Epilepsy, a condition manifested by aberrant brain networks from a combination of disparate etiologies, is in continuing need for better biomarkers for diagnosis and treatment. While seizures identified using electrophysiological tools is most recognized in detecting seizure onset and spread, their rare occurrence limits its ability to identify epileptic foci. On the other hand, events such as interictal spikes are a potential and commonly suggested alternative. Although these spikes are often associated with seizure onset regions, their precise roles are poorly understood. Additionally, the convoluted structural abnormalities alter the way the brain communicates amongst its different regions. What is needed is an integrated approach to understanding the functional networks of the diseased brain that accounts for its underlying structures. Using a combination of long term electrocorticography recordings and magnetic resonance imaging, we developed a set of algorithms and mapped the spatial-temporal network of interictal spike propagations and related it to brain topography and cortical lesions loci in 43 pathologic brains. We found that our method of isolating the time-epochs when time-locked interictal spikes were occurring led to the best identification of seizure onset regions, are related to brain topography, and are related to lesion loci. From a series of work, we have found the following observations. Interictal spike network is highly reproducible at different time points and frequency bands. While spikes are mostly closer to the epileptic foci, they are also sparsely distributed across the neocortex amongst which only a subset of spikes propagate to different cortical regions. Among these propagating spike regions, the spike onset regions often do not correspond to high spiking regions. While spike onset regions partially co-localize to seizure onset regions, they are also found on other cortical regions not identified as seizure onset zones. The specificity of these spike network is significantly higher compared to total brain network which is evaluated using the entire dataset irrespective of spiking.

The spike propagation is not uniform across all cortical region and is dependent on the geodesic distance between electrodes as well as the sulcal patterns. While the total brain network connection density decreases with geodesic distance, the rate of decrease in propagation density for spike network is more drastic. The large sulci such as central sulcus acts as a rigid barrier of propagation and restricts 80% of the propagations. While brain topography is a natural barrier of propagations, the pathologic lesions also play an important role. The spike networks are often found reverberating in the perilesional regions and colocalize with seizure onset. Such evaluations delineate the relationship between subtle electrical signatures and the underlying pathology, suggesting further attention to these biomarkers to enable better surgical precision, treatment plans, and drug development.