SBIR Phase I: A Novel Analytical Tool to Localize the Epileptogenic Zone in Medically-Refractory Epilepsy

SBIR一期项目需要开发EEG分析软件应用程序，用于识别医学难治性癫痫（MRE）患者的癫痫区（EZ -癫痫发作开始于大脑的地方）。美国有超过100万人患有MRE，这意味着他们对药物没有反应。MRE患者经常住院，承受着癫痫相关残疾的负担，在美国每年用于治疗癫痫患者的160亿美元中，MRE患者占80%。有两种治疗方法：(i)手术切除EZ， （ii）神经刺激，电刺激EZ以抑制癫痫发作。成功的治疗结果主要取决于从侵入性脑电图记录中准确识别EZ，这是一个漫长而昂贵的过程，导致30%-70%的治疗患者继续癫痫发作。因此，人们一直在寻找一种准确的数据分析工具，以减少侵入性监测的时间、风险和成本。该项目涉及进一步开发这样一种工具，该工具可以从脑电图数据中生成视觉“热”图。该工具以动力系统理论和神经工程为基础，已通过20例患者的数据进行验证，预测手术结果的准确率达到95%。减少监测时间可以减少大脑暴露感染的风险，并减少与长时间住院和临床工作人员审查数据相关的医院费用。通过提供更准确的EZ定义，该工具还将允许使用精确的全新激光消融程序，在目标结构中形成微小病变，而不是去除大脑的大部分。如果成功，该工具将在可持续的商业模式下更接近商业化。主要的EEG供应商和医疗设备公司正在寻找癫痫治疗中精确的软件应用程序，以增强他们的产品套件，并且对授权该工具非常感兴趣。这个小企业创新研究第一阶段项目涉及开发一种尖端的脑电图工具，该工具使用动态网络建模和高度创新的专利理论，即动态网络中节点的“脆弱性”，从侵入性脑电图记录中定位EZ，同时考虑到大脑中神经元的广泛互连。一个EEG通道越“脆弱”，它就越有可能出现在EZ。该项目的目标是：(i)在大量患者队列中验证该工具，在癫痫发作之前、期间和之后使用侵入性脑电图数据；(i)测试工具？（iii）设计用户界面，并将该应用程序集成到现有的临床工作流程中，以促进前瞻性研究。这些里程碑将最大限度地降低将这项创新推向市场的关键风险，包括采用、感知责任、监管批准和报销。如果该工具准确、快速且易于使用，只需按下按钮即可接收脆弱性地图，则采用风险将得到降低。如果我们完成的回顾性研究，包括网络模型的改进，显示出与我们的初步数据相当的性能，那么准确性风险将会降低。通过开发与现有EEG数据采集和可视化工具重要集成的直观界面，可以降低快速和易于使用的风险。由于存在谓词装置，监管风险很低。该工具在误诊方面的感知责任风险较低，因为该工具并不打算取代临床医生的分析，而是为临床工作流程中已经收集和分析的脑电图数据提供增强的可视化（如我们的回顾性研究所示）。最后，如果使用该工具准确识别EZ，有可能显著减少甚至消除病灶MRE部分，从而每年减少60亿美元的癫痫相关费用，则可以降低报销风险。因此，医疗保健和保险提供者将有强烈的动机为癫痫诊所支付或偿还该工具。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

This SBIR Phase I project entails the development of an EEG analysis software application that identifies the epileptogenic zone (EZ - where seizures start in brain) in medically refractory epilepsy (MRE) patients. Over 1 million people in the US have MRE, meaning that they do not respond to medication. MRE patients are frequently hospitalized, burdened by epilepsy-related disabilities, and contribute to 80% of the $16 billion dollars spent annually in the US treating epilepsy patients. There are 2 treatments: (i) surgical removal of the EZ, and (ii) neurostimulation, where the EZ is electrically stimulated to suppress seizures. Successful outcomes depend critically on accurately identifying the EZ from invasive EEG recordings, which is a long costly process, leading to grim outcomes where 30%-70% of treated patients continue to have seizures. There has thus been an intensive search for an accurate data analytics tool to reduce time, risks and costs of invasive monitoring. This project involves further development of such a tool that generates visual "heat" maps from EEG data. The tool, grounded in dynamical systems theory and neuroengineering, has been validated with data from 20 patients, achieving 95% accuracy in predicting surgical outcomes. Reducing monitoring time reduces the risk of infection from the brain being exposed, and reduces hospital costs associated with lengthy stays and clinical staff reviewing data. By providing more accurate definition of the EZ, the tool will also enable use of a precise and entirely new laser ablation procedure that makes tiny lesions in targeted structures as opposed to removing large portions of the brain. If successful, the tool will be closer to commercialization under a sustainable business model. Major EEG vendors and medical device companies are looking for accurate software applications in epilepsy treatment to enhance their product suites, and will be very interested in licensing the tool. This Small Business Innovation Research Phase I project involves development of a cutting-edge EEG tool that uses dynamic network modeling and a highly innovative and patented theory of "fragility" of nodes in a dynamic network to localize the EZ from invasive EEG recordings, taking into account the extensive interconnection of neurons in the brain. The more "fragile" an EEG channel, the more likely it is in the EZ. Project aims are to (i) validate the tool on a large patient cohort, using invasive EEG data before, during and after seizure events; (i) test the tool?s efficacy using noninvasive scalp EEG recordings and (iii) design the user-interface and integrate this application into the existing clinical workflow to facilitate prospective studies. These milestones will minimize key risks in bringing this innovation to market, which are adoption, perceived liability, regulatory approval and reimbursement. Adoption risk will be mitigated if the tool is accurate, quick and easy-to-use, requiring essentially the push of a button to receive fragility maps. Accuracy risk will be mitigated if our completed retrospective study, including refinement of network models, shows comparable performance to our preliminary data. The quick and easy-to-use risks will be mitigated with the development of an intuitive interface that importantly integrates with the existing EEG data acquisition and visualization tools. Regulatory risk is low as a predicate device exists. Perceived liability of the tool in mis-diagnosis is a low risk as the tool is not intended to replace the clinician's analysis, but rather it provides an enhanced visualization of the EEG data (as demonstrated in our retrospective study) already being collected and analyzed in the clinical workflow. Finally, reimbursement risks will be mitigated if accurate identification of the EZ using the tool has the potential to significantly reduce or even eliminate the focal MRE segment reducing epilepsy-related costs by $6 billion/year. Consequently, healthcare and insurance providers will have a strong incentive to pay for, or reimburse epilepsy clinics for the tool.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.