Implantable neural readout frontend design for closed-loop neuromodulation systems

The goal of the PhD research is to introduce progresses in the performance of neural read-out electronics, with a focus on miniaturization for high channel counts in implantable chips. The
research is framed in the SCATMAN SBO project, which aims at developing an implantable
neural interface able to fit on an abnormal Brain Cavity (aBC) wall for the study and treatment
of chronic post-stroke symptoms. In particular, this research explores techniques to reduce the neural readout circuitry footprint while maintaining a high signal quality, even in the presence of neural stimulation. 

A first prototype in 180nm CMOS has been fabricated, featuring both readout and stimulation circuitries. The achieved readout area per channel is only 0.0018 square millimiters, which is the smallest reported in a stimulation-compatible technology, while the total harmonic distortion shows a best-in-class peak of -72dB. Experiments on the tolerance of stimulation artifacts based on the error between the baseline and the reconstructed local field potential biomarkers have been performed as well. They show how a simple linear inteprolation technique, in combination with the frontend's capability to quickly clear the memory of the artifacts, yields very small errors and can thus enable an uninterrupted extraction of brain features in a true real-time closed-loop system.

A second prototype in 40nm CMOS that extends the channel count to 64 and features on-chip interpolation logic has been designed and is currenlty under measurement.