Conclusions

Non-invasive functional neuroimaging techniques have been actively investigated in human subjects since the discovery of EEG in the late 1920s (Berger, 1929). Tremendous efforts have been made in order to improve the temporal and spatial resolutions of non-invasive approaches. Along this line, EEG and MEG source imaging (Baillet et al., 2001; Hamalainen and Sarvas, 1989; Michel et al., 2004), fMRI (Bandettini et al., 1992; Kwong et al., 1992; Ogawa et al., 1992), and multimodal EEG- fMRI imaging (Dale and Halgren, 2001; He and Liu, 2008; Liu et al., 1998) have improved our capability of noninvasive imaging to a resolution with a millimeter scale in space and a millisecond scale in time. While neuroimaging techniques have been extensively used in neuroscience research, it is, to a lesser extent, applied to clinical diagnoses and treatments. An important application of functional neuroimaging techniques in the diseased brain is the pre-surgical evaluation of epilepsy. Collectively, information from multiple independent modalities, such as semiology, neuroimaging, and neurological tests, provides clues to delineate a surgical plan. The current pre-surgical imaging procedure has noninvasive components including the imaging of anatomy (MRI), imaging of hemodynamic and metabolic responses (SPECT, fMRI, PET), and scalp mapping of electrophysiological signals (Engel et al., 2008; Rosenow and Luders, 2001). A potential trend in the clinical investigations is to include the localization of interictal activity from EEG or MEG as an imaging tool. However, with the many

component that can image electrophysiological signals during seizures. The development of such a component can be extraordinarily important, because the seizure is the most clinically important event in epilepsy patients, the cure of which is the ultimate goal of any epilepsy treatment.

In the present dissertation research, we developed a non-invasive seizure imaging protocol, which combines high-resolution EEG recording and spatiotemporal source imaging. By testing the new protocol in epilepsy patients, we demonstrated the feasibility of the high resolution EEG recording protocol for clinical long-term monitoring. The extended recording time is crucial for capturing spontaneous and infrequent seizure events. We then evaluated the capability of the spatiotemporal EEG seizure imaging approach by information obtained through the most invasive procedures: resective surgery and chronic invasive recording of iEEG. The concordance with the invasive results suggests the clinical feasibility of such a noninvasive electrophysiological seizure imaging protocol. It is widely applicable to every patient because of its noninvasiveness. As a potential pre-surgical evaluation tool, it can be used as priori information to assist the implant of iEEG electrodes. If it is consistent with other non-invasive neuroimaging or neurological tests, the proportion of patients that must undergo invasive iEEG monitoring can be reduced. By adding this seizure imaging component, the non-invasive neuroimaging aspect of the pre-surgical evaluation in epilepsy can be significantly enhanced.

Epileptic seizures impact widespread brain networks beyond the extent of the epileptogenic zone. As a result, patients can experience a variety of manifestations in

behaviors. One common phenomenon is the loss or decrease of consciousness and responsiveness to the external world (Arthuis et al., 2009; Englot et al., 2010; Guye et al., 2006; Lee et al., 2002; Lux et al., 2002). In the analysis of abnormal slow-wave rhythms following epileptic seizures, we found changes of cortical electrophysiology corresponding to the different levels of cognitive deficits in epileptic patients. In line with previous neuroimaging and intracranial EEG studies (Lux, et al., 2002; Lee et al., 2002; Englot et al., 2010; Arthuis et al., 2009; Guye et al., 2006), yet providing more cortical coverage, our results suggest that widespread neocortical regions are affected by seizures. The spectral shifts to lower-frequency and the spatial shifts to neocortical regions (most prominently to frontal regions) outside of the epileptogenic zone may be an electrophysiological representation of the impact of epileptic seizures on the brain. These changes in slow rhythms may also be related to the altered cognitive functions in epilepsy patients.

Although continuous spatiotemporal imaging can be achieved noninvasively by EEG source imaging techniques, by adding multimodal information from other modalities, the spatial and temporal resolutions can be significantly improved. In the dissertation research, we also developed an EEG-fMRI integrated imaging approach, which combines the temporal resolution of EEG and spatial resolution of fMRI. We have shown in our computer simulations and experimental study in normal healthy subjects that we are able to reconstruct the continuous rhythmic activities, as well as their modulation that corresponds to the behavioral changes. Such a high-resolution spatiotemporal imaging technique can be important for neuroscience research. Potential

applications also include the localization of the epileptic networks in epilepsy patients (Grouiller et al., 2011; Laufs et al., 2007; Tyvaert et al., 2008).

In summary, the present dissertation research developed and evaluated the spatiotemporal source imaging approaches for the non-invasive localization of dynamic neural activity in the diseased and normal brains. An evaluation has been conducted in patient groups in order to test the clinical applicability of such a pre-surgical noninvasive seizure imaging tool. An investigation has been conducted to study the widespread impact of epileptic seizures on the brain and, as a result, on the behaviors and cognitions of epilepsy patients. The spatial resolution has been further improved by adding the component of fMRI through an EEG-fMRI integrated imaging framework. All of the results that were obtained suggest the importance of noninvasive spatiotemporal neuroimaging approaches for solving clinical problems and for investigating neuroscience questions. Furthermore, an improved understanding of neurological diseases and their mechanisms would help us to develop and deliver curative treatments of neurological diseases.