TL;DR
This review explores the current state and future potential of magnetoencephalography (MEG) for localizing epileptic foci in patients with drug-resistant epilepsy. It highlights MEG's strengths in spatial and temporal resolution compared to EEG and discusses its correlation with the gold standard intracranial EEG (iEEG), suggesting a growing role for MEG in guiding surgical interventions.
What Problem Does This Paper Address?
Epilepsy affects a significant portion of the population, and for about a third, medication is insufficient, making them candidates for neurosurgery. A critical step before surgery is accurately identifying the epileptic foci – the brain regions responsible for seizures. While various non-invasive imaging techniques exist (EEG, MRI, PET, SPECT, MEG), they sometimes yield inconsistent results. When this happens, invasive intracranial EEG (iEEG) is the current "gold standard" for pinpointing these foci. However, iEEG is invasive and only monitors a portion of the brain. This paper addresses the challenge of improving the accuracy and reliability of non-invasive methods, specifically MEG, to better guide presurgical localization and potentially reduce the need for or optimize the use of iEEG. It also explores how insights from cognitive neuroscience, particularly regarding brain network dynamics, can enhance MEG's application in this clinical context.
Methods
The paper discusses two primary approaches for source localization using MEG data:
- Single Equivalent Current Dipoles (sECD): This method models the electrical activity of neuronal assemblies as a single dipole. It involves visually extracting the tightness and orientation of dipole clusters from MEG signals. While it can be effective for localized sources, its subjectivity and potential to misrepresent spatially extended generators are noted limitations.
- Distributed Source Modeling (DSM): This approach aims to explain observed magnetic fields by modeling the distribution of source intensities across the brain, rather than a single point. DSM has gained prominence, especially in cognitive MEG studies, due to its ability to capture simultaneous activation across multiple brain areas. The paper highlights the nonlinear Maximum Entropy on the Mean (MEM) algorithm as a promising DSM technique for predicting cortical regions with interictal epileptic discharges (IEDs).
The review also emphasizes the importance of correlating MEG findings with iEEG data (electrocorticography - ECoG, and stereo-EEG - sEEG) to validate MEG's effectiveness.
Datasets
Not discussed in the paper.
Key Insights and Ideas
- MEG's Superior Resolution: MEG offers significantly better spatial and temporal resolution than scalp EEG for detecting epileptic spikes, potentially identifying generators within a 4 cm² area compared to EEG's 10 cm².
- MEG as a Complement to iEEG: While iEEG remains the gold standard, MEG shows strong promise in localizing epileptic zones and has demonstrated good concordance with iEEG results. It can provide a broader view of brain activity, aiding in the interpretation of iEEG data.
- The Network Perspective: Viewing epilepsy as a network disease, rather than solely a focal issue, opens new avenues for MEG application. Understanding functional connectivity within and between brain areas is crucial for predicting surgical outcomes.
- Resting-State MEG Potential: Preliminary studies suggest that resting-state MEG analysis, without the need to capture interictal spikes, may be sufficient for accurate epileptogenic focus localization, offering a more accessible approach.
- Cognitive Neuroscience Synergy: Advances in understanding brain networks from cognitive studies can inform and improve MEG analysis for epilepsy, particularly in areas like functional connectivity and network dynamics.
Key Results
- MEG can detect 50–90% of spikes from superficial cortex and 50–63% from mesial temporal structures, which is generally considered acceptable for presurgical assessment.
- Recent studies show MEG can detect epileptic activity from deep brain structures like the insula.
- Studies correlating MEG with iEEG (ECoG and sEEG) show that when iEEG completely samples areas identified by MEG, seizure-free outcomes after surgery are significantly higher.
- DSM algorithms, particularly nonlinear MEM, show promise in better predicting cortical regions with IEDs compared to other distributed models.
Limitations and Future Work
- Immature Source Reconstruction Algorithms: The paper notes that MEG source reconstruction algorithms are still developing, which is a significant limitation for clinical application.
- MEG vs. iEEG: MEG cannot yet fully replace iEEG as an independent modality for presurgical assessment, especially when non-invasive outcomes are inconsistent.
- Need for Large Databases: To fully leverage functional connectivity studies for predicting individual surgical outcomes, a large database of normal resting-state MEG mapping is required.
- Future Directions: The authors suggest that future research should focus on developing more advanced MEG analysis technologies, creating objective biomarkers for foci localization, and integrating resting-state fMRI and MEG studies to investigate epileptic networks. The establishment of normative MEG databases, like those contributing to the Human Connectome Project, is also highlighted as a crucial step.
Summary updated