A MIMO channel approach for characterizing electrode-tissue interface in long-term chronic microelectrode array recordings

Characterizing the encapsulation layer caused by glial scar formation surrounding microelectrode arrays in chronic implants has been the subject of extensive research. Typically, an equivalent circuit model is used to characterize the reactive tissue response by nonlinearly fitting the electrical impedance spectroscopy (EIS) data. This model assumes a time invariant adjacent layer of encapsulation tissue to have the same structure on every electrode site. In this paper, an alternative approach is proposed based on modeling the encapsulation layer as a time varying communication channel. The channel is characterized by a multi-input multi-output (MIMO) transfer function with time varying coefficients. This model circumvents spatial resolution limitations of existing EIS equivalent circuit models. It further allows capturing the observed changes in neural signal quality over time. We show that "equalizing" the channel using this model can yield a substantial improvement in signal quality. With tendency towards high-density electrode arrays for cortical implantation, the proposed model is better suited to equalize the fading channel and interpret the recorded signals with higher accuracy. We also show conceptually how patterned waveforms can periodically be used to probe the channel if adverse effects can be avoided. This can potentially improve the channel estimator performance, particularly when cell migration occurs