I am an electrical engineer whose research focuses on the use of advanced algorithmic techniques to improve the reconstruction of biomedical images from physical data. My graduate work focused on the development of algorithms to improve pre-clinical imaging of fluorescent proteins in small animals. Since the fall of 2008, I have held a postdoctoral position at Children's Hospital Boston studying the use of scalp EEG data to non-invasively localize seizure foci for the planning of surgical interventions. My immediate career goals are to establish myself in the field of functional neuroimaging using electroencephalography. In the near term, I seek to demonstrate that these techniques can offer a non- invasive approach to the current clinical gold standard of electrocorticography. In the longer term, I want to translate high leadcount EEG to being a part of the standard clinical care of epilepsy. Additionally, I want to expand my research into the application of EEG localization to more general neuroscience topics such as auditory and visual evoked stimuli. This K25 will give me the clinical and medical experience necessary to complete a successful RO1 application, and transition to being an independent researcher. The proposed career development plan incorporates a combination of classroom learning and practical clinical shadowing to develop a greater understanding of basic neurology and its clinical applications. I will take several classes in neurology from the Harvard/MIT Health Science and Technology (HST) division. Additionally, I will shadow both epileptologists and neurosurgeons at Children's Hospital to attain a greater understanding of clinical procedure and the ways in which various sources of information affect diagnostic decision making. I will regularly attend surgical planning conferences, and observe the final resection procedures. Professionally, I will attend local presentations on neurology and neuroimaging, and a minimum of two national and international conferences on imaging and epilepsy per year. In as many as 30% of epilepsy patients, their condition is poorly controlled with existing medications. For these individuals, surgical resection of afflicted nervous tissue may be the only effective treatment approach. However, poor localization of the seizure focus means only a small number of these patients ever see an operating room. Current state-of-the-art clinical procedure uses a wide range of structural and functional tests to localize seizure activity. The current gold standard for localization of the epileptic focus is electroencephalography (ECoG). However, because of the highly invasive nature of this procedure only a small number of patients are evaluated. Advanced algorithmic processing of scalp EEG data has the potential to offer similar accuracy from a noninvasive screening technique. This would allow far more patients to be comprehensively screened for surgical potential, both after conventional drug therapy has failed and early in the progression of the disease, before epileptic networks have developed which may reduce the effectiveness of surgical interventions. The EEG signal is highly dependent upon the physical, and thereby electrical, structure of the head. The use of structural information as prior knowledge in the source localization problem can be essential to enhancing the capability for identifying and localizing seizure foci. The long term goal is to improve scalp based EEG source localization to a point where the use of highly invasive sub-dural electrodes can be reduced or eliminated. Furthermore, highly accurate localization would potentially enable minimally invasive treatment approaches to be pursued, improving patient prognosis.
The specific aims of this proposal are to 1) Construct and evaluate improved patient-specific modeling of electrical propagation. We will examine improved methods for modeling the skull and cerebrospinal fluid regions of the brain using new MR imaging approaches such as ultrashort echo time (UTE) imaging, and model based methods to account for partial volume effects in sulci. 2) Develop statistical models for data fusion within the EEG source localization problem. Constructed from structural and functional information extracted from MR imaging studies, these will be used in a Bayesian inversion framework to obtain statistically optimal maps of source activity. 3) Evaluate and quantify the benefits of high lead-count EEG systems. We will use the 128-lead EEG system currently at Children's Hospital to perform additional EEG studies and compare the accuracy of source localizations to those obtained with subsampled electrode data. We expect the additional information provided by the higher number of electrodes to yield increased localization accuracy, and thereby improved diagnoses. 4) Evaluate the relationship between localization and surgical outcome. Using quantitative metrics presented in Specific Aim 3, we will evaluate the interaction between accuracy and surgical outcome. Additionally, we will investigate whether a metric applied to the localization itself can act as a statistically significant predictor of surgical outcome.
Electromagnetic source localization provides three dimensional images of seizure activity within the brains of pediatric epilepsy patients. This proposal will use structural and functional information extracted from MR images to help improve the accuracy of source localization techniques. Improved maps of seizure activity will have a significant impact upon human health by allowing neurosurgeons to perform curative surgery in a larger portion of patients whose epilepsy is poorly controlled by current drug therapy.
