Question icon
Your Current Search
Choose below to refine your search
Research Topic
Download abstract book

Download the NI2012 abstract book here. The page numbers in the index are clickable for easy browsing.


Tools for analysis of high-density multielectrode recordings in the neonatal mouse retina

Filed under:

Matthias H. Hennig (School of Informatics, University of Edinburgh), Oliver Muthmann (NCBS Banglaore, India and School of Informatics, University of Edinburgh), Alessandro Maccione (Department of Neuroscience and Brain Technologies , Italian Institute of Technology (IIT)), Mauro Gandolfo (3Brain GmbH, Landquart), Stephen Eglen (Department of Applied Mathematics and Theoretical Physics, University of Cambridge), Luca Berdondini (Department of Neuroscience and Brain Technologies , Italian Institute of Technology (IIT)), Evelyne Sernagor (Medical School, Newcastle University)

A new generation of high-density multielectrode arrays (MEAs) now allows the characterisation of neural population activity in unprecedented detail. Here, we present several methods developed for the analysis of such recordings from the neonatal mouse retina with the 4,096 channel Active Pixel Sensor (APS) MEA, which allows near-cellular resolution recordings (electrode diameter of 21um ; 42um pitch; Berdondini et al., Lab on a Chip, 2009, Neural activity in the neonatal mouse retina consists of spontaneous, correlated bursts of neighbouring retinal ganglion cells (RGCs), resulting in propagating waves. Retinal waves undergo several developmental stages, early waves are mediated by gap junctions (Stage I), followed by lateral connections between cholinergic starburst amacrine cells (late gestation to P9; Stage II), and finally by glutamatergic transmission (P10; Stage III) before they disappear at the time of eye opening (P12). Despite such major developmental changes in network connectivity, the low spatial resolution of conventional MEAs has, so far, prevented a systematic observation of concomitant changes in population activity. Here, we report several important developmental changes in wave dynamics as revealed by pan-retinal high-density recordings. A challenge in analysing such data is the quantitative analysis of the spiking activity of thousands of active channels, which typically show considerable heterogeneity. As histogram-based approaches are impractical for such data sets, we have developed an efficient, real-time capable method based on the representation of spike trains as random walks, which allows for reliable classification into bursting, regular or randomly active neurons. After detecting bursts of propagating activity (cf. Hennig et al, 2012,, waves were grouped and analysed with alpha-shapes, yielding measures of the size and recruitment within wave areas. We found that Stage II waves showed low recruitment and highly variable sizes, becoming larger and denser up to P6, and smaller and slower at P8-9. Stage III waves were denser and more restricted spatially. Finally, wave trajectories were obtained and similarities within recordings analysed by clustering them. This revealed that Stage II waves propagated slowly with a high degree of randomness, while Stage III waves were faster and follow few repetitive, non-random propagation patterns.
Tools for analysis of high-density multielectrode recordings in the neonatal mouse retina
Examples of retinal waves recorded in the mouse retina at P5 and P11. Panels on the left show raster plots of bursts extracted from the recorded spike trains and then grouped into individual waves, as indicated by different colours. Three individual waves are shown on the right (numbers in the left panels indicate the events shown). Waves propagation is indicated by a change from bright to dark colors. Grey areas indicate MEA channels where spikes were recorded during the experiment that did not participate in the waves shown.
Preferred presentation format: Poster
Topic: Electrophysiology

Filed under: