植入式医疗装置能够显著提高人类对相关难治性疾病的治疗水平，是生物医疗领域当前的热点研究方向。然而，随着植入式医疗装置电极通道数量的提升以及对更高精度的要求，传统的基于高性能处理器的智能识别方法所消耗的动态功耗显著增加，这对植入式医疗装置的进一步低功耗化带来严峻挑战。本项目在对植入式心脏起搏器专用芯片和植入式癫痫治疗仪专用芯片研究的基础上，进一步探索具有生理信号异常识别功能的低功耗智能模拟前端电路，研究基于异常事件驱动的粗细两级分类方案，以在实现精准识别的同时大大降低装置的整体功耗；探索新的低功耗低噪声感知电路结构，使其能有效抑制刺激脉冲自干扰、运动和工频干扰以及电极失调，确保准确、可靠的信号感知和识别；探索有源运放和无源开关电容相结合的带宽可重构高阶噪声整形SAR ADC，以实现适应多种生物信号的高效AD转换。各电路初步分析已完成，项目结束时预期在生物医疗集成电路研究领域达到国际先进水平。

Because implantable medical devices can significantly improve the treatment of refractory diseases, they have become the important research fields in biomedical engineering. However, with the increase in the number of electrode channels of implantable medical devices and the pursuit of higher accuracy, the dynamic power consumed by traditional smart classifier based on high-performance processors has increased significantly, which poses severe challenges on the ultra-low power design for future implantable devices. Based on the research on ASICs for the cardiac pacemaker and neuromodulator, in this project the low power and smart analog frontend circuits with classification of abnormal signals will be further explored, and the event-driven two-level classification scheme will be studied in order to achieve accurate classification and reduce the overall power at the same time. To achieve stable and accurate signal sensing and classification, a low power and low noise sensing amplifier would be proposed to suppress the effects of stimulation artifacts, power interferences, motion artifacts and electrode offsets. Furthermore, a reconfigurable hybrid high-order integrator combining an active amp and passive switched capacitors would be used in the noise shaping SAR ADC to fulfill low power and high resolution for different biomedical signals. The preliminary analysis of each circuit has been completed, and it is expected to reach a level equal to or higher than that of international peers in the field of biomedical integrated circuit research at the end of the project.