Brain Function: Novel Technologies Driving Novel Understanding - Bioinspired Approaches for Human-Centric Technologies - page 314

10.11 Experimental Capabilities of Large-Scale Electrode
Arrays

The APS multielectrode-array platform was initially validated on cultured neuronal
networks, ex vivo cortico-hippocampal brain slices, and mice retina preparations.
The high spatial resolution of the device allows monitoring of large neural
networks yet with the resolution to finely track spatial signal propagation down to
cellular dimensions. This has necessitated the development of a new range of
analysis approaches to track the spatial distribution of propagating extracellular
signals. At the same time, the high-sampling frequency allows monitoring of fast
spiking activity and slow oscillations, such as LFPs in brain slices. Recordings
performed on different types of preparations are illustrated in Fig. 10.9 . While
high-quality extracellular voltage traces can be acquired for conventional displays
of single-unit activity, spatial propagations of neuronal activity are represented
as image sequences of the 64 64 multielectrode-array area, incorporating extra-

cellular signals sensed over the entire array. The CMOS Multielectrode Arrays
(CMOS-MEAs) are able to sense both fast oscillations in the band of 700-6,000 Hz
and local field potentials or multiunit activity in the frequency band of 10-100 Hz.
The noise of the system is low enough to allow reliable spike detections from single
units.
In Maccione et al. ( 2010 ), we showed that high-density arrays provided a more
statistically reliable description of in vitro neural network activity compared to
standard MEAs, decreasing intra- and inter-experiment variability and making
these devices a potential tool for in vitro screening of drugs or toxins.
Furthermore, fine characterization of signal propagation in networks can identify
spatiotemporal interactions between different regions of a network at cellular
resolution. For instance, by combining optical microscopy imaging with high-
resolution recordings with CMOS-MEAs, in vitro networks can be structurally
and functionally characterized at the cellular scale (Maccione et al. 2012 ) with
respect to the activity of specific neural populations in the network (e.g.,
GABAergic vs. non-GABAergic) or by quantifying changes in functional connec-
tivity estimates. We have also developed adaptive algorithms to track and quantify
neural activity propagations: trajectories of spatiotemporal propagations in net-
works (Fig. 10.10a ) can be estimated based on the “center of activity trajectory”
analysis presented in Garofalo et al. ( 2009 ). We used a similar approach to
characterize chemically induced interictal events in cortico-hippocampal brain
slices (Fig. 10.10b ), by recording field potentials with high-density MEAs (Ferrea
et al. 2012 ). Such events were successively classified by generating maps of activity
based on clustering of LFP shapes recorded by multiple electrodes. This approach
has the potential to provide screens for novel neuropharmacological and neurotox-
icologal targets.