Theriogenology 67 (2007) 661–672

www.theriojournal.com

Development of a novel CASA system based on open

source software for characterization of zebrafish
sperm motility parameters
Jonas G. Wilson-Leedy, Rolf L. Ingermann *
Department of Biological Sciences and Center for Reproductive Biology, University of Idaho, Moscow, ID 83844-3051, USA
Received 15 August 2006; accepted 16 October 2006

Abstract
Although computer-assisted sperm analysis (CASA) outperforms manual techniques, many investigators rely on non-automated
analysis due to the high cost of commercial options. In this study, we have written and validated a free CASA software primarily for
analysis of fish sperm. This software is a plugin for the free National Institutes of Health software ImageJ and is available with
documentation at http://rsb.info.nih.gov/ij/plugins/casa.html. That it is open source makes possible external validation, should
improve quality control and enhance the comparative value of data obtained among laboratories. In addition, we have improved
upon the traditional velocity straight line (VSL) algorithm, eliminating inaccurate characterization of highly curved fish sperm
paths. Using this system, the motion of zebrafish (Danio rerio) sperm was characterized relative to time post-activation and the
impact of acquisition conditions upon data analysis determined. There were decreases in velocity and path straightness (STR), but
not linearity (LIN), relative to time. From 30 to 300 frames/s, frame rate significantly affected curvilinear velocity (VCL) and STR
measurements. Sperm density in the field of view did not affect any measured parameter. There was significant inter-male variation
for VCL, VSL, velocity average path (VAP), percent motility, path character (STR, LIN), and duration of motility. Furthermore,
relative sperm output (a measure reflecting both semen volume and concentration) was positively correlated to percent motility. For
all motion parameters measured (except duration), the average CV was 10%, comparable to values obtained using commercial
systems.
# 2006 Elsevier Inc. All rights reserved.

Keywords: CASA; Motility; Open source; Sperm; Zebrafish

1. Introduction analysis of these indicators of sperm quality, computer-

assisted sperm analysis (CASA) is the most objective and
Fish populations represent substantial natural and comprehensive quantification currently available and
commercial resources worldwide. One limiting factor in facilitates rapid assessment of percent motility, velocity
the reproduction of many species is low fertility [1]. and other descriptors. In studies of fish sperm motility,
Percent motility and sperm velocity have been correlated CASA has been employed, for example, in determination
to reproductive success in fish; therefore, reductions in of sperm variability among males [2], cryopreservant
these parameters may decrease fecundity [2–4]. For effectiveness [5], the effects of heavy metals [6],
quantification of differences in duration and initial speed
between sperm of bluegill adopting different reproduc-
* Corresponding author. Tel.: +1 208 885 6280. tive tactics [7,8], and elucidation of the effect of various
E-mail address: [email protected] (R.L. Ingermann). ions on path curvature of trout sperm [9].

0093-691X/$ – see front matter # 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.theriogenology.2006.10.003
662 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672

Despite CASA’s quantitative portrayal of motility immobilizing solution (ZSI, modified from Krasznai
parameters and rapid provision of descriptive data for a et al. [14], in mM: 140 NaCl, 10 KCl, 2 CaCl2, 20
sample, many investigators continue to rely on manual HEPES titrated to pH 8.5 with 1.0 M NaOH). Gentle
or semi-manual methods of motility assessment due to abdominal pressure was applied and a capillary tube
the high cost and corresponding low availability of was held near the vent to collect the semen, which was
CASA software. Semi-CASA systems have been subsequently diluted with 20 mL ZSI. Except for the
developed that describe sperm motion parameters based determination of the effect of relative sperm concentra-
upon a technician’s manual identification of sperm tion on motility analysis, below, all studies relied on this
paths. However, these systems may be impacted by the dilution to generate the initial semen suspension used
technician’s choice of sperm for quantification and are throughout the study. As determined by observation of
time intensive [3]. Like semi-CASA systems, visual fluid height in the capillary, the volume of semen
assessment at the time of sperm activation is out- obtained from each fish was approximately the same
performed by its computerized counterpart in terms of (about 1.5 mL). Lahnsteiner et al. [15] noted that
objectivity, reproducibility, and comprehensive descrip- seminal plasma components act to stabilize sperm
tion of sperm motion [10]. motility. To avoid differentially diluting seminal plasma
In the current study, we have constructed the only components in ZSI or in subsequent activations, and
freely distributable, open-source CASA system capable thus potentially generating differences among males
of measuring the motility parameters most commonly due to semen concentration rather than to male-to-male
used by investigators. In addition, we have validated the variability, the concentration of each sample was not
measurements generated, described the components of adjusted. Samples were kept on wet ice and were used
the system, and determined the impact of frame rate and within 1 h of collection.
sperm concentration on description of fish sperm Reagents were obtained from Sigma–Aldrich (St.
motion. Verstegen et al. [11] stated that for studies Louis, MO, USA).
based on CASA, methods need to be clearly and fully
described to facilitate reproducibility. A goal of the 2.2. Activation of sperm for analysis
current presentation was to make readily available this
versatile methodology and increase its utility by Sperm were activated by addition of 1 mL of semen
describing the components of our system. In addition, suspension in ZSI to 20 mL of aged tap water [16,17],
the present work sought to demonstrate the system’s using a Pipetman pipet (Ranin, Oakland, CA, USA),
value by analyzing the motility of sperm of the followed by 2 s of gentle stirring with the pipet tip.
zebrafish, Danio rerio. Although this species is a Approximately 0.5 mL of activated sperm was quickly
widely used biological model system with a wealth of applied to a single well in a 12-well multitest slide (MP
genetic and developmental information, little is known Biomedicals, Irvine, CA, USA; approximately 12 mm
about the sperm of this fish, with the 1995 study by deep [3]) and covered with an 18 mm  18 mm
Takai and Morisawa [12] being the only one to examine coverslip (VWR, West Chester, PA, USA) to provide
factors affecting sperm motility. The results of the a consistent fluid depth. Due to technical limitations in
current study substantially expand the database on the the activating procedure followed, the earliest time that
motility characteristics of zebrafish sperm. video could be reliably collected was 15 s after
activation. The slide and coverslip were coated with
2. Materials and methods 1% (w/v) polyvinyl alcohol to reduce sticking of sperm
[3]. Activations were conducted at room temperature
2.1. Collection and handling of biological samples (20  1 8C).

Adult zebrafish were obtained from Scientific 2.3. Construction of a video microscopy system for
Hatcheries (Scientific Hatcheries, Huntington Beach, use with fish
CA, USA) and maintained at the University of Idaho
zebrafish facility (Aquaneering, San Diego, CA, USA). The microscope used was an inverted Zeiss Axiovert
Fish were at least 9 months old at the time of sample 40, with a 0.4 numerical aperture condenser, and 0.25
collection. Fresh semen samples were collected by the numerical aperture 10 A-plan phase objective (Zeiss,
method of Westerfield [13]. Briefly, each male was Thornwood, NY, USA). To obtain the brightest possible
anesthetized using 170 mg/mL TRIS buffered tricaine image, and thus the fastest possible shutter speed, no
(pH 7.5), blotted dry, and rinsed in zebrafish sperm diffusion or contrast enhancing filters were used. The
 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672 663

video camera was coupled to the microscope through a by Nico Stuurman, http://valelab.ucsf.edu/nico/IJplu-
1 c-mount adapter using the 0/100% photoport of the gins/MTrack2.html) was used within the plugin to
microscope. A firewire Prosilica EC640 camera generate x, y coordinates for the centroid of each sperm
(Prosilica, Burnaby, BC, Canada) with a resolution of present throughout the frames analyzed and assemble
659 pixel  493 pixel at 97 frames/s (fps) was used. these coordinates into tracks for individual sperm.
Gain was turned on at a relative setting of 2 and the Objects larger than 40 pixels in area were likely to be
shutter speed set to 1/10,000 s to effectively freeze non-sperm or sperm in contact due to agglutination or
sperm motion in each image captured. Final resolution collision and so were not analyzed by the plugin. Sperm
obtained was 1.075 mm/pixel, with a field of view of that could not be tracked throughout all frames due to
708 mm  530 mm. collisions or movement out of view or plane of focus
Images were recorded directly to a RAID 0 array of were not analyzed by the plugin. The maximum search
7200 rpm hard drives using National Instruments radius for sperm path assembly was set at 8 pixels.
LabVIEW, IMAQ 1394 and Vision software (National Path coordinates generated for each sperm present
Instruments, Austin, TX, USA) on a computer with throughout all frames were then further analyzed by the
1 GB of RAM and a 3.4 GHz Hyper Threading Pentium plugin to describe the motion characteristics for later
IV processor (Intel, Santa Clara, CA, USA). Each frame determination of percent motility and of the average
was processed to a binary image by threshholding, such motion characteristics for all sperm in the sample. The
that the centers of all sperm (the brightest area produced sperm motility characteristics analyzed and their
by phase contrast due to the shape of the sperm) were abbreviations are given in Table 1. Excluding VSL
retained and other image data discarded; this was calculations, algorithms are consistent with those widely
performed by Vision software at a constant threshold reported by users of commercial CASA systems [3]. Only
setting of 0–50. Image sequences were written at 97 fps sperm that were determined by the plugin to be actively
from 15 to 95 s after the sample was activated. motile (as opposed to moving due to bulk flow, see below)
were included in calculations of average motility
2.4. Attributes of an ImageJ plugin for sperm parameters for the sample. The duration of motility
motility analysis was estimated as the rate at which motility decreased for
an individual analysis; specifically, it was estimated as
For each analysis, 97 frames, representing 1 s of the percent decrease of motility from 15 s post-activation
motion (verified by dividing the number of images to 55 s post-activation ((percent motility 15 s percent
saved by the length of time for which images were motility 55 s)/percent motility 15 s). Duration was
written), were imported into ImageJ and analyzed using estimated in this manner because measurements of
our plugin. ImageJ is an open source JAVA application duration based upon time until cessation of motion by all
distributed by the National Institutes of Health (NIH, sperm may be dependent upon the number of sperm
Bethesda, MD, USA) and is available at http:// viewed [7]. In addition, the plugin is able to output all x, y
rsb.info.nih.gov/ij/. A tracking algorithm from Mtrack2 coordinates for a single sperm and motion parameters for
(distributed as open source software for academic use each individual motile sperm.

Table 1
Sperm motility characteristics calculated by the ImageJ plugin
Percent motility Percent of tracked sperm identified by the plugin as exhibiting motility during the 1 s period of analysis
Velocity curvilinear (VCL) The total point to point distance traveled by the sperm over the time period analyzed averaged to a per
second value
Velocity average path (VAP) Velocity over an average path, generated by a roaming average of sperm position from one-sixth of the video’s
frame rate, such that each point is generated by averaging the coordinates of a set number of locations on the
VCL path (using 97 fps, the location of a sperm in 16 frames are averaged to generate one VAP point and the
first VAP point represents frames 1–16, the second point frame 2–17); averaged to a per second value
Velocity straight line (VSL) The maximum distance moved on the VAP path by the sperm from the first VAP point during in the video
segment analyzed, calculated to a per second value based on the number of frames for which VAP points
were calculated
Linearity (LIN) A measure of path curvature determined by dividing VSL by VAP
Straightness (STR) A measure of VCL side to side movement determined by dividing VAP by VCL
Duration estimate (Percent motility at 15 s post-activation percent motility at 55 s post-activation)/percent motility at 15 s
post-activation
664 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672

2.5. Verification of algorithms would be counted as motile whereas sperm moving due
to bulk flow would be counted as immotile and not used
Internal calculations of the program were verified by in future determination of the average motion char-
comparing manual analysis of single sperm with the acteristics of the sample. Determination of motility for a
results generated by the program. The x, y coordinates sperm was completed in two passes, such that slow
for 10 sperm from each of three males were analyzed moving sperm moving in a manner uncharacteristic of
using equations in a Microsoft Excel spreadsheet movement due to bulk flow would be identified as
(Microsoft, Redmond, WA, USA). Sperm moving in motile (Table 2). For each analysis performed, the
uniform circular, linear and non-circular, non-linear program displayed an image showing the paths of all
paths were chosen so that parameters could be sperm analyzed with motile sperm highlighted. This
compared to the visually apparent differences. Correct was used to verify that motile sperm were not being
tracking of sperm by the coordinates generated was counted as inactive and that inactive sperm were not
verified by comparison of projections of video (showing being included as motile by looking for the pattern of
the path of a sperm by superimposing its image from motion occurring with bulk flow. No samples were used
each frame) for three males with the image of paths where movement of inactive sperm due to bulk flow
generated automatically using the coordinates calcu- surpassed the second-pass minimum velocity require-
lated by the plugin. ments for detection of motile sperm.

2.6. Establishment of settings for non-motile sperm 2.7. Evaluation of the effect of sperm concentration
identification upon motility measurements

When using polyvinyl alcohol-coated slides, one key To estimate the effect of sperm concentration on
problem encountered in the analysis of zebrafish sperm motility analysis, sperm suspension from three indivi-
motility was associated with motion of inactive sperm duals was initially diluted with 5 mL ZSI; directly prior
due to bulk flow of fluid, characterized by the linear to use, 1 mL of this semen suspension was further
motion of many sperm in the same direction at the same diluted in ZSI to yield approximately 200, 100, 80, 50
velocity. This occurred directly after application of the and 30 sperm/video frame. The actual concentrations of
sample to the slide, often preventing analysis of motility sperm were not determined; this should not be relevant
during and shortly after the sperm had been activated, a to the ability of the system to assess motility. Instead,
period crucial to the analysis in light of the short the number of sperm visible, and thus the number of
duration of motility of these sperm. Characteristics of potential sperm–sperm collisions, was used as a relative
motion for a given sperm were used to determine if it concentration, as this may affect the measured motility
was actively motile or moving due to bulk flow. The parameters. The motility for each relative concentration
average characteristics of motion for six recordings was recorded in triplicate.
performed with inactive samples that exhibited bulk
flow (sperm diluted with immobilizing rather than 2.8. Determination of the impact of acquisition
activating solution) were compared to the character- frame rate upon measured fish sperm motility
istics calculated for slow moving sperm obtained by characteristics
examining motility at the end of the motile time period.
Based upon histograms of all motion parameters To assess the effect of frame rate upon the system’s
from both sperm populations analyzed, cut-off values ability to describe sperm motion, as well as potential
were selected so that the slow moving motile sperm pitfalls in comparing data captured at different frame
Table 2
Parameters used to determine whether a specific sperm is actively motile or moving due to bulk flow
First pass (a sperm must meet this set of conditions to be Minimum VSL of 3 mm/s, minimum VCL of 25 mm/s, minimum VAP of
considered motile) 20 mm/s, and a minimum STR of 2%
Second pass (a sperm must meet one of these sets of conditions Set A: maximum STR of 80%, maximum LIN of 80, no more than 25%
as well as the first pass conditions to be considered motile) of frame to frame VAP movements may be <5 mm/s, and no more than
1% of frame to frame VAP movements may be zero. Set B: minimum VCL
of 35 mm/s, and minimum VAP of 25 mm/s. Set C: maximum STR of 50%
or maximum LIN of 60%
Sets of parameters presented are used to evaluate the motility status of each individual sperm analyzed by the ImageJ plugin.
 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672 665

rates, motility characteristics obtained using video for 3. Results

three males at different frame rates were compared.
Video was collected three times for each male with the 3.1. Verification of parameters measured
camera set at 300, 150, 97, 60 or 30 fps. Frame rates
above 97 fps were generated by decreasing the video The coordinate locations determined by the program
field, and thus decreasing the number of sperm were verified for ten males by comparing a projection of
recorded. Due to the decrease in velocity and in STR the original video with a plot of the tracks generated by
recorded for slower frame rates, percent motility for the program; Fig. 1 illustrates one representative
frame rates below 97 fps was calculated by adjusting the comparison. Differences were evident in a limited
motility determination algorithm such that the only number of cases where sperm collided and one or both
specification used in determining motility for each paths were eliminated, or where sperm were eliminated
sperm was a VCL > 20 mm/s. from analysis because they swam out of the plane of
focus or out of the frame. In addition, the coordinates
2.9. Characterization of zebrafish sperm motility, generated by the program for VCL, VAP, and VSL were
discrimination among males, and determination of
the system’s repeatability

Samples from 13 male zebrafish were analyzed for

motility as indicated above. Motility analysis was
performed in triplicate for all but one male (two
replicates were completed). Relative sperm output for a
male was determined as the average number of sperm
present in the field of view during motility analysis and
reflects both the volume of semen obtained and the
concentration of sperm in the semen.

2.10. Statistical analysis

SAS version 9.1 (SAS Institute, Cary, NC, USA) was

used for all statistical analysis. Pearson correlations
coefficients were estimated using Proc Corr to determine
if sperm concentration or acquisition frame rate
influenced the ability to measure motility characteristics.
When a correlation was identified, Proc GLM was used to
further test the effect of acquisition frame rate with one-
way analysis of variance (ANOVA), followed by post-
hoc pair-wise comparisons using Fisher’s protected LSD.
One-way ANOVA was used (performed as above) to
test the effect of individual males on sperm character-
istics. Relationships among relative sperm output and
motility characteristics were examined using Pearson
correlation coefficients (determined as above). Differ-
ences in values for VCL, LIN and STR between 15 and
65 s post-activation were examined using one-way
ANOVA (performed as above). Using replicate values
of motion parameters for an individual male, the CV Fig. 1. (A) Projection of all video frames in 1 s of video taken at
(S.D. of replicates/mean value of replicates) for a given 97 fps. (B) Set of tracks generated by the plugin, plotted based upon
parameter and male was determined. Reproducibility of the calculated x, y coordinates from the video used in (A). In (B),
tracks indicated by an arrow have been identified as immotile by the
a motion parameter was described as the average CV
plugin, the calculated percent motility for this analysis was 88% based
and repeatability (determined using one-way ANOVA) on 90 tracks; tracks not seen in (B) were excluded by the plugin due to
for that parameter. movement out of the frame or focal plane or collisions such that they
Data are presented as mean  S.D. were not tracked in all 97 frames. The scale bar represents 100 mm.
666 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672

used to plot individual tracks and compare the visually 3.2. Evaluation of the effect of sperm concentration
evident character of these paths with the parameters on measurement of parameters
determined by the program (Fig. 2). The accuracy of the
plugin’s calculation of parameters of interest was Relative sperm concentrations from three males with
verified by comparison of spreadsheet-based computa- three replicates each were analyzed producing between
tion using track coordinates generated by the program to 6 and 344 sperm in the field of view. Within this range,
the computerized output; calculations based upon the there was no discernible relationship (linear or
coordinates were not significantly different between the otherwise) between any of the motility characteristics
plugin output and spreadsheet-based determinations and the adjusted relative concentration of sperm,
(Fig. 2, legend). determined as the number of sperm in the field of
view (r < 0.3 and P > 0.1 for all comparisons).

3.3. Determination of the influence of acquisition

frame rate upon measured fish sperm motility
characteristics

Samples from three males were analyzed in

triplicate. Between 30 and 250 tracks were analyzed
at 30, 60 and 97 fps. Between 8 and 60 tracks were
analyzed at 150 and 300 fps rates due to the decreased

Fig. 2. Tracks represent sperm motion over 1 s, taken 15 s post-

activation. Individual tracks were chosen to demonstrate the soft-
ware’s ability to quantify motion parameters for sperm moving in a
variety of manners: linear, circular and erratic. The black path
represents VCL points utilized by the program, dashed path represents
the VAP path as calculated by the program, diamonds on the VAP path
represent the points used in calculating VSL. For motion parameter
measurements, the first value for a parameter was that output by the
plugin and the second was determined with spreadsheet-based calcu-
lations: (A) VCL 103, 103 mm/s; VAP 92, 92 mm/s; VSL 90, 89 mm/s; Fig. 3. Average VCL (A) and average STR (B) for sperm as measured
LIN 98, 97%; STR 89, 89%. (B) VCL 111, 111 mm/s; VAP 82, 82 mm/ at different frame rates (n = 3, with three replicates each). VCL
s; VSL 25, 25 mm/s; LIN 31, 31%; STR 74, 74%. (C) VCL 104, (r = 0.93, P < 0.0001) and STR (r = 0.91, P < 0.0001) correlated
104 mm/s; VAP 62, 62 mm/s; VSL 53, 53 mm/s; LIN 85, 85%; STR 60, strongly with frame rate. Values without a common letter differed
60%. (P < 0.05).
 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672 667

field of view necessary for increasing frequency above

the camera’s native specifications. Both VCL and STR
were correlated with frame rate (Fig. 3, r = 0.93,
P < 0.0001 and r = 0.91, P < 0.0001, respectively)
and significant differences were identified among all
frame rates. Percent motility, VSL, VAP and LIN had no
visible relationship (linear or otherwise) with frame rate
(r < 0.15 and P < 0.5 for all comparisons).

3.4. Characterization of zebrafish sperm motility

and the system’s reproducibility

Significant differences were detected among males

for all parameters measured. Except for duration, the
average CV for all measures of sperm motion was
10% and the repeatability was 0.6 (Table 3).
Average percent motility was strongly correlated with
relative sperm output, defined as the number of sperm
visible in the field of view and reflecting both the
volume of semen obtained and sperm concentration in
the semen (r = 0.83, P = 0.004). There were no
significant correlations between any motility character-
istic and relative sperm output (r < 0.5 and P > 0.05).
The decline in percent motility was determined relative
to time post-activation (Fig. 4). Very few sperm samples
were motile beyond 65 s post-activation and, as such,
velocity and path curvature parameters could only be Fig. 4. (A) The time course of sperm percent motility once activated,
measured up to this time point (Fig. 5). There was no n = 13 with at least two replicates for each male. (B) The time course
of sperm motility once activated, expressed for individual males
significant difference between initial and final LIN.
(n = 13) as a percentage of value at 15 s. The initial time point,
However, both VCL and STR decreased significantly, 15 s post-activation, represented the earliest reliable time point that
relative to their initial values. could be obtained (due to limitations in activation procedure).

4. Discussion 7,18–21]. A 1990 comparison of two commercial

systems using the same video of sperm motility with
Although CASA systems represent the most quanti- comparable settings led to the conclusion that the
tative and comprehensive measurement of sperm systems produce results differing by up to 30% for a
motility currently available, their utilization appears given parameter [22]. In contrast, a similar 1992
to be reduced by high costs; this may have resulted in comparison found no clinically relevant differences
manual data acquisition by many investigators [e.g., [23]. Considering that a previous investigation found

Table 3
Motility characteristics of zebrafish sperm
Relative Percent VCL VAP VSL LIN STR Duration
sperm output motility (mm/s) (mm/s) (mm/s) (%) (%) (%)
Mean 87  33 90  6 104  9 77  15 66  17 84  10 74  10 58  14
CV (%) 36 9 5 7 10 5 6 22
Repeatability 0.39 0.6 0.61 0.87 0.89 0.85 0.83 0.14
Values presented as mean  S.D. for 13 males, analyzed at 15 s post-activation; data represent motion parameters as calculated by CASA as well as
an estimate of variation among replicates. Repeatability and CV determined based upon three replicates for 12 males and two replicates for a single
male. Percent duration was determined as ((percent motility 15 s percent motility 55 s)/percent motility 15 s). Relative sperm output was defined
as the number of sperm in the field of view, reflecting the concentration of sperm once diluted in activating solution (intermale variation for this
parameter is the result of variation in both semen volume and sperm concentration in the semen).
668 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672

a sample [9,24], appropriate quantification of para-

meters describing such motion for fish sperm was
uniquely important. The current study has established
and verified the validity of algorithms for VCL, VAP,
VSL, LIN and STR. Attempts to analyze beat cross
frequency (BCF) as the frequency of the sperm head
crossing the average path in either direction [25]
generated errors due to the highly circular nature of
some sperm paths. In cases of a tight circular track, the
average path was occasionally located within the
curvilinear path, resulting in very few crosses; this
error was identified for a number of individual sperm
(Fig. 2), as well as in the data obtained by Ravinder et al.
Fig. 5. Values for velocity (VCL, squares) and path curvature (LIN, [26] using a commercial system. Because of erroneous
triangles; STR, diamonds), expressed as a percentage of the value at calculations when analyzing sperm with circular paths,
15 s post-activation vs. time post-activation. VCL and STR, but not BCF may not be an accurate descriptor of fish sperm
LIN, decreased from 15 to 65 s (P < 0.05). motility. Measurements for the amplitude of lateral
head displacement based on the distance of points on
drastic variation among commercial systems, no study the curvilinear path from the average path [27] have not
to date has compared multiple systems using identical been included as such a calculation would be similarly
semen samples and as no comparison of modern skewed.
systems exists, the reproducibility among different Percent motility, VCL, VAP, STR and LIN were
systems may be suspect. calculated as described by investigators using com-
In this study, we have written free CASA software mercial systems [3,25,28]. In place of common VSL
which should enable standardization of data among calculations based on the distance from a sperm’s first
laboratories due to its accessibility, described the location to its last, our algorithm utilized the maximum
components employed and verified the validity and distance traveled from the first VAP point during the
biological relevance of the system for fish sperm time period analyzed (Fig. 2). Therefore, our VSL
analysis. The source code and complete documentation calculation was not subject to errors in which a sperm
are available at http://rsb.info.nih.gov/ij/plugins/ with a circular path traveled back to its origin over 1 s
casa.html. and was thus attributed a near zero VSL measurement
(Fig. 2B).
4.1. Validity of measurements and algorithm
capability for describing motion 4.2. Verification of motility determination
algorithm
The CASA-generated tracks were verified in terms of
the system’s ability to accurately track sperm and to Our algorithm was visually verified by examining
calculate motility parameters. As expected for all every analysis for incorrect motility determination
CASA systems, errors due to collisions of sperm were using the program’s color-coded differentiation of
identified [11]. Similar to commercial systems, the motility status for all sperm tracked. Parameters for
prevalence of these errors has been minimized by identification of motile sperm can be modified by the
excluding any track where the sperm was larger than our end user to best characterize motility for a given
maximum size. Analysis of a frame where two sperm species. Furthermore, we have verified the plugin’s
touched will result in the exclusion of the tracks of both ability to identify sperm moving due to bulk flow, even
sperm. Cases where sperm collide and do not present an in cases where the velocity of these sperm exceeds that
image greater than the maximum size result in inclusion of the minimum velocity for an actively motile sperm.
of one path and possibly erroneous tracking of a sperm Previous work [29] used velocity cut-off values as low
by combination of two paths. We have found no means as 5 mm/s. Such values are probably useful only when
of automatically detecting such errors and they will using a slide that has not been coated with an effective
exist in data as noise. non-sticking agent; our observations suggested that
Because of the potential for specific activating inactive sperm applied to coated slides usually moved
conditions to induce highly circular paths in all sperm of more freely in response to bulk flow.
Table 4
Motility parameters of zebrafish sperm in the current study compared to those of sperm from a variety of other fish species obtained by CASA or semi-CASA systems
Parametersa % Motile VCL (mm/s) VAP (mm/s) VSL (mm/s) LIN (%) Study

J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672

Freshwater species
Zebrafish, Danio rerio 97 fps, 20 8C, 15 s, 1:21, TAP, n = 13, present 83  13 104  10 77  15 66  18 84  11 Present
Lake sturgeon, Acipenser 200 fps, 10 8C, 5 s, 1:750, saline, n = 5, CellTrack/S 56  22 217  68 62  14 33  15 [27]
fulvescens
Danube bleak, Chalcalburnus 32 fps, 15 8C, 10 s, 1:500–1000, saline, n = 10, 70  27 86  17 [29]
chalcoides Stroemberg Mika cell motility analysis program
Carp, Cyprinus carpio 50 fps, ?, 15 s, 1:50, distilled water, n = 11, 98  2 139  14 [36]
Olympus MicroImage software with special macro
Carp, Cyprinus carpiob 200 fps, 23–25 8C, 15–20 s, 1:20, saline, n = 10, 60 145 60 [37]
Cell Track/S CASA system
Bluegill, Lepomis macrochirus ?, 20 8C, 5 s, ?, lake water, n = 22, HTM-CEROS v. 12 60 195 [8]
Rainbow trout, 32 fps, 4 8C, 10 s, 1:50, saline, n = 3, Stroemberg 66  6 91  21 [38]
Oncorhynchus mykiss Mika cell motility analysis program
Artic charr, Salvelinus aplinus 50 fps, 5–7 8C, 15 s, 1:37.5, water, n = 45, 90 117 [39]
HTM-CEROS v. 12
Marine species
European eel, Anguilla anguilla ?, ?, ?, 1:1001, seawater, n = 29, 40 40 16 12 [40]
sperm class analyzer
Halibut, Hippoglossus 50 fps, 8 8C, 30 s, 1:37.5, seawater, 83  7 145  12 110  17 76  4 [30]
hippoglossusb n = 5, HTM-CEROS v. 12
Turbot, Psetta maxima 50 fps, 20 8C, 10 s, 1:25, artificial seawater, 85–90 185–215 [41,42]
n = 6–15, Modcell computer-assisted sperm analysis
a
Parameters are: data acquisition rate in fps, temperature of activation, time after activation that data collection was initiated, dilution of semen or sperm suspension with activating solution,
activation solution, replicates, system used to collect data. Question marks represent unknown conditions in the original studies.
b
Data show seasonal variation; early season data reported in the table.

669
670 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672

4.3. Evaluation of the effect of sperm concentration 4.5. Characterization of sperm motility parameters
on measurement of parameters among males and of the system’s reproducibility

When the number of sperm analyzed ranged from 6 The average CV was 10% for all parameters
to 344 in the visual field, there was no effect on percent measured (except duration). This degree of reproduci-
motility or motion parameters. Rijsselaere et al. [25] bility was comparable to that obtained using a
demonstrated an effect of dog sperm concentration on commercial system for analysis of dog sperm [31].
motility analysis using a commercial system; we have Repeatability was 0.85 for VAP, VSL, LIN and STR
no means of comparing results, as the average number and 0.6 for percent motility and VCL, indicating that
of sperm analyzed when using different sperm these measures were reproducible.
concentrations was not reported in their study. The Average values of velocity and path curvature were
absence of a correlation between the density of sperm in obtained for zebrafish sperm. Unlike salmonid sperm
the field of view and motility determinations agreed [9,24], zebrafish sperm had no significant change in
with the findings of Babiak et al. [30] using halibut path curvature (LIN) over time post-activation. This
sperm; these findings may hold for fish sperm in was in accordance with the effect of time on carp sperm
general. Nonetheless, it is likely that at very high path noted by Ravinder et al. [26] and apparent in the
concentrations, sperm concentration will have an effect data of Lahnsteiner et al. [29] for Chalcalburnus
on measured motility characteristics due to an increase chalcoides and may represent a difference between
in the frequency of collisions. cyprinids and salmonids in reproductive strategies. Both
VCL and STR decreased with time following activa-
4.4. Determination of the impact of acquisition tion; the decrease in velocity agreed with but the
frame rate upon measured fish sperm motility decrease in STR contrasted with that found for carp
characteristics [26]. The lower frame rate used in the carp study may
have been unable to detect a change in STR, or for carp
Previous studies using carp [26], dog [25] and human sperm, this value may not vary relative to the time post-
[11] sperm suggested that increases in frame rate activation.
drastically increased measured VCL without substantial In addition to providing the first description of
impact on VAP, resulting in a decrease in STR. zebrafish sperm motility, the current study demon-
Increases in frame rate are expected to result in greater strated our system’s capability to distinguish differences
visualization of sperm motion; the optimal frame rate among males in terms of every variable analyzed.
would generate enough locations in each track to Although semen volume and concentration were not
describe the angle of curvature and magnitude for all determined directly, our data suggested that relative
deviations relative to the average path. The minimum sperm output among males corresponded to percent
frame rate with optimal descriptive capability could be motility in zebrafish, i.e. a male with low sperm output
identified as the rate beyond which increases in frame usually had a low percent motility. The basis for this
rate result in nominal increases in VCL. Toth et al. [27] observation was not clear.
suggested that frame rates >60 fps should be used when Zebrafish sperm motility characteristics determined
analyzing fish sperm. As in previous studies, measured in this study compared favorably with those obtained
VCL in the present work increased with acquisition for different species by other laboratories using CASA
frame rate, indicating that comparison across systems systems (Table 4). Nonetheless, it was not possible to
using different frame rates may not be valid and that rigorously compare these data among laboratories, as
systems operating at 60 or 30 fps acquire little of the there was little consistency among studies in the
side-to-side movement that can be visualized in the path conditions and methodologies used. Data from studies
of fish sperm. The trend identified for zebrafish in the employing the Hobson sperm tracker were not included
present work suggested that frame rate and VCL were in Table 4; this method of sperm analysis averages
linearly related between 30 and 300 fps, and thus we are motility parameters over a 15 s period and thus lacks the
hesitant to suggest that a frequency of even 300 fps fully temporal resolution of other systems [e.g., 32–34].
described the path of zebrafish sperm. The demon-
strated trend towards increased descriptive capability 5. Conclusion
relative to frame rate indicates that use of the highest
commonly available frame rate (97 fps) is preferable to Our free CASA software represents an alternative to
lower rates. high cost commercial systems and is fully capable of
 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672 671

rapidly generating reproducible and quantitative mea- [6] Rurangwa E, Roelants I, Huyskens G, Ebrahimi M, Kime DE,
surements for fish sperm. Like commercial systems, a Ollevier F. The minimum effective spermatozoa: egg ratio for
artificial insemination and the effects of mercury on sperm
graphic user interface is employed; sperm identification motility and fertilization ability in Clarias gariepinus. J Fish
and tracking are automated. Because this software is Biol 1998;53:402–13.
open source, independent investigation and validation [7] Casselman SJ, Montgomerie R. Sperm traits in relation to male
of the algorithms used is possible, potentially increasing quality in colonial spawning bluegill. J Fish Biol 2003;64:1700–
11.
quality control for data gathering [35]. Replication of
[8] Burness G, Casselman SJ, Schulte-Hostedde AI, Moyes CD,
our system should be possible by any investigator, as Montgomerie R. Sperm swimming speed and energetics vary
should modification to fit a particular need (including with sperm competition risk in bluegill (Lepomis macrochirus).
analysis of sperm of other taxa), increasing the Behav Ecol Sociobiol 2004;56:65–70.
versatility of the system as a whole. With these [9] Boitano S, Omoto CK. Trout sperm swimming patterns and role
qualities in mind, this study should increase availability of intracellular Ca2+. Cell Mot Cytoskel 1992;21:74–82.
[10] Cooper TG, Yeung CH. Computer-aided evaluation of assess-
of CASA systems to investigators and represent a ment of ‘‘grade a’’ spermatozoa by experienced technicians.
contribution to future attempts to understand the Fertil Steril 2006;85:220–4.
influence of motility characteristics upon sperm [11] Verstegen J, Iguer-Ouada M, Onclin K. Computer assisted semen
fertilizing capacity. analyzers in andrology research and veterinary practice. Ther-
iogenology 2002;57:149–79.
[12] Takai H, Morisawa M. Change in intracellular K+ concentration
Acknowledgements caused by external osmolality change regulates sperm motility of
marine and freshwater teleosts. J Cell Sci 1995;108:1175–81.
We thank Gail Leedy and Morgan Wilson for [13] Westerfield M. The zebrafish book. A guide for the laboratory
insightful comments/critical review of this manuscript use of zebrafish (Danio rerio), 4th ed., Eugene: University of
Oregon Press; 2000.
and Robert Drew and Barrie Robison for assistance with [14] Krasznai Z, Marian T, Izumi H, Damjanovich S, Balkay L, Tron
statistics. This study was supported in part by the L, et al. Membrane hyperpolarization removes the inactivation of
National Research Initiative Competitive Grants pro- Ca2+ channels, leading to Ca2+ influx and subsequent initiation
gram (grant number IDAR-2005-01500) from the of sperm motility in the common carp. Proc Natl Acad Sci USA
USDA Cooperative State Research, Education, and 2000;97:2052–7.
[15] Lahnsteiner F, Mansour N, Berger B. Seminal plasma proteins
Extension Service and in part by the Columbia River prolong the viability of rainbow trout (Oncorhynchus mykiss)
Inter-Tribal Fish Commission, through contract with the spermatozoa. Theriogenology 2004;62:801–8.
Bonneville Power Administration as part of the [16] Billard R, Cosson MP. Some problems related to the assessment
Northwest Power and Conservation Council’s Colum- of sperm motility in freshwater fish. J Exp Zool 1992;261:122–
bia Basin Fish and Wildlife Program (BPA project no. 31.
[17] Woolsey J, Holcomb M, Cloud JG, Ingermann RL. Sperm
2001-017-00). motility in the steelhead Oncorhynchus mykiss (Walbaum):
influence of the composition of the incubation and activation
References media. Aquacult Res 2006;37:215–23.
[18] Hoysak DJ, Liley NR. Fertilization dynamics in sockeye salmon
[1] Billard R, Cosson J, Crim LW, Suquet M. Sperm physiology and and a comparison of sperm from alternative male phenotypes. J
quality. In: Bromage NR, Roberts RJ, editors. Broodstock Fish Biol 2001;58:1286–300.
[19] Liley NR, Tamkee P, Tsai R, Hoysak DJ. Fertilization dynamics
management and egg and larval quality. Cambridge: Blackwell
Science; 1995. p. 25–52. in rainbow trout (Oncorhynchus mykiss): effect of male age,
[2] Lahnsteiner F, Berger B, Weisman T, Patzner RA. Determination social experience, and sperm concentration and motility on in
of semen quality of the rainbow trout, Oncorhynchus mykiss, by vitro fertilization. Can J Fish Aquat Sci 2002;59:144–52.
[20] Vladic TV, Afzelius BA, Bronnikov GE. Sperm quality as
sperm motility, seminal plasma parameters and spermatozoal
metabolism. Aquaculture 1998;163:163–81. reflected through morphology in salmon alternative life histories.
[3] Kime DE, Van Look KJW, McAllister BG, Huyskens G, Rur- Biol Reprod 2002;66:98–105.
[21] Morita M, Fujinoki M, Okuno M. K+-independent initiation of
angwa E, Ollevier F. Computer-assisted sperm analysis (CASA)
as a tool for monitoring sperm quality in fish. Comp Biochem motility in chum salmon sperm treated with an organic alcohol,
Physiol C 2001;130:425–33. glycerol. J Exp Biol 2005;208:4549–56.
[4] Rurangwa E, Kime DE, Ollevier F, Nash JP. The measurement of [22] Jasko DJ, Lein DH, Foote RH. A comparison of two computer-
automated semen analysis instruments for the evaluation of
sperm motility and factors affecting sperm quality in cultured
fish. Aquaculture 2004;234:1–28. sperm motion characteristics in the stallion. J Androl
[5] Rurangwa E, Volckaert FAM, Huyskens G, Kime DE, Ollevier F. 1990;11:453–9.
Quality control of refrigerated and cryopreserved semen using [23] Davis RO, Katz DF. Standardization and comparability of CASA
instruments. J Androl 1992;13:81–6.
computer-assisted sperm analysis (CASA), viable staining and
standardized fertilization in African catfish (Clarias gariepinus). [24] Cosson M-P, Cosson J, Billard R. Synchronous triggering of
Theriogenology 2001;55:751–69. trout sperm is followed by an invariable set sequence of move-
672 J.G. Wilson-Leedy, R.L. Ingermann / Theriogenology 67 (2007) 661–672

ment parameters whatever the incubation medium. Cell Mot [34] Le Comber SC, Faulkes CG, Van Look KJW, Holt WV, Smith C.
Cytoskel 1991;20:55–68. Recovery of sperm activity after osmotic shock in the three-
[25] Rijsselaere T, Van Soom A, Maes D, de Kruif A. Effect of spined stickleback: implications for pre-oviposition ejaculation.
technical settings on canine semen motility parameters measured Behaviour 2004;141:1555–69.
by the Hamilton-Thorne analyzer. Theriogenology 2003;60: [35] Martens L, Monsieur G, Ampe C, Gevaert K, Vandekerckhove J.
1553–68. Cell motility: a cross-platform, open source application for the
[26] Ravinder K, Nasaruddin K, Majumdar C, Shivaji S. Computer- study of cell motion paths. BMC Bioinf 2006;7:289.
ized analysis of motility, motility patterns and motility para- [36] Linhart O, Rodina M, Gela D, Kocour M, Vandeputte M.
meters of spermatozoa of carp following short-term storage of Spermatozoal competition in common carp (Cyprinus carpio):
semen. J Fish Biol 1997;50:1309–28. what is the primary determinant of competition success? Repro-
[27] Toth GP, Ciereszko A, Christ SA, Dabrowski K. Objective duction 2005;130:705–11.
analysis of sperm motility in the lake sturgeon, Acipenser [37] Christ SA, Toth GP, McCarthy HW, Torsella JA, Smith MK.
fulvescens: activation and inhibition conditions. Aquaculture Monthly variation in sperm motility in common carp assessed
1997;154:337–48. using computer-assisted sperm analysis (CASA). J Fish Biol
[28] Kato M, Fukunishi K, Ikegawa S, Higuchi H, Sato M, Horimoto 1996;48:1210–22.
M, et al. Overview of studies on rat sperm motion analysis using [38] Lahnsteiner F, Berger B, Grubinger F, Weismann T. The effect of
a Hamilton–Thorne sperm analyser—collaborative working 4-nonylphenol on semen quality, viability of gametes, fertiliza-
study. J Toxicol Sci 2001;26:285–97. tion success, and embryo and larvae survival in rainbow trout
[29] Lahnsteiner F, Berger B, Weismann T. Sperm metabolism of the (Oncorhynchus mykiss). Aquat Toxicol 2005;71:297–306.
teleost fishes Chalcalburnus chalcoides and Oncorhynchus [39] Rudolfsen G, Figenschou L, Folstad I, Tveiten H, Figenschou M.
mykiss and its relation to motility and viability. J Exp Zool Rapid adjustment of sperm characteristics in relation to social
1999;284:454–65. status. Proc R Soc Br 2006;273:325–32.
[30] Babiak I, Ottesen O, Rudolfsen G, Johnsen S. Quantitative [40] Asturiano JF, Perez L, Garzon DL, Marco-Jimenez F, Penaranda
characteristics of Atlantic halibut, Hippoglossus hippoglossus DS, Vicente JS, et al. Physio-chemical characteristics of seminal
L., semen throughout the reproductive season. Theriogenology plasma and development of media and methods for the cryo-
2006;65:1587–604. preservation of European eel sperm. Fish Physiol Biochem
[31] Iguer-ouada M, Verstegen JP. Evaluation of the ‘‘Hamilton 2004;30:283–93.
Thorn computer-based automated system’’ for dog semen ana- [41] Dreanno C, Suquet M, Quemener L, Cosson J, Fierville F,
lysis. Theriogenology 2001;55:733–49. Normant Y, et al. Cryopreservation of turbot (Scophthalmus
[32] Kime DE, Tveiten H. Unusual motility characteristics of sperm maximus) spermatozoa. Theriogenology 1997;48:589–603.
of the spotted wolffish. J Fish Biol 2002;61:1549–59. [42] Dreanno C, Cosson J, Suquet M, Seguin F, Dorange G, Billard R.
[33] Elofsson H, Mcallister BG, Kime DE, Mayer I, Borg B. Long Nucleotide content, oxydative phosphorylation, morphology,
lasting stickleback sperm; is ovarian fluid a key to success in and fertilizing capacity of turbot (Psetta maxima) spermatozoa
fresh water? J Fish Biol 2003;63:240–53. during the motility period. Mol Reprod Dev 1999;53:230–43.