Design of A System for Cancelling Stimulus Artifact in Multi-Channel Neural Interfaces

Bidirectional neural interfaces record electrical activity in neurons and modulate their signaling by suppressing or enhancing the activity via an electronic device. Such devices have been shown to treat and rehabilitate subjects suffering from neural diseases as well as to conduct research to advance the treatment methods. These applications require a new paradigm, which is currently widely explored in neural interface designs: the ability to stimulate and record simultaneously to explore complex tissue responses and to automatically adjust the stimulation in real-time based on the recorded data, thus “closing the loop”. The recordings are contaminated by a stimulation artifact which is formed when the stimulation current flows into the device-tissue interface and is recorded at the time of stimulation event as an undesired waveform. This waveform often overwhelms the recording device and disturbs the recording data long enough to overlap with the response of the neural tissue to the stimulation. To isolate the response from the contaminating artifact, a sophisticated artifact cancellation methodology is needed. Current state-of-the-art works offer solutions to the artifact problem but are practically constrained in the maximum artifact amplitude they can accommodate. A generalized framework and a corresponding methodology are thus developed and described in this work.The proposed architecture uses a combination of a coarse feedback in analog domain with a template generation in the digital domain to suppress artifacts and separate them from the underlying neural signals. Two variants of the design are discussed and then tested for studies of neural networks. The design is validated in-vivo in a deep brain stimulation of a hippocampus in a rodent model. The design is also validated in-vitro in an emulated stimulation and recording of a rodent spinal cord, based on preliminary in-vivo work conducted prior. The architecture notably performed artifact cancellation and recording of neural signals in-vivo on the stimulating electrode with a large artifact. A ~1V artifact was generated and suppressed with an unprecedented ratio of 80-100dB, while recovering the neural signals