Increasing Reproducibility and Reliability of Novel Object Tests Through Standardization and Automation

Alicia F. Brantley, PhD
Scientific Director
Mouse Behavior Core
Scripps Research
Increasing Reproducibility and
Reliability of Novel Object
Tests Through Standardization
and Automation

An expert shares her work involving novel
object tests, the challenges associated with
these protocols, and the solutions she is
exploring to overcome these challenges.
and Automation

Copyright 2021 A. Brantley and InsideScientific. All Rights Reserved.
Scientific Director
Mouse Behavior Core
Scripps Research
and Automation

Novel Object Tests
• First developed by Ennaceur and Delacour in rats as a “pure
working memory test” with a 1-min retention interval
• Basic protocol: rodents explore 2 alike objects after habituation to
arena; after a delay (varies from short- to long-term, 1-24 hrs
usually), one object is replaced with a novel object (Novel Object
Recognition) or one object is moved to a novel location within the
arena (Novel Object Location)
• Analysis varies, usually by protocol; discrimination ratio vs.
comparison to chance

Replicating Novel Object Tests
• Protocols vary greatly
• Published methodological details are vague
• Objects can be anything, difficult to duplicate
• Observer-based scoring parameters vary
• Analysis choice affects result

Added Complication – Core Assay
How to take a task that is so lab-specific and make it work for
multiple labs with varied research experience and still be a valid,
replicable assay?
Some degree of standardization is necessary.
Automated scoring of exploration is preferable.

Object Placement
• Floor insert for open
field with a circular
cutout in each corner
• Blank disks fill in holes
when not needed to
make solid floor
• LED surround light bar
for even lighting of
objects and shadow
reduction

Object Choice
• First iteration – found lab
items epoxied to blank
disks
• Exploration was not
great
• Mice would climb on
them often,
complicating
automation of
exploration detection

Object Choice
What can mice see?
What is an “interesting” object and
what is a “scary” object?
How different do two “different”
objects need to be?
How tall should the objects be?
Will the objects affect automated
tracking?
• 3D printed objects
• Can create whatever I
want and replicate at
any time
• Objects can be like
common objects but
without seams or
difficulty replacing
• How to determine an
object to design?

Object Choice
• First 3D printed items
• Climbable
• One was highly
preferred over the
other
• Easy to revise designs
and print new options

Object Choice
• Current 3D printed items
• Approachable
• Difficult to climb/stand
on
• Differ in
texture/temperature
due to addition of
metal coil on one
object

Automating Object Exploration
First: which protocol?
• Fixed duration protocols – training and testing phases each given a fixed
amount of time (usually 10 minutes) during which the animal can freely
explore the object
• Exploration-based cutoff protocols – training only or training and testing
phases allowed to continue until a predetermined cumulative exploration time
is met (usually ~30 seconds)
• Controls for individual differences in exploration
Tuscher JJ, Fortress AM, Kim J, Frick KM. Regulation of object recognition and object placement by ovarian sex steroid hormones.
Behav Brain Res. 2015 May 15;285:140-57. doi: 10.1016/j.bbr.2014.08.001. Epub 2014 Aug 15. PMID: 25131507; PMCID:
PMC4329280.

How do we define exploration?
Ennaceur and Delacour (1988) defined exploration as “the directing the nose at a
distance 2 cm to the object and/or touching it with the nose,” while turning around or
sitting on the object was not considered exploration.
• Orientation of animal’s snout toward the object, sniffing or touching with snout
• Sometimes defined as the animal’s head oriented within 45 degrees of the object
• Primary difference between studies was the distance from the snout to the object;
varies usually within 1–4 cm distance from the object
Antunes M, Biala G. The novel object recognition memory: neurobiology, test procedure, and its modifications. Cogn Process. 2012
May;13(2):93-110. doi: 10.1007/s10339-011-0430-z. Epub 2011 Dec 9. PMID: 22160349; PMCID: PMC3332351.

How do we define exploration using EthoVision XT?
• Accurately assign nose point – cannot designate active interest in an object
unless the mouse’s face is at the very least in proximity to the object.
• Determine proximity distance threshold
• Identify orientation towards object

Improving Detection – Deep Learning
Deep learning refers to the use of deep neural networks, where “deep” indicates that
the networks are made of multiple, hidden layers of “neurons” or decision nodes.
Deep learning in EthoVision XT
• A deep neural network can find structures in unstructured data, like for pictures and
video images, and can recognize recurring patterns and relationships between those
patterns.
• Because of the layered structure of decision nodes, deep neural networks can learn to
represent data at various levels.
• EthoVision XT uses a trained network, that is, the network has learned to extract
features from a number of video images of rodents of various colors and in various
backgrounds, where the nose and the tail-base were previously annotated. During
tracking, the network analyzes a portion of the image that includes the detected
subject and makes an estimate of the position of the nose point and the tail-base
point.

Initial Cutoff Criteria
• Based on tracking of nose
point inside small area
around objects
• Set cutoff time to 45
seconds, longer than end
goal of 30 seconds

Refining Variables
• Choosing most accurate
variable
• Defining appropriate search
areas for higher accuracy
• Determination of
exploration is now easily
replicable

Redefine Cutoff
Criteria
• Apply refined
detection of
exploration to Trial
Control Settings
• Can now accurately
automate
exploration in real
time for future
experiments

Providing Replicable Methodological Details
All equipment and software details are easily shared
• Object 3D print file can be saved to cloud and shared in methodology.
• Open field and floor insert dimensions can be provided in equipment
description.
• Precise arena zone dimensions can be listed.
• “Exploration” condition definitions (from Trial Control Settings) can be
provided.
• Actual EthoVision XT template can be shared from cloud source.

Data Analysis – Which Method?
All equipment and software details are easily shared
• Discrimination ratios:
(Tnovel− Tfamiliar)/(Tnovel+ Tfamiliar) or Tnovel/(Tnovel+ Tfamiliar)
values near zero indicate chance performance, values greater than or
equal to 0.5, or greater than 0.5 (respectively) indicate preference
• One-sample t-test compared to chance performance
if total allowed exploration time = 30s, chance = 15s

Raw data
• Compare novel exploration
time (out of 30s) to chance
(15s) using one sample t-test
• Same data converted to a
discrimination ratio using
familiar object exploration time
Group
Time Novel Object
Exploration (s)
Tnovel/
(Tnovel+Tfamiliar)
(Tnovel− Tfamiliar)/
(Tnovel+Tfamiliar)
old 147.88 -0.15 4.92
17 12.24 -0.19 0.41
18 18.16 0.21 0.60
19 22.48 0.50 0.75
20 19.44 0.29 0.65
23 4.80 -0.68 0.16
25 13.60 -0.09 0.45
26 7.68 -0.49 0.26
27 19.32 0.29 0.64
28 16.64 0.11 0.55
30 13.52 -0.10 0.45
young 207.84 1.84 6.92
1 9.60 -0.36 0.32
10 9.96 -0.34 0.33
11 20.52 0.37 0.68
12 17.84 0.19 0.59
13 16.80 0.12 0.56
14 22.64 0.51 0.75
2 17.56 0.17 0.58
3 22.48 0.50 0.75
4 21.04 0.40 0.70
7 23.96 0.60 0.80
8 11.96 -0.20 0.40
9 13.48 -0.10 0.45
old-c 60.08 0.00 2.00
31 12.72 -0.15 0.42
32 21.04 0.40 0.70
132 9.00 -0.40 0.30
131 17.32 0.15 0.58
young-c 120.16 0.00 4.00
15 22.00 0.46 0.73
16 19.56 0.30 0.65
5 5.60 -0.63 0.19
6 5.56 -0.63 0.19
105 24.44 0.63 0.81
116 10.48 -0.30 0.35
106 24.48 0.63 0.81
115 8.04 -0.46 0.27
Grand Total 535.96 1.68 17.84

NOR Outcomes
Young mice Old mice
0
5
10
15
20
25
30
time
exploring
novel
object
(s)

NOR Outcomes
Y
o
u
n
g
m
i
c
e
O
l
d
m
i
c
e
y
o
u
n
g
c
o
n
t
r
o
l
o
l
d
c
o
n
t
r
o
l
0
5
10
15
20
25
30
time
exploring
novel
object
(s)
y
o
u
n
g
r
a
t
i
o
o
l
d
r
a
t
i
o
y
o
u
n
g
c
o
n
t
r
o
l
r
a
t
i
o
o
l
d
c
o
n
t
r
o
l
r
a
t
i
o
0.0
0.2
0.4
0.6
0.8
1.0
discrimination
ratio
y
o
u
n
g
r
a
t
i
o
c
s
v
o
l
d
r
a
t
i
o
c
s
v
y
o
u
n
g
c
o
n
t
r
o
l
r
a
t
i
o
c
s
v
o
l
d
c
o
n
t
r
o
l
r
a
t
i
o
c
s
v
-1.0
-0.5
0.0
0.5
1.0
discrimintion
ratio

Replicating Novel Object Tests
• Utilize available technology
• Reporting methods with detail
• Rely on automated tracking (always validate)
• Share with other labs/institutions

Thank you!
Ruud Tegelenbosch, MSc
Matt Feltenstein PhD

Scientific Director
Mouse Behavior Core
Scripps Research
Thank you for participating!
CLICK HERE to learn more and
watch the webinar

Thank you for participating!
Before you go…
1. Complete the Survey – we’d love to get your
feedback
2. Still have questions? Use the Ask a Question
panel
3. Interested in learning more? Explore the
Resources or visit www.noldus.com

Increasing Reproducibility and Reliability of Novel Object Tests Through Standardization and Automation

More Related Content

What's hot (20)

Similar to Increasing Reproducibility and Reliability of Novel Object Tests Through Standardization and Automation (20)

More from InsideScientific (20)

Recently uploaded (20)

Increasing Reproducibility and Reliability of Novel Object Tests Through Standardization and Automation