A 256-Channel 0.227µJ/class Versatile Brain Activity Classification and Closed-Loop Neuromodulation SoC with 0.004mm<sup>2</sup>-1.51 µW/channel Fast-Settling Highly Multiplexed Mixed-Signal Front-End
Uisub Shin, Laxmeesha Somappa, Cong Ding, Yashwanth Vyza, Bingzhao Zhu, Alix Trouillet, Stéphanie P. Lacour, Mahsa Shoaran
Abstract
Closed-loop neuromodulation can alleviate disease symptoms and provide sensory feedback in various neurological disorders and injuries [1]. Energy-efficient realization of closed-loop devices with on-site classification is critical to enhancing therapeutic efficacy. Despite recent advances, existing SoCs with integrated machine learning are constrained by low channel count (8–32) [2]–[5] and poor generalizability. To address these limitations, this paper presents a versatile neuromodulation SoC that integrates: (1) a 256-channel area-efficient dynamically addressable analog front-end (AFE), (2) information-rich multi-symptom biomarkers, (3) a low-power tree-structured hierarchical neural network (NeuralTree) classifier, and (4) a 16-channel high-voltage (HV) compliant neurostimulator.