Evaluation of Epilepsy Seizure Onset Zone (SOZ) Implantation Quality through Spatial Perturbation Framework in Pre-Surgical Planning

Background

The success of epilepsy surgery relies heavily on the accuracy of pre-surgical planning, which typically involves identifying the epileptic zone (EZ) using clinical symptomatology, electroencephalography (EEG), and magnetic resonance imaging (MRI). However, in complex cases, these tools may not be sufficient to determine the EZ, necessitating the insertion of stereoelectroencephalography (SEEG) electrodes for higher spatial resolution. Bancaud and Talairach defined the EZ as the network or region where seizures originate and propagate earliest. While SEEG is extremely useful, it is limited by the number of electrodes that can be implanted. Typical SEEG explorations only cover about 5-10% of the brain volume.

In clinical practice, the seizure onset zone (SOZ) serves as an electrophysiological marker of the EZ for pre-surgical planning. However, even with meticulous identification of potential epileptogenic foci, only about half of the patients selected for surgery achieve seizure freedom post-operatively. A major reason for surgical failure is the incorrect sampling of the “true” SOZ, leading to misidentification of the EZ. Such inaccuracies can be attributed to multiple factors, including the ambiguous hypotheses originating from the initial pre-surgical evaluation and inaccuracies in the stereotactic implantation of SEEG electrodes.

Research Origin

This paper was co-authored by Kassem Jaber, Tamir Avigdor, Daniel Mansilla, Alyssa Ho, John Thomas, Chifaou Abdallah, Stephan Chabardes, Jeff Hall, Lorella Minotti, Philippe Kahane, Christophe Grova, Jean Gotman, and Birgit Frauscher. The authors are affiliated with institutions including the Montreal Neurological Institute and Hospital (McGill University), the Pratt School of Engineering at Duke University, the Multimodal Functional Imaging Laboratory (McGill University), the Asenjo Neurosurgery Research Institute, the Duke University Medical Center, and the Grenoble-Alpes University Brain Science Institute. This paper was published on June 4, 2024, and was accepted and published online in Nature Communications.

Detailed Process of the Original Study

Study Workflow

1. Patient Selection and Data Collection: The study involved two groups of patients: 50 patients from the Montreal Neurological Institute (17 Engel IA, 33 Engel IIB+) and 26 patients from the Grenoble Alpes University Hospital Center (18 Engel IA, 8 Engel IIB+). Engel IB-IIA patients were excluded as they were considered ambiguous in the assessment of EZ implantation. SEEG recordings from patients with drug-resistant focal epilepsy were sampled.

2. Extraction of Epileptic Biomarkers: The rate of detecting epileptic biomarkers at each SEEG electrode was used, specifically the IED-γ rate (γ activity between 30-100Hz occurring as interictal epileptiform discharges). IED-γ is considered a highly relevant biomarker for the EZ.

3. Construction of a Spatial System and Perturbation Analysis: A spatial system was constructed, coupling IED-γ rates with their distances to specific spatial reference points, defined as channels with maximal IED-γ rates. The system’s response to perturbations (such as the removal of the SOZ) was explored and measured.

Experimental Methods and Data Analysis

1. Virtual Removal of SOZ: The spatial system was perturbed by virtually removing the SOZ, and changes in coupling strength before and after removal were measured. This was achieved by calculating the Pearson correlation coefficient (ρ) before and after removal.

2. Random Removal of Non-SOZ Channels: Non-SOZ channels were randomly removed to test whether the impact of removing arbitrary parts of the configuration was different from removing the SOZ.

3. Data Comparison and Statistical Testing: Statistical methods were used to verify whether the perturbed system significantly affected the SOZ. Results for both patient cohorts (seizure-free and non-seizure-free) were compared.

4. Classification by Perturbation Strength: The perturbation strength (absolute log ratio of ρ before and after) was used as an important indicator for successfully classifying implantation quality and surgical outcomes.

5. Construction of a Spatial Perturbation Map: By spatially ranking the system’s response, a spatial perturbation map was constructed to qualitatively assess implantation plans. This map visually displayed the impact on each channel and compared the implantation plan to the SOZ location.

Main Research Findings

• In the Montreal Neurological Institute cohort, the removal of the SOZ significantly reduced spatial coupling (ρ significantly decreased) in seizure-free patients but had little impact on non-seizure-free patients. This result was consistently validated in the Grenoble Alpes University Hospital Center (similar statistical outcomes).
• Perturbation strength was significantly higher in seizure-free patients and was successfully used to classify implantation plan quality.
• A spatial perturbation map was constructed, showing a strong system response in seizure-free cases, while poor-quality implantation cases displayed irregular reductions in the map.

Conclusion and Significance

This study proposes a novel SEEG-based evaluation framework, effectively validating SOZ implantation in seizure-free patients. During pre-surgical evaluation, this new framework can help avoid unsuccessful surgeries due to poor implantation plans, reducing the risk of post-operative seizure recurrence. It provides an important tool for improving pre-surgical planning and implantation strategy.

Additionally, this study highlights the critical role of precise sampling in the success of epilepsy surgeries. By using computational models and the spatial distribution of biomarkers, it ensures that implantations accurately cover the seizure onset zones. The clinical application of this new framework can enhance the success rate of pre-surgical assessments and ultimately protect patients from unnecessary surgical complications and sequelae.

Highlights

• Quantitative Evaluation of Implantation Quality: Introduced a novel quantitative evaluation method based on perturbation strength, which is more effective than traditional methods that measure the Epileptogenic Zone (EZ) through event rates.
• Efficient Pre-Surgical Evaluation Tool: The spatial perturbation map provides an intuitive visual tool to help epilepsy experts assess the quality of implantations pre-surgically, preventing unsuccessful surgeries.
• Adaptation and Integration of Various Epileptic Biomarkers: The framework can adapt to various epileptic biomarkers, not limited to IED-γ, further validating its applicability in complex clinical settings.

This study not only introduces a new theoretical framework but also validates its effectiveness through actual clinical data, providing new methods and tools for pre-surgical evaluation in epilepsy surgery. The introduction of this framework can significantly improve surgical success rates and patients’ quality of life.