% 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”.
@PHDTHESIS{Stella:909350,
author = {Stella, Alessandra},
title = {{H}igher-order correlation analysis in massively parallel
recordings in behaving monkey},
volume = {81},
school = {RWTH Aachen University},
type = {Dissertation},
address = {Jülich},
publisher = {Forschungszentrum Jülich GmbH Zentralbibliothek, Verlag},
reportid = {FZJ-2022-03140},
isbn = {978-3-95806-640-3},
series = {Schriften des Forschungszentrums Jülich Reihe Information
/ Information},
pages = {xiv, 184},
year = {2022},
note = {Dissertation, RWTH Aachen University, 2022},
abstract = {It has been hypothesized that information processing in the
cortical network evolves through the subsequent activation
of groups of neurons called cell assemblies, and correlated
activity is thought to be the signature of their activation.
Numerous studies have assessed the presence of
precisely-timed spatio-temporal spike patterns (STPs),
defined here as sequences of spikes emitted by a set of
neurons with fixed time delays between the spikes, repeating
in the same configuration in all occurrences. SPADE
(Spatio-temporal PAttern Detection and Evaluation) was
introduced as an analysis method for the detection of
synchronous patterns, and then extended for the detection of
spike patterns with temporal delays in parallel spike
trains. However, the method was evaluated for the STP
detection on simple artificial data, and not yet applied on
experimental spike trains. In this thesis we introduce an
extension of the original statistical test of SPADE,
accounting for the temporal duration of the patterns, the
order of correlation and the frequency of pattern
occurrence. In this way, we assess that statistical
performances are strongly improved. Additionally, we propose
an optimized implementation of the mining algorithm of
SPADE. We test the implementation on a wide range of
different hardware and on real experimental data, showing
that it results to be between one and two orders of
magnitude faster andmore memory efficient. We also propose
five artificial data sets, reproducing with increasing
degree the statistical complexity of experimental data,
still being completely artificial and generated by point
process models. Such data sets may be employed as ground
truth for analysis methods of parallel spike trains.
Furthermore, we compare different surrogate techniques to
evaluate their effect on parallel spike trains statistics
and on the evaluation of STP significance. Our results show
that the most classical method of uniform dithering fails as
an appropriate surrogate, since it leads to underestimation
of significance. Thus, we propose an alternative method with
better performance. Finally, we analyze with SPADE
experimental data from the neural activity recorded from the
motor cortex of two macaque monkeys, trained to execute a
reaching-and-grasping task. We find that significant STPs
occur in all phases of the behavior, and are highly specific
to the behavioral context, suggesting that different cell
assemblies are active in the context of different behaviors.
Moreover, our analysis reveals neurons that are involved in
several patterns in differentbehavioral contexts, and are
not clustered in space},
cin = {INM-6 / IAS-6},
cid = {I:(DE-Juel1)INM-6-20090406 / I:(DE-Juel1)IAS-6-20130828},
pnm = {5231 - Neuroscientific Foundations (POF4-523)},
pid = {G:(DE-HGF)POF4-5231},
typ = {PUB:(DE-HGF)3 / PUB:(DE-HGF)11},
urn = {urn:nbn:de:0001-2022090718},
url = {https://juser.fz-juelich.de/record/909350},
}