Validation of a Novel Automated Seizure Detection and EEG Trending System for Con

描述（由申请方提供）：对重症患者非惊厥性癫痫发作（NCS）发生率的认识不断提高，导致重症监护室（ICU）对连续EEG监测的需求不断增长。然而，在ICU中扩大使用连续EEG监测的最大挑战之一在于难以由EEG专家提供对临床数据的及时审查。因此，尽管EEG被连续记录，ICU医生可能直到几个小时后才被警告癫痫发作的发生，随后由专业人员对原始EEG进行彻底审查。最终结果是患者护理受损，无法做出适当的实时治疗决策。自动癫痫发作检测软件偶尔用于协助EEG审查过程。然而，现有的癫痫发作检测软件是不够的，因为大多数ICU患者的异常背景EEG和高度可变的癫痫发作放电，发生在脑病患者。这些EEG模式与癫痫监测单元（EMU）中出现的模式有很大不同。因此，市售软件在用于重症监护患者时表现不佳，导致错过癫痫发作和错误检测的高发生率。虽然可用的EEG趋势软件更有用，但现有算法存在显著的技术限制：1）不能显示bref、局灶性或缓慢演变、低频率癫痫发作的明显变化，2）在区分需要紧急治疗的NCS与通常不用抗癫痫药物治疗的其他异常方面的特异性低，和3）临床实用性的有限科学证据-没有FDA批准用于ICU。该SBIR项目的总体目标是开发和商业化一个准确，可靠和用户友好的ICU癫痫发作监测和警报系统，CereScope。该系统将采用一种新型的自动癫痫发作检测算法ICU-ASDA，具有高灵敏度（> 85%）和低错误检测率（< 0.2/小时或5/天）。为了提高系统灵敏度， 还将与伪影减少的新型定量EEG（qEEG）趋势，即有助于通过目视检查快速识别潜在癫痫发作模式的癫痫发作指数（SI）相结合。CereScope”将自动创建并传输包含检测到的事件的数字图形文件，以供立即进行专家审查。在完成ICU-ASDA的设计和培训研究后，在此I期项目中，我们提出了一项临床研究，以统计学方式评估和验证符合FDA要求的算法性能。此外，我们将在四个神经ICU进行研究，以调查如何将qEEG趋势与检测算法的结果相结合，以提高系统的整体性能。本I期可行性研究的具体目的是：1）进行一项临床研究，以评价新型癫痫发作检测算法ICU-ASDA在急性成人患者EEG记录中的性能（灵敏度和误检率），以及2）研究新型qEEG趋势与ICU-ASDA联合使用时在长期ICU EEG记录中识别NCS时提高灵敏度的效用。CereScope系统的成功商业化将改善危重患者癫痫发作的识别和管理。 公共卫生相关性：对危重患者癫痫发作高发生率的认识不断提高，导致重症监护室（ICU）对连续脑监测的需求不断增长。然而，在ICU中扩大使用连续脑监测的最大挑战之一是难以由专家及时审查大量EEG（脑电活动信号）数据，导致无法做出适当的实时治疗决策并损害患者护理。该SBIR项目的总体目标是开发一种用于EEG分析的新型系统，该系统将允许ICU医生，技术人员和护士快速识别癫痫发作以及脑功能的其他异常变化。

DESCRIPTION (provided by applicant): Increased recognition of the incidence of nonconvulsive seizures (NCSs) in critically ill patients has led to a growing demand for continuous EEG monitoring in Intensive Care Units (ICUs). However, one of the biggest challenges to the expanding use of continuous EEG monitoring in the ICU lies in the difficulty of providing a timely review of clinical data by EEG experts. As a result, though EEG is being recorded continuously, ICU physicians may not be alerted to the occurrence of seizures until several hours later, following thorough review of the raw EEG by specialized personnel. The end result is compromised patient care and the inability to make appropriate real-time treatment decisions. Automated seizure detection software is occasionally used to assist in the EEG review process. However, existing seizure detection software is inadequate for most ICU patients because of the abnormal background EEG and highly variable seizure discharges that occur in encephalopathic patients. These EEG patterns differ greatly from those patterns that occur in epilepsy monitoring units (EMUs). As a result, commercially available software performs poorly when used with critical care patients, resulting in missed seizures and a high incidence of false detections. While available EEG-trending software is more useful, there are significant technical limitations in existing algorithms: 1) inability to show clear changes for bref, focal or slowly evolving, low frequency seizures, 2) low specificity in differentiating NCSs that require urgent treatment from other abnormalities, which are usually not treated with anti-seizure medicines, and 3) limited scientific evidence of clinical utility - none are FDA approved for ICU use. The overall goal of this SBIR project is to develop and commercialize an accurate, reliable, and user-friendly ICU seizure monitoring and alert system, CereScope". The system will feature a novel automated seizure detection algorithm, ICU-ASDA, with a high sensitivity (> 85 percent) and low false detection rate (< 0.2/hr or 5/day). To enhance the system sensitivity, it will also be interfaced with artifact-reduced novel quantitative EEG (qEEG) trending, a Seizure Index (SI) which facilitates rapid recognition of potential seizure patterns by visual inspection. CereScope" will automatically create and transmit digital graphic files containing detected events for immediate expert review. Having completed the design and training studies for the ICU-ASDA, in this Phase I project we propose a clinical study to statistically evaluate and validate the performance of the algorithm that will meet the FDA's requirements. In addition, we will conduct studies at four Neuro-ICUs to investigate how combining qEEG trending with the results from the detection algorithm can enhance overall performance of the system. The specific aims of this Phase I feasibility study are: 1) Conduct a clinical study to evaluate the performance (sensitivity and false detection rate) of a novel seizure detection algorithm, ICU-ASDA, in EEG recordings from acutely ill adult patients, and 2) Investigate the utility of novel qEEG trends for enhancing sensitivity in identifying NCSs in long-term ICU EEG recordings when used in conjunction with the ICU-ASDA. Successful commercialization of the CereScope" system will improve the recognition and management of seizures in critically ill patients. PUBLIC HEALTH RELEVANCE: Increased recognition of the high incidence of seizures in critically ill patients has led to a growing demand for continuous brain monitoring in Intensive Care Units (ICUs). However, one of the biggest challenges to the expanding use of continuous brain monitoring in the ICU is the difficulty of providing a timely review of the high volumes of EEG (brain electrical activity signal) data by experts, resulting in an inability to make appropriae real-time treatment decisions and compromised patient care. The overall goal of this SBIR project is to develop a novel system for EEG analysis that will allow ICU physicians, technicians, and nurses to rapidly identify seizures as well as other abnormal changes in brain function.