CSR: Small: Ultra-Low Power Analog Computing and Dry Skin-Electrode Contact Interface Design Techniques for Systems-On-A-Chip with EEG Sensing and Feature Extraction

脑电图（EEG）用于分析许多神经系统疾病，如癫痫、睡眠障碍、脑病和昏迷。EEG信号的永久监测和处理有助于医院环境内外的治疗。然而，对于可穿戴无线EEG设备，信号采集、处理和通信中涉及的功耗仍然很高。本研究将开发一个超低功耗脑电采集与处理系统芯片，采用一种新的模拟计算技术代替传统的数字处理系统。该SoC将能够识别模拟域中的癫痫发作事件，结合学习和连续信号处理。模拟处理和特征提取能力将通过精确的放大器和滤波器设计技术实现，以实现增益和滤波器截止频率的稳定性低至10秒/百万分之一（ppm）/度。特征提取方法将通过利用这些精确的模拟放大器和滤波器来测量各种EEG谱带中的功率水平，以检测癫痫发作。功率电平阈值设置和基于简单矢量模型的训练方法将在片内实现，用于癫痫发作表征和检测。将设计具有在线校准的电容消除方案，以获取需要长期监测的脑机接口的具有较高输入阻抗的EEG信号。这项研究的结果将改善EEG信号的采集，用于预测小型便携式设备的癫痫发作，这影响了世界2%的人口。所提出的SoC将特别有益于未来的小型化可穿戴设备，用于医院环境之外的连续EEG信号监测。从该项目获得的知识将被纳入研究生和本科生教育;该项目的成果将通过期刊文章和会议介绍传播。本科研究人员和高中实习生将参与该项目并接受培训。 作为本研究的一部分收集的公开共享数据将存入东北大学的数字存储库服务（DRS），这是一个由图书馆开发和维护的数字档案（https：repository.library.northeastern.edu）。它为它存储的文件提供安全性，以及访问管理控制和对各种元数据标准的支持，以帮助确保数据在现在和将来都可以访问和使用。所有项目参与者都可以访问存储在本地服务器上的项目管理数据库，其中包含设计和模拟数据。项目数据将在项目结束后保留至少3年。该奖项反映了NSF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。

Electroencephalography (EEG) is used for the analysis of many neurological disorders such as epilepsy, sleep disorders, encephalopathy, and coma. Perpetual monitoring and processing of EEG signals helps the treatment inside and outside of the hospital environment. However, the power consumption involved in signal acquisition, processing, and communication has remained high for wearable wireless EEG devices. This research will develop an ultra-low power (ULP) EEG acquisition and processing system-on-a-chip (SoC) using a new analog computing technique instead of conventional digital processing system. This SoC will be able to identify a seizure event in the analog domain, incorporating learning and continuous signal processing.The analog processing and feature extraction capability will be realized with precise amplifier and filter design techniques to achieve stabilities down to 10s of parts-per-million (ppm)/degree for gains and filter cutoff frequencies. The feature extraction method will measure the power levels in various EEG spectral bands by utilizing these precise analog amplifiers and filters to detect the onset of seizures. Power level threshold setting and simple vector model based training methods will be implemented on-chip for seizure characterization and detection. A capacitance cancellation scheme with online calibration will be devised to acquire EEG signals with higher input impedance for brain-computer interfaces requiring long-term monitoring. The results from this research will improve the acquisition of EEG signals for predicting the onset of seizures with small portable devices, which impacts 2% of the world's population. The proposed SoC will be particularly beneficial in future miniaturized wearable devices for continuous EEG signal monitoring outside of hospital environments. Knowledge obtained from this project will be integrated into graduate and undergraduate education; results from the project will be disseminated through journal articles and conference presentations. Undergraduate researchers and high school interns will be involved and trained in the project. Publicly shared data collected as part of this research will be deposited into Northeastern University's Digital Repository Service (DRS), which is a digital archive developed and maintained by the library (https://repository.library.northeastern.edu). It provides security for the files it stores, as well as access management controls and support for various metadata standards to help ensure that data is as accessible and usable in the present and the future. All project participants will have access to a project management database stored on local servers with design and simulation data. The project data will be maintained for at least 3 years after the conclusion of the project.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.