EEG Source Imaging: Correlating Source Locations and Extents With Electrocorticography and Surgical Resections in Epilepsy Patients

Lei Ding, Christopher Wilke, Bobby Xu, Xiaoliang Xu, Wim van Drongelen, Michael Kohrman, Bin HeJournal of Clinical Neurophysiology 2007 · DOI

TL;DR

This paper introduces a novel noninvasive method using EEG to pinpoint the location and size of epileptogenic zones in epilepsy patients. By combining a subspace source localization technique (FINE) with a local thresholding strategy, the researchers can estimate the extent of abnormal brain activity, which is crucial for surgical planning and has shown promising correlation with invasive ECoG data and successful surgical outcomes.

What Problem Does This Paper Address?

For patients with intractable epilepsy, precisely identifying the brain regions responsible for seizures (epileptogenic zones) is critical for effective surgical treatment. Currently, the "gold standard" for this is electrocorticography (ECoG), which involves surgically implanting electrodes directly on the brain's surface. While accurate, ECoG is invasive and carries risks. There's a significant need for noninvasive techniques that can provide comparable information, specifically regarding both the location and the spatial extent of these problematic zones. Existing noninvasive methods, particularly those based on equivalent current dipole (ECD) models, often struggle to accurately estimate the extent of the epileptic activity, which is vital for determining the precise area to be resected.

Methods

The core of this study lies in the application and evaluation of a subspace source localization method called FINE (First Principle Vectors). Unlike traditional dipole models, FINE scans the entire possible source space and calculates a "subspace correlation" (SC) metric for each point. This SC value indicates how well a potential source at that point explains the measured scalp EEG data.

The key innovation here is the local thresholding strategy applied to the FINE results. Instead of using a single, global threshold for the SC values, the authors propose a method that sets a threshold for each identified source based on its own local minimum SC value. This approach aims to better account for the varying sensitivities of scalp EEG to different source configurations (location, orientation, depth, and strength). By applying this local thresholding, the researchers can then estimate the spatial extent of the identified current sources. The performance of this combined FINE and local thresholding approach was evaluated using interictal spikes (IISs) from three pediatric epilepsy patients.

Datasets

The study utilized data from three pediatric patients with medically intractable partial epilepsy. For each patient, the following data were collected:

  • Scalp EEG: Recorded using 24 electrodes at a 400 Hz sampling rate.
  • Invasive ECoG: Recorded from subdural electrode arrays placed on the cortex during postoperative monitoring.
  • Structural MRI: Used for creating realistic head models.
  • CT scans: Used for precise localization of implanted subdural electrodes.

The study did not introduce a new benchmark dataset but rather used clinical data from these three patients for validation.

Key Insights and Ideas

  • Beyond Location: Extent Matters: The study strongly suggests that estimating the extent of epileptogenic zones, not just their location, is crucial and achievable with noninvasive EEG.
  • FINE as a Promising Tool: The FINE subspace source localization method, when combined with local thresholding, shows promise in identifying both the location and extent of abnormal brain activity.
  • Local Thresholding's Advantage: A local thresholding strategy appears more effective than global thresholding for estimating source extents, as it can adapt to the varying signal characteristics of different sources.
  • EEG-ECoG Correlation: The noninvasively derived EEG source locations and extents showed good correlation with invasive ECoG recordings, validating the method's potential.
  • Clinical Relevance: The identified epileptogenic zones from EEG source imaging were well-aligned with the areas surgically resected, and satisfactory surgical outcomes (seizure reduction or freedom) were observed in all three patients.

Key Results

  • The thresholded subspace correlation from FINE scanning was identified as a favorable marker for the extents of current sources associated with epileptic activities.
  • In Patient #1, the relative extent ratios of multiple activated areas identified by EEG were similar to those from ECoG.
  • In Patient #2, the EEG source with a larger extent corresponded to a larger subdural potential extent in ECoG, and vice versa.
  • In Patient #3, the EEG source location and extent were consistent with the ECoG findings, and the subsequent medial temporal lobe resection resulted in the patient becoming seizure-free.
  • The study successfully identified multiple current sources in patients, even when they were closely located, demonstrating FINE's capability.

Limitations and Future Work

  • Limited Sample Size: The study was conducted on only three pediatric patients. The authors acknowledge that further investigations with a larger cohort are desirable.
  • Simultaneous EEG-ECoG: While the study showed consistency between preoperative EEG and postoperative ECoG, performing validation with simultaneously recorded EEG and ECoG signals would provide a more direct comparison. The authors note that the presence of low-conductive ECoG pads can introduce inhomogeneities that need to be handled.
  • Optimizing Thresholding: Further research is needed to establish an optimal and robust thresholding value for the subspace correlation metric under various conditions. The authors suggest exploring techniques like Receiver Operating Characteristic (ROC) curves or confidence volumes for this purpose.
  • Deep Sources: The paper mentions that FINE can estimate depth information, which might not be obtainable from ECoG, suggesting potential for localizing deep epileptogenic zones. Future work could explore this further.

Summary updated

Abstract

It is desirable to estimate epileptogenic zones with both location and extent information from noninvasive EEG. In the present study, the authors use a subspace source localization method (FINE), combined with a local thresholding technique, to achieve such tasks. The performance of this method was evaluated in interictal spikes from three pediatric patients with medically intractable partial epilepsy. The thresholded subspace correlation, which is obtained from FINE scanning, is a favorable marker, which implies the extents of current sources associated with epileptic activities. The findings were validated by comparing the results with invasive electrocorticographic (ECoG) recordings of interictal spike activity. The surgical resections in these three patients correlated well with the epileptogenic zones identified from both EEG sources and ECoG potential distributions. The value of the proposed noninvasive technique for estimating epileptiform activity was supported by satisfactory surgery outcomes.
BibTeX
@article{ding2007eeg,
  author = {Lei Ding and Christopher Wilke and Bobby Xu and Xiaoliang Xu and Wim van Drongelen and Michael Kohrman and Bin He},
  title = {{EEG Source Imaging: Correlating Source Locations and Extents With Electrocorticography and Surgical Resections in Epilepsy Patients}},
  journal = {Journal of Clinical Neurophysiology},
  year = {2007},
  doi = {10.1097/wnp.0b013e318038fd52},
}
Figure