Automated Algorithm for Synchronized Quantification of LFP
Recordings and Individual Behavioural Parameters
in an Animal Model for OCD
T. Tambuyzer
1
, H. Wu
2
, K. Bauweleers
1
, K. van Kuyck
2
, B. Nuttin
2,3
and J.-m. Aerts
1
1
M3-BIORES: Measure, Model & Manage Bioresponses, KU Leuven, Leuven, Belgium
2
Laboratory of Experimental Neurosurgery and Neuroanatomy, KU Leuven, Leuven, Belgium
3
Department of Neurosurgery, University Hospitals Leuven, Leuven, Belgium
1 OBJECTIVES
The knowledge on the origin of local field potential
(LFP) signals is making progress (Buzsaki et al.,
2012). To prove the clinical relevance of recording
LFP signals, synchronous comparison with
pathological behavioural parameters in animal
models for diseases is essential. This work describes
an algorithm, which is developed for automated
analyses of LFP recordings and behavioural
parameters in freely moving rats. Several studies
showed the advantages of behaviour monitoring
with video recordings (e.g. Zarringhalam et al.,
2012), but there are limited studies describing
correlations between neural and behavioural
recordings for freely moving rodents (Venkatraman
et al., 2010; Fan et al., 2011). Here, we present the
algorithm applied on a rat model for OCD
(schedule-induced polydipsia, SIP; Falk, 1961;
Moreno and Flores, 2012). The algorithm allows to
extract detailed behavioural parameters based on
images of a top view camera. Changes during the
disease conditioning period can be tracked and
correlated with synchronously measured changes in
the LFP recordings. We believe that such automated
algorithms can highly contribute to a deeper
understanding of recorded LFPs and their link with
pathological behaviour of individual animals.
2 MATERIALS AND METHODS
2.1 Animal Conditioning
Electrodes (twisted bipolar platinum electrode,
single strand diameter: 0.203mm, part number
E363/8-2TW, PlasticsOne) were implanted in the
bed nucleus of the stria terminalis (BNST)
bilaterally in one Wistar rat before SIP conditioning.
After three days of recovery the conditioning
commenced. During SIP conditioning, the rat
received food pellets under a fixed-time schedule
each day, and gradually developed compulsive
drinking behaviour (Moreno and Flores, 2012). A
sampling rate of 10000 samples/s was used for all
LFP recordings during conditioning.
2.2 Image Analysis
A webcam (Logitech HD Webcam Pro C910) was
fixed on top of the conditioning cage to record the
rat behaviour during conditioning (sampling rate:
15 images/s). The processing of the images
consisted of three main steps: (i) Cage detection:
Based on colour properties of the cage floor (dark
brown-black), the shape of the cage and the exact
locations of drinking and feeding tube were first
extracted. (ii) Rat detection: Rats could be
recognised using a specific threshold for
(approximately) white pixel values, which represent
the rat. After fixing the threshold to distinguish
between rat and background, a binary image was
obtained. For each video sample, the threshold was
automatically recalculated based on the expected
size of the rat to deal with changes of recorded light
intensities. (iii) Quantification of behavioural
parameters: Based on the rat detection, four
behavioural parameters were determined: rat
location in cage (in x- and y direction), drinking
behaviour (0 or 1), eating behaviour (0 or 1), and
walking patterns (i.e. correlation of changes in x-
and y- direction). All image processing steps were
executed in Matlab using the Image Processing
Toolbox.
Tambuyzer T., Wu H., Bauweleers K., van Kuyck K., Nuttin B. and Aerts J..
Automated Algorithm for Synchronized Quantification of LFP Recordings and Individual Behavioural Parameters in an Animal Model for OCD.
Copyright
c
2013 SCITEPRESS (Science and Technology Publications, Lda.)
2.3 LFP Analysis
After band-pass filtering (0.1-300 Hz), detrending
and noise cancellation, the power of specific
bandwidths was calculated using the fast Fourier
transform algorithm of Matlab. To create a direct
link between brain signal and the individual animal
in the graphical interface, these power calculations
were afterwards used for replacing the white colour
of the rat with a colour on a cold-hot scale: for high
powers, the rat was represented in dark red and
for low powers the rat was represented in dark
blue.
3 RESULTS
The developed algorithm required two inputs: a top
view video of the cage in .avi format and a
simultaneously recorded LFP signal. The average
time (+-standard deviation) needed by the algorithm
to analyse the video- and LFP recordings is 0.14s (+-
0.015s), which implies that the developed algorithm
could be used for real-time monitoring. The
accuracy of the rat detection was tested based on a
video of 20 minutes, which was not used for
algorithm training. In all samples of the test video
the rat could be correctly detected. Figure 1 shows
the output of the algorithm. The feeding tube and the
drinking tube are marked. The colour of the rat
corresponds to the power of the LFP in a specific
bandwidth, which can be selected by the user
Figure 1: Output of the automated LFP-video algorithm.
Location of food magazine (green) and water bottle (blue)
are shown by rectangular. The rat is represented in a
colour, which corresponds to the power of a specified
frequency band (e.g. delta band power of the left
hemisphere).
(e.g. the delta band in figure 1).Using this graphical
representation, one can easily visualize correlations
between behaviour (image at time t) and LFP
recordings (signal of a one second time interval [t-1
,
t]).
4 DISCUSSION
The developed algorithm presents an attractive way
to synchronize, visualize and analyse LFP data with
simultaneously recorded behavioural video data. The
algorithm allows to gain insight in neuronal
recordings, to assess animal model validity (e.g.
Nestler and Hyman, 2010) and to quantify severity
of disease in individual animals (e.g. Hooks et al.,
1994). It is expected that such automated
comparison of synchronised video- and LFP-
recordings could substantially contribute to finding
potential neurological biomarkers of psychiatric
disorders.
REFERENCES
Buzsaki, G., Anastassiou, C. A., Koch, C., 2012. The
origin of extracellular elds and currents—EEG,
ECoG, LFP and spikes. Nature Reviews
Neuroscience, 13, 407-420.
Falk, J. L., 1961. Production of polydipsia in normal rats
by an intermittent food schedule. Science, 133, 195-
196.
Fan, D., Rich, D., Holtzman, T., Ruther, P., Dalley, J. W.,
et al., 2011. A Wireless Multi-Channel Recording
System for Freely Behaving Mice and Rats. PLoS
ONE 6(7): e22033.
Hooks, M. S., Jones, G. H., Juncos, J. L., Neill, D. B.,
Justice, J.B., 1994. Individual differences in schedule-
induced and conditioned behaviors. Behavioural Brain
Research, 60, 199-209.
Moreno, M., Flores, P., 2012. Schedule-induced
polydipsia as amodel of compulsive behaviour:
neuropharmacological and neuroendocrine bases.
Psychopharmacology (Berl), 219, 647–659.
Nestler, E. J., Hyman, S. E., 2010. Animal models for
neuropsyciatric disorders. Nature Neuroscience,
13, 1161–1169.
Venkatraman, S., Jin, X., Costa, R. M., Carmena, J. M.,
2010. Investigating neural correlates of behaviour in
freely behaving rodents using inertial sensors. Journal
of Neurophysiology, 104, 569-575.
Zarringhalam, K., et al. , 2012. An open system for
automatic home-cage behavioral analysis and its
application to male and female mouse models of
Huntington’s disease. Behav. Brain. Res., 229, 216–
225.