Using Network Dynamic fMRI for Pre-surgical Localization of Epileptogenic Foci

Pi：Mujica-Parodi，LiliannProposal编号：1264440智力优点：对于这项提议，我们将开发一种基于神经生物学的仪器来识别癫痫的癫痫灶。癫痫发作会逐渐破坏大脑；因此，早期干预对于最大限度地增加完全康复的机会至关重要。对于所有无法通过药物控制癫痫发作的患者中，大约三分之一的人最常见的治疗选择是手术切除被认为是癫痫发作灶的脑区。然而，该程序假定病灶可以被准确地识别。当它们不能通过神经成像的标准技术来识别时，就像大约一半的耐药局灶性癫痫患者一样，选择要么不治疗，要么手术反复试验，这两者都可能导致与疾病本身一样严重的神经退化。基于动力学系统和统计物理的熵方法已被应用于脑电信号的癫痫识别，最近又被应用于癫痫灶的识别，并取得了一定的成功。然而，这些技术的临床应用失败了，因为EEG的空间分辨率很差，而且无法接触到通常与癫痫发作有关的皮质下区域。功能磁共振成像具有识别神经外科目标所需的三维全脑覆盖和空间分辨率。遗憾的是，血流动力学时间序列通常太短和太稀疏，不能应用大多数标准的信息论方法，而fMRI采集和处理技术在多大程度上保留了信号动力学是一个尚未解决的基本工程问题。拟议的三年研究计划提供全面优化，包括在采购、硬件、软件和分析技术方面的创新，包括四个部分。首先，我们将评估和提高fMRI动力学的保真度。这包括开发用于功能磁共振成像的动态体模，首次允许在图像处理管道的每个阶段对已知的“血流动力学”输入和信号输出进行定量比较/校正，以及使用颅内脑电(“金标准”)、头皮脑电和功能磁共振从相同空间坐标获取的时间序列之间的定量比较/校正。其次，为了帮助识别网络异常，我们将得出标准的掩码来控制刺激或默认网络激活模式。第三，我们将研究信号复杂性异常与网络连通性之间的关系，后者对于在突触水平上理解癫痫易感性的病因具有重要意义。最后，我们将使用支持向量机来开发自动算法来识别假定的癫痫灶，如手术切除后的颅内电极和/或癫痫发作自由所证实的。如果成功，我们提出的方向将最终为神经外科医生在难治性隐源性癫痫的外科治疗方面取得潜在的革命性进展，从而满足NSF资金机制的使命，旨在支持“基础工程和科学知识的显著进步”，而不是渐进的改进。这项建议将计算技术和仪器的发展与直接的临床应用相结合。广泛的影响：2007年，美国国家科学院、国家工程院和美国医学研究所受国会委托成立了一个委员会，以应对与在日益全球化的经济中保持科学创新和经济竞争力相关的挑战。对于这项提案，我们将重点解决具体建议。一项行动项目是通过加强科学教师本身的科学和工程教育来加强儿童的K-12科学技术准备。行动项目将通过为1-6年级开发实践工程设计和创新课程来解决。这一课程将在一所社会经济多元化的学校反复测试和完善，并通过我们的网站传播，并将包括后续的NWEA个性化评估，以衡量提高学生STEM表现的有效性。通过在实验室培训学生在自己的研究中发展工程设计和创新，然后培训他们在更基本的层面上向教师和小学生传授相同的概念工具，我们能够最大限度地将我们的研究和教育目标结合在一起。

PI: Mujica-Parodi, LilianneProposal Number: 1264440Intellectual Merit: For this proposal, we will develop a neurobiologically-based instrument to identify seizure foci in epilepsy. Seizures gradually destroy the brain; therefore, early intervention is critical to maximize chances of full recovery. For the roughly one third of all patients whose seizures cannot be managed through medication, the most common treatment option is surgical removal of the brain areas thought to be seizure foci. However, the procedure assumes that foci can be identified with precision.When they cannot be identified through standard techniques in neuroimaging, as in about half of all patients with medication-resistant focal epilepsy, the options are either no treatment or surgical trial and error, both of which can lead to neurodegeneration as severe as the disease itself. Entropic methods adapted from dynamical systems and statistical physics have been applied to EEG to identify seizures, and most recently have been applied to the identification of seizure foci, with some success. However, clinical adoption of these techniques has failed due to EEG's poor spatial resolution and inability to access sub-cortical regions commonly implicated in seizures. FMRI has the three-dimensional whole brain coverage and spatial resolution required for identifying neurosurgical targets. Unfortunately, hemodynamic time-series are typically too short and sparse to permit application of most standard information theoretic methods, and the degree to which fMRI acquisition and processing techniques preserve signal dynamics is a fundamental engineering question that remains unaddressed. The proposed three-year research plan provides for comprehensive optimization, including innovations in acquisition, hardware, software, and analytical techniques, and is comprised of four parts. First, we will assess and improve the fidelity of fMRI dynamics. This includes both instrumentation development of a Dynamic Phantom for fMRI, for the first time permitting quantitative comparison/correction between known "hemodynamic" inputs and signal outputs at each stage of the image processing pipeline, as well as between time-series acquired from the same spatial coordinates using intracranial EEG (the "gold standard"), scalp EEG, and fMRI. Second, in order to aid in identification of network abnormalities, we will derive normative masks to control for stimulus or default-network activation patterns. Third, we will investigate the relationship between abnormalities of signal complexity and network connectivity, the latter of which has critical implications for understanding the etiology of seizure vulnerability at the synaptic level. Finally, we will use support vector machine to develop automated algorithms for identification of putative seizure foci, as confirmed by intracranial electrodes and/or seizure freedom following surgical resection. If successful, our proposed direction would culminate in providing neurosurgeons with a potentially revolutionary advance in surgical treatment of intractable cryptogenic epilepsy, and thus satisfies the mission of the NSF funding mechanism, designed to support "significant advancement of fundamental engineering and scientific knowledge" rather than incremental improvements. This proposal integrates development of computational techniques and instrumentation with direct clinical applications.Broader Impact: In 2007, the National Academy of Science, National Academy of Engineering, andInstitute of Medicine were charged by Congress to form a committee to address the challenges associated with maintaining scientific innovation and economic competitiveness within an increasingly global economy. For this proposal, we will focus on addressing specific recommendations. an action item was to strengthen children's K-12 preparation in science and technology by enhancing the science and engineering education of the science teachers themselves. The action item will be addressed through the development of a hands-on engineering design and innovation curriculum for grades 1-6. This curriculum will be iteratively tested and refined in a socioeconomically-diverse school, disseminated through our website, and will include follow-up NWEA individualized assessment to measure efficacy in improving student STEM performance. By training students in the lab to develop engineering design and innovation in their own research, and then training them to teach teachers and elementary school students the same conceptual tools at a more basic level, we are able to integrate our research and educational goals to the fullest extent possible.