Multichannel Front - End circuit with Action Potentials / Local Field Potentials separation for wireless neural recording

HE bottleneck of the Systems on Chip (SoC) for simultaneous multichannel wireless neural recording is the RF transmitter. It is due to the disproportion between its throughput limited by the requirements of low power consumption and the amount of data acquired from the multielectrode probes. In order to reduce the amount of data without loss of neurobiological information, a sub-band coding technique can be applied, which divides the recorded signal into two or more different frequency bands encoding each of them independently. In the case of neural signals, the sub-bands are in the range of 0.5Hz-200Hz for Local Field Potentials (LFP) and 200Hz-7.5kHz for Action Potentials (AP). By applying different gains and sampling rates for the APs and LFPs, significant data reduction can be achieved [1]. The presented integrated circuit consists of 8 signal-conditioning channels with outputs connected to an 8:1 analogue on-chip multiplexer (Fig. 1a). Each channel consists of one common preamplifier and two different amplifiers for the AP and LFP separation with outputs connected to a 2:1 analogue multiplexer (Fig. 1b). The amplifiers have different gains and digitally adjustable (8-bit DACs) cut-off frequencies. The lower cut-off frequency of the preamplifier can be tuned individually for each channel in order to achieve a low spread of low cut-off frequencies in the multichannel system. Cut-off frequencies of the AP and LFP amplifiers can be adjusted globally. The circuit was fabricated in CMOS 0.18 μm technology (Fig. 2). A single channel occupies 0.066 mm 2 . Table I shows basic parameters of the circuit. Neural activity has been recorded by extracellular methods from a hippocampus of an anesthetized rat. Fig. 1a shows that action potentials with amplitudes as low as 60μV can be detected applying the presented circuitry. Fig. 1b shows the APs and LFPs recorded simultaneously from the same electrode. A significant correlation between AP frequency and LFP level can be seen. The above results confirm that the presented circuitry is able to simultaneously amplify and separate the APs and LFPs with small amplitudes. Low power consumption and small size predispose the circuit to be used in SoCs for multichannel wireless neural activity recordings.