% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@MASTERSTHESIS{Canova:185738,
author = {Canova, Carlos},
title = {{S}tatistical {A}ssessment and {N}euronal {C}omposition of
{A}ctive {S}ynfire {C}hains},
school = {Universität Tübingen},
type = {Dipl.},
reportid = {FZJ-2014-07163},
pages = {75},
year = {2014},
note = {bitte dem korrekten Teilprojekt von : SPP 1665:
Aufschlüsselung und Manipulation neuronaler Netzwerke im
Gehirn von Säugetieren: Von korrelativen zur kausalen
Analyse zuordnen; Universität Tübingen, Diplomarbeit,
2014},
abstract = {The synfire chain (SFC) model has been suggested (Abeles,
1982, 1991) as a network model for cortical cell assemblies
(Hebb, 1949). It is composed of consecutive groups of
neurons, where each group is connected to the next in a
feedforward fashion by a large number of convergent and
divergent inputs. This connectivity structure enables stable
propagation of packets of synchronous spiking activity
through the network after stimulation of the first group
(Diesmann et al., 1999a). Recent advances in
electrophysiological recording techniques enable to record
one hundred or more individual neurons simultaneously,
thereby increasing the chance to detect active cell
assemblies. Schrader et al. (2008) and Gerstein et al.
(2012) suggested a method based on an intersection matrix to
detect active SFCs. Each entry in the matrix contains the
degree of overlap of identical neurons being active at two
different time bins. If a particular SFC isa ctivated twice,
a diagonal structure, composed of consecutive time bins of
high intersection values, appears in the matrix. Further
convolution of the matrix with a diagonal linear filter
enhances diagonal structures as a whole compared to isolated
high intersection values due to chance overlaps, allowing to
distinguish among the two. Several features of the data are
expected to interfere with the analysis. First, the contrast
between diagonal structures and the rest of the matrix can
be severely diminished by high firing rates of the neurons,
thereby increasing the intersection values of individual
background pixels. Second, undersampling of the system
and/or stochastic participation in assembly activation may
lead to diagonal structures that fluctuate in intensity or
even become discontinuous. Third, diagonal structures may
become wiggly due to interference of the analysis bin width
and the propagation speed of the SFC. This thesis introduces
a quantitative statistical analysis method for the presence
of SFCsbuilt on the original approach which features an
automatic identification of the neurons participating in the
SFC. In the first stage, six steps are performed in addition
to the construction of the original intersection matrix
(step 1 in Figure 6.1). Diagonal structures are enhanced by
convolving the intersection matrix with a block filter,
which enables to cope with wiggly structures (step 2). Then,
a statistical test is performed to assess if there are
significant sequences of high intersection values (step 3)
by generating multiple realizations of surrogate matrices.
In these surrogates, the positions of the original matrix
entries are randomized before filtering, thereby
implementing the null hypothesis of repeated synchronous
activations of specific neuronal groups but not in a
consecutive manner in time. The entries of all surrogate
matrices form thedistribution of matrix entries under the
stated null hypothesis. By choosing an upper quantile (e.g.
$0.1\%),$ a threshold is defined to identify statistically
significant entries (step 3). The resulting matrix is
converted into a mask of regions of interest (step 4), which
is then applied to the original intersection matrix (step
5). A second threshold is calculated from a further set of
surrogate matrices which are generated by dithering the
spike times, in order to implement a null hypotheses of no
spike synchrony in the data. The masked intersection matrix
is thresholded with this new significance level, yielding
the final matrix in step 6. Ideally, this final matrix
contains only significant entries which are the signature of
repeated SFC activity. The second stage of the method
characterizes the significant entries from the final matrix.
Using a clustering algorithm on those entries, the diagonal
structures are labeled as repetitions of an active synfire
chains (step 7). The IDs of the neurons participating in the
chain(s) are then extracted to identify the neurons
composing the SFC and their group membership (step 8). The
method was calibrated using stochastic simulations
consisting of repeating consecutive synchronous spike
patterns embedded in otherwise independent data. Performance
results from the calibrations show that in realistic
parameter regimes, e.g. with realistic spike rates and
downsampled networks, the new analysis method is able to
successfully identify large portions (> $90\%)$ of the
embedded SFCs while having low false positive and false
negative levels. Possible applications to
electrophysiological data were demonstrated, identifying
where the method has to be improved for practical use.},
keywords = {Unveröffentlichte Hochschulschrift (GND)},
cin = {INM-6 / IAS-6},
cid = {I:(DE-Juel1)INM-6-20090406 / I:(DE-Juel1)IAS-6-20130828},
pnm = {331 - Signalling Pathways and Mechanisms in the Nervous
System (POF2-331) / 89571 - Connectivity and Activity
(POF2-89571) / HBP - The Human Brain Project (604102) / SMHB
- Supercomputing and Modelling for the Human Brain
(HGF-SMHB-2013-2017)},
pid = {G:(DE-HGF)POF2-331 / G:(DE-HGF)POF2-89571 /
G:(EU-Grant)604102 / G:(DE-Juel1)HGF-SMHB-2013-2017},
typ = {PUB:(DE-HGF)10},
url = {https://juser.fz-juelich.de/record/185738},
}