Brain Function: Novel Technologies Driving Novel Understanding - Bioinspired Approaches for Human-Centric Technologies - page 311

sampling capabilities beyond the few tens or hundreds of neurons that can be
currently measured simultaneously (Buzsaki 2004 ; Stevenson and Kording 2011 )
as well for improved signal quality (Spira and Hai 2013 ). Recent achievements in
nanostructuring capabilities and microelectronic circuits applied to large-scale
neural recordings open new perspectives that have the potential to dramatically
scale-up the performance of multielectrode devices.
Conventional multielectrode probes are realized using microfabrication pro-
cesses to integrate tens of microelectrodes on structured substrates (typically
silicon) and to embed the electrical wires on-probe to connect each electrode to
on-chip or off-chip signal conditioning and acquisition circuits. Currently, two
major types of probes are commercially available, using either “in-plane” fabrica-
tion approaches with micron-scale photolithography (Wise et al. 2004 , 2008 ),
compatible with on-probe integration for signal conditioning and multiplexing
(Sodagar et al. 2009 ) or “out-of-plane” processing from a single block of silicon,
using etching, doping, and heat treatments to realize a three-dimensional array of
“needlelike” electrodes (Maynard et al. 1997 ; Rousche and Normann 1998 ;
Nordhausen et al. 1994 ). Current technologies allow access to tens to hundreds of
neurons simultaneously, but multielectrode recording must be dramatically scaled
up to measure signals from thousands of neurons. This requires novel array
architectures to individually address each electrode while spatially constraining
the geometry and size of the probe.
10.10 Large-Scale CMOS Multielectrode Arrays
IIT scientists have contributed to developing novel generations of dense active
multielectrode arrays with several thousand micro-/nanoelectrodes (Berdondini
et al. 2009 ; Hierlemann et al. 2011 ). These arrays are realized with standard
complementary metal-oxide-semiconductor (CMOS) technologies. The adoption
of CMOS technology and the development of microelectronic circuits for
multielectrode-array recordings have drastically increased during the last decade
(Jochum et al. 2009 ). Circuits have been developed to provide signal conditioning
close to the microelectrodes, to multiplex signals to reduce output wires, and to
wirelessly transmit data (Perlin and Wise 2010 ), but CMOS circuits are increas-
ingly used to record from a larger number of electrodes simultaneously. Hybrid
architectures of application-specific integrated circuits (ASICs) realized in CMOS
technology and connected to passive electrode arrays have been developed
(Dabrowski et al. 2004 ; Blum et al. 2007 ; Bottino et al. 2009 ; Grybos et al. 2011 )
and used to record from hundreds of retinal ganglion cells in ex vivo retina, with
single-cell resolution (Field et al. 2010 ). However, active multielectrode-array
architectures could dramatically increase the number and density of microelec-
trodes (Berdondini et al. 2009 ; Eversmann et al. 2003 ; Heer et al. 2004 ). Berdondini
and collaborators introduced the first high-resolution active MEA, in 2001, incor-
porating 4,096 electrodes (Berdondini et al. 2001 ). In this device (today also