Abstracts

Rationale: We have developed an electromagnetic source imaging algorithm which enables us to image underlying brain electrical activity from non-invasive scalp EEG (or MEG) measurements. This method (IRES) works on the principles of sparse signal processing and iterative reweighting and can identify both the location and extent of underlying brain activity objectively, i.e. without applying a subjective threshold to the solution [1]. Simulation study results have been presented in [1]. We have used this technique to image the inter-ictal spikes (IIS) of patients suffering from medically intractable focal epilepsy, to investigate if IRES can determine the nodes of the epilepsy network. This can be helpful in clinical management of focal epilepsy, as IRES can provide an estimate of the epileptic activity extent aiding surgery planning or intra-cranial grid placement in the pre-surgical planning phase. Methods: We studied five focal epilepsy patients who suffered from medically intractable seizures and underwent surgical resection to become seizure-free (temporal and extra-temporal lobe cases). The pre-operational EEG recordings of these patients (76 channels of EEG) were screened for IIS to select spikes that were consistent in shape and had similar scalp maps. The IIS were averaged and input into IRES to find the underlying sources. The obtained solution was compared to clinical findings such as resection surface (obtained from post-operational MRI) and intra-cranial EEG electrodes marked as seizure onset zone (SOZ) by the physician (obtained from CT images). Results: The estimated results by IRES show a high value of overlap with resection surface and SOZ electrodes. Furthermore, the normalized overlap values (defined as overlap area of solution and resection divided by solution area or overlap area divided by resection surface area) both show a high value of about 0.8 in the cohort of studied patients. The normalized overlap ratios for the SOZ shows a high value (~ 0.9 on average) for the overlap area normalized by SOZ area but a smaller value (~ 0.5 on average) for the overlap area normalized by solution. This means that while the solution covers SOZ electrodes pretty well, it is a bit extended beyond. This could be due to the fact that IIS arise from the irritative zone which is typically larger than the SOZ. Conclusions: We have tested the feasibility of applying IRES to estimate underlying epileptic networks and observed that IRES solution is comparable to clinical findings such as SOZ or resection area, in focal epilepsy patients who became seizure-free after surgical resective operation. The unique feature of IRES, which is to estimate extended sources without applying subjective thresholds, can help improve patients' quality of life by providing a more accurate mapping of epileptogenic brain noninvasively. The present promising results suggest the approach warrants further investigation. [1] Sohrabpour, A., Lu, Y., Worrell, G., & He, B. (2016). Imaging brain source extent from EEG/MEG by means of an iteratively reweighted edge sparsity minimization (IRES) strategy. NeuroImage. Funding: This work was supported in part by NIH NS096761-01 and EB021027-01A1.