Imaging epileptogenic brain using high density EEG source imaging and MRI

Noninvasive imaging of epileptogenic zone is of great clinical importance to improve the success rate of surgical resective treatment of intractable epilepsy. A number of efforts have been made to improve our ability to localize and image epileptogenic zones using structural and functional neuroimaging modalities including metabolic/hemodynamic and electrophysiological signals (He et al., 2013). Despite the significant progress made in the past decade, current clinical practice is still based upon invasive intracranial EEG (iEEG) recordings, to make definite surgical planning decisions; particularly in patients without MRI visible lesions. Most studies using noninvasive modalities have involved small number of patients due to the challenges involved in comparing estimation results obtained by such modalities with invasive recordings and surgical resection outcomes, in patient populations. In the present issue, Lascano et al. (2016) report encouraging results in 190 operated patients using electrical source imaging (ESI), magnetic resonance imaging (MRI), Positron emission tomography (PET) and single-photon emission computed tomography (SPECT) (Lascano et al., 2016). It represents an important advance in furthering our efforts to establish noninvasive neuroimaging for aiding presurgical planning in partial epilepsy. Several modalities have been used to localize and image epileptogenic zones to date, including MRI, functional MRI (fMRI), EEG/MEG based ESI, PET, and SPECT. Different imaging techniques and modalities complement each other, with their advantages and limitations. fMRI, SPECT and PET record neuronal activity indirectly, as the blood flow, blood perfusion or glucose metabolism is measured rather than directly measuring the electromagnetic fields generated by neuronal activity. These imaging modalities provide high spatial resolution or contrast but have low temporal resolution (la Fougere et al., 2009; Velez-Ruiz and Klein, 2012). On the other hand, EEG/MEG directly measures instantaneous electromagnetic fields generated by synchronized neuronal excitation, resulting in high temporal resolution, but with a low spatial resolution due to the volume conduction effect. Structural MRI has been shown to be useful in localizing lesions which are highly correlated with epileptogenic substrates (Engel, 2013), although many partial epilepsy patients do not exhibit MRI visible lesions. EEG based ESI has been pursued by a number of groups for diagnosis and pre-surgical planning of epilepsy patients. EEG is a relatively inexpensive, portable and accessible device which is available in most clinical settings. There have been many studies to show the usefulness and effectiveness of ESI in determining the epileptogenic foci (Brodbeck et al., 2011; Holmes et al., 2010; Lantz et al., 2003; Michel et al., 2004; Michel and He, 2011; He and Ding, 2013), including source localization during inter-ictal spikes (Kaiboriboon et al., 2012; Noe et al., 2013; Plummer et al., 2010; Sohrabpour et al., 2015; Wang et al., 2011) and seizures (Assaf and Ebersole, 1997; Boon et al., 2002; Ding et al., 2007; Holmes et al., 2010; Koessler et al., 2010; Lu et al., 2012a; Yang et al., 2011). In ESI, of importance is to model individual patient’s head volume conductor, develop inverse imaging algorithms for accurate reconstruction of epileptic activity, and increase the number of scalp electrodes to avoid possible spatial aliasing. Since introduction of the boundary element method (BEM) to model the geometry of human head more realistically for EEG source localization (He et al., 1987; Hämäläinen and Sarvas, 1989), the BEM modeling is fairly well established based on structural MRI scans of patients. Innovations in source localization and imaging algorithms have led to a rich body of imaging algorithms including weighted minimum norm estimates (LORETA (Pascual-Marqui et al., 1994), LAURA (Grave de Peralta et al., 2001)) as well as subspace and spatiotemporal dipole fitting methods (MUSIC (Mosher et al., 1992), FINE (Xu et al., 2004)). Studies have also explored the desirable number of scalp electrodes necessary to obtain accurate, robust and reliable ESI results. Studies have shown the merits of increased number of scalp electrodes beyond the clinical convention of 32 electrodes (Lantz et al., 2003; Holmes et al., 2010; Wang et al., 2011; Yang et al., 2011). It has also been shown recently that while increasing the number of electrodes will improve results, there is a plateauing effect in the localization accuracy, thus increasing the electrodes beyond a certain point will not be as beneficial (Lu et al., 2012b; Sohrabpour et al., 2015). Brodbeck et al. showed in a study containing 152 patients that the sensitivity and specificity of high density EEG (40 patients with 128 electrodes and 14 patients with 256 electrodes) source imaging results are higher than MRI, SPECT, PET and low-density EEG (19–29 electrodes) results, suggesting the accuracy and usefulness of ESI when enough number of electrodes are used (Brodbeck et al., 2011). SPECT and PET are functional imaging modalities where the neuronal metabolism or blood flow (both of which are modulated