Abstracts

Rationale: Identifying the seizure onset zone (SOZ) accurately is of utmost importance to guide surgery or intra-cranial grid placement. The availability of such method can be useful and clinically relevant in diagnosing and managing medically intractable epilepsy. The goal of this work is to develop an objective and computationally efficient technique to image dynamic spatio-temporal activity of the brain from scalp EEG. We have tested this method in computer simulations as well as imaging ictal activity of focal epilepsy patients who have undergone surgery and became seizure free.  Methods: We have performed a Monte-Carlo simulation, placing extended sources on a real geometry cortex randomly, and then solved the forward problem to simulate EEG recordings (different signal to noise ratio were considered such as 5, 10 and 20 dB). Solving the inverse problem using FAST-IRES we estimated the location, extent and the time-course of activity for all these sources and compared our estimates to simulated sources. Additionally we applied this method to the ictal recordings of one temporal lobe epilepsy patient who underwent surgery and became seizure-free afterwards (1 year) and compared our results to clinical findings such as surgical resection and SOZ determined from intra-cranial studies, for validation. Results: The estimated results by FAST-IRES match well with the location, extent and time-course activity of simulated sources. Our initial results suggest a localization error of about 5 mm (distance between the simulated and estimated source). The estimated time-course of activity between the estimated and simulated sources is over 95%. This is an important feature of this work, as estimating underlying dynamics and intermodal connectivity of seizure networks can only be achieved if a precise estimate of the activity is provided. Furthermore our initial explorations with ictal data suggest that the SOZ can be estimated fairly accurately with FAST-IRES even from surface measurements, i.e. EEG. Our results indicate that applying FAT-IRES to ictal recordings from the scalp EEG of the patient prior to surgery to estimate the source of ictal activity in the brain matches well with the SOZ determined from invasive studies and falls within the surgical resection volume, obtained from post-operational MRI images (Fig. 1).  Conclusions: We have proposed a new spatio-temporal inverse algorithm to trace back dynamic brain electrical activity from non-invasive surface measurements such as EEG and MEG. The capability of this algorithm to accurately localize, objectively determine the extent and efficiently extract the activity time-course of underlying brain sources is of crucial interest in studying ictal sources and signals. This algorithm has the potential to be used as an investigational tool to non-invasively study ictal networks and also make an impact on patient population by providing a precise estimate of the SOZ, which can potentially limit the need to use invasive intra-cranial studies; it can also help guide the placement of the electrodes (in case of invasive studies) to more optimally place the limited electrodes in regions which are more probable to be epileptogenic.  Funding: This work was supported in part by NIH R01 NS096761 and EB021027. Abbas Sohrabpour was supported in part by an Interdisciplinary Doctoral Fellowship and a Doctoral Dissertation Fellowship from the University of Minnesota.