Memristor based ultrasonic optimized excitations for mesoscopic nonlinear characterization of biomedical tissues

The electronic calibration process of the nonlin-ear time reversal based ultrasonic (US) imaging device (TR-NEWS) including the memosducer is presented and fabricated experimentally using Knowm Inc. memristors. Due to the wide frequency bandwidth of the instrumentation, an amplitude mod-ulated chirp-coded optimized US excitation is suggested in order to satisfy all constraints induced by the use of memristors, ultrasonic sensors and the biomedical tissues under analysis. The proof of concept is proposed and optimized with a 8-components memosducer prototype providing low and high resistance states values statistically distributed respectively around 200Ω and 5k Ω and associated to the [100kHz-5MHz] TR-NEWS imaging device bandwidth. Experimentally tested on a polymer phantom with analogue properties of the human skin, the demonstrator constitutes a promizing reservoir computing system based on the principle of neuromorphic devices, providing new generation of multiscale data highly compatible within the Health 4.0 community.