Organic Electrochemical Transistors based Decoding Hardware for Near-Sensor Brain Signal Processing

As the thriving of brain-computer interface (BCI) technologies, signal decoding algorithm and hardware have been rapidly developing for information interpretation as a crucial part of the BCI system. Although state-of-the-art algorithms can be capable for successful brain wave decoding to realize tasks such as speaking intention prediction or robot arm manipulation, unconventional processing and computing hardware are indeed desired for low-cost and high-speed decoding. Neuromorphic hardware provides solutions in this regard, for which low-complexity protocol and true-hardware implementations have been pursued. Herein, reservoir computing (RC) hardware is constructed based on memristive organic electrochemical transistors (OECTs) for intention of decoding signals from brain. The OECT devices with PEDOT:TOS/PTHF channel layer can be electrically regulated to realize transition from short-term memory (STM) element to long-term memory (LTM) element, which benefits the one-batch fabrication of reconfigurable dual-modal memory device. Based on this merit, the reservoir unit of the RC system is constructed with STM elements and designed as an adjustable network to fit the feature of inputs, at the same time LTM elements are used as trainable weights of the neural network in the readout layer. It is beneficial that the reservoir maps temporal inputs into reduced-size vectors, allowing feasible hardware demonstration of the readout layer in a small network size and leading to desirable short-delay decoding. With this hardware system, simulated epileptic seizure signals can be detected online at decent accuracy, which further direct the generation of stimulus to form a close-loop BCI system for early depression of seizures. This function is transferable for varied types of brain signals at appropriate configuration, paving the way for further design of feasible decoding hardware for efficient online interpretation of brain disease or intentions.