to conventional EEG could reduce the need for invasive sEEG and make advanced seizure mapping more accessible, especially in non-specialist epilepsy centers.
• TEDIE’s brain activity mapping technology could also support basic neuroscience research.

Advantages

• Accurately reconstructs seizures for precise EZ identification, outperforming existing algorithms.
• First-in-class automated sEEG planning tool that optimizes electrode placement and improves localization accuracy.
• Standardized, quantitative, and computational tool for clinician decision-making.
• Enhances patient safety and improves surgical outcomes through better sEEG planning and EZ localization.
• Delivers personalized care based on patient-specific brain geometry.