Human Reproduction vol.13 no.3 pp.611–619, 1998
Factors influencing human sperm kinematic
measurements by the Celltrak computer-assisted sperm
analysis system
Michel Kraemer1, Corinne Fillion1,
Brigitte Martin-Pont1 and Jacques Auger2,3
1Laboratoire de Biologie de la Reproduction, Hôpital Jean Verdier,
Bondy and 2Service d’Histologie-Embryologie Biologie de la
Reproduction, CHU Cochin, Paris, France
3To
whom correspondence should be addressed at:
Service d’Histologie-Embryologie Biologie de la Reproduction/
CECOS, Pavillon Cassini, Groupe Hospitalier Cochin Port Royal,
123 Bd de Port Royal, 75679 Paris Cedex 14, France
This study examines the effect of varying several
factors, both extrinsic and technology-dependent, on
the reconstruction of human sperm trajectories and the
derived kinematic measurements using videotapes and the
Motion Analysis Celltrak/S instrument. In semen samples
from normal healthy men, curvilinear (VCL) and straight
line velocities (VSL) were found to increase 1.5-fold, and
linearity (LIN) of trajectories and amplitude of lateral
head displacement (ALH) increased 1.25-fold when the
temperature of analysis was raised from 24 to 37°C. Only
VCL and VSL were found to increase significantly between
24 and 37°C for sperm samples selected by Percoll gradient
and incubated in a capacitating medium. An analysis
chamber of 20 µm depth was found to be adequate for
seminal sperm samples while for Percoll-selected sperm
samples the analysis in a 50 µm depth provided the highest
proportions of spermatozoa with the highest VCL and the
largest ALH. The grey level detection threshold required
careful adjustment: using a threshold lower than the
optimal threshold produced spurious sperm trajectories
for seminal sperm samples and rejected some trajectories
for Percoll-selected sperm samples. Definition of the appropriate frame rate and maximum burst speed was critical
for valid trajectory reconstruction and therefore adequate
derived kinematic measurements. Optimal values of these
parameters were found to be 30 Hz and 400 µm/s for
seminal spermatozoa and 60 Hz and 700 µm/s for selected
spermatozoa. The optimal values of ‘ALH path-smoothing
factor’ used to calculate average path and ALH were 5–
10 points for seminal spermatozoa analysed at 30 Hz and
15–20 points for selected spermatozoa analysed at 60 Hz.
We propose a set of standard conditions for reliable
kinematic analysis of human spermatozoa using the
Celltrak/S system.
Key words: computer-assisted sperm analysis/sperm kinematic
measurements
© European Society for Human Reproduction and Embryology
Introduction
Computer-assisted sperm analysis (CASA) makes possible
accurate and precise analysis of various parameters depicting
the pattern of sperm kinematic movement (Boyers et al., 1989;
Holt et al., 1994). A number of reports have demonstrated
the diagnostic and prognostic value of kinematic parameters
(Aitken et al., 1985; Feneux et al., 1985; Jeulin et al., 1986;
Barratt et al., 1993; Aitken et al., 1994; MacLeod and Irvine,
1995). CASA is valuable for monitoring subtle changes in
sperm motion under various experimental conditions (Auger
et al., 1993; Skiba-Lahiani et al., 1995; Gandini et al., 1997;
Mahadevan et al., 1997). CASA machines may also be of
value in the analysis of human sperm morphology (Davis
et al., 1992b).
However, CASA instruments are not ‘ready-to-use’ robots,
and the reliability of CASA depends largely on the expertise
and training of the user (Comhaire et al., 1992; Holt et al.,
1994). Unfortunately, this widely used technology is still
accepted uncritically by many of its users. Therefore, it is
important to identify all the factors affecting the results of an
analysis and to define standardized procedures to facilitate the
wider use and value of CASA (Davis and Katz, 1993).
Some of the problems associated with CASA machines arise
from fundamental limitations of this technology (Boyers et al.,
1989). Others arise from the non-optimization of semen
preparation and/or calibration of the instrument. The effect of
the depth of preparation on sperm kinematics is critical for
capacitating/hyperactivated spermatozoa due to an enlarged
flagellar curvature which in turn induces larger displacements
of the sperm head in the z-direction (Mortimer et al., 1995).
The machine setting requires careful optimization (Mortimer
et al., 1995). CASA systems are based on similar principles,
but they differ in terms of the optics used and the software
for sperm identification and trajectory reconstruction. Moreover, the set-ups recommended by the manufacturers must be
tested to evaluate their efficiency in the full range of clinical
uses given the considerable heterogeneity of, and rapid changes
in, human sperm motion.
Such detailed testing was not done for the Motion Analysis
Celltrak instrument first described by Katz et al. (1985) and
Katz and Davis (1987). A few studies of human spermatozoa
using this system have recently been reported (Moohan et al.,
1993; Holt et al. 1994; Paston et al., 1994; Hurowitz et al.,
1995).
In the present study, we investigated the effect of varying
several extrinsic and technology-dependent factors on trajectory reconstruction and derived kinematic measures to determine the optimal conditions for use of the Motion Analysis
611
M.Kraemer et al.
Celltrak instrument in human sperm kinematic studies. The
extrinsic factors studied were the temperature of analysis and
the depth of the sperm preparation. The technology-dependent
factors tested were the detection threshold, the frame rate, the
maximum burst speed and the ‘amplitude of lateral head
(ALH) path-smoothing factor’.
Materials and methods
Semen samples
Semen samples were collected from normal healthy men by masturbation at the laboratory after sexual abstinence for 3–5 days. The men
were referred for semen analysis before in-vitro fertilization (IVF) of
their partner because of tubal obstruction. Samples were allowed to
liquefy for 1 h at 37°C before routine semen analysis according to
World Health Organization recommendations (WHO, 1992). All
semen samples used in this study were normal according to WHO
reference values (WHO, 1992). We used only semen samples with a
concentration of 20–603106/ml as appropriate sperm tracking by
CASA is concentration-dependent (Davis and Katz, 1993).
Configuration of the CASA instrument
A Nikon Optiphot-2 microscope (Tokyo, Japan) with a heated stage
(Motion Analysis Corp., Santa Rosa, CA, USA) was used, with a
phase contrast condenser, a 310 objective and a 36.7 photo eyepiece.
A high resolution video camera (NEC Corporation, Tokyo, Japan)
fed the video signal to a Panasonic AG6300 video recorder (Panasonic
Corp., Secaucus, NJ, USA) for processing using the Motion Analysis
CellTrak/S machine (CTS; VP 110 video processor; software version
3.20) (Motion Analysis Corp., Santa Rosa, CA, USA). The CTS
machine uses a fixed-length, weighted, running average to calculate
the average path (Boyers et al., 1989). The shape of the running
average is a cosine taper, and its width, the ALH path-smoothing factor
(APSF), can be set by the user. Other principles and specifications of
the hardware and software of the CTS instrument are given elsewhere
(Katz and Davis, 1987; Boyers et al., 1989).
Experimental design and parameter setting
The effect of various temperatures (24, 28, 32 and 37°C) on the
movement characteristics was tested on matched seminal samples
and selected sperm samples. The selected samples were prepared
using centrifugation on simplified Percoll density gradients (WHO,
1992) followed by incubation for 30 min in Ménézo’s B2 capacitating
medium (CCD, Paris, France). All the samples were videotaped in a
20 µm disposable microcell chamber (Conception Technologies, La
Jolla, CA, USA) and the subsequent CTS analyses were carried out
on at least 200 motile spermatozoa per sample. For these experiments,
the following standard kinematic parameters (WHO, 1992) were
calculated: curvilinear velocity (VCL; µm/s), straight line velocity
(VSL; µm/s), linearity of trajectory (LIN; %) and amplitude of lateral
head displacement (ALH; µm). Two different set-ups of the CTS
instrument previously shown to be optimal (data not shown), were
used as follows: for semen samples, frame rate, 30 Hz; duration of
data capture, 15 frames; minimum motile speed, 10 µm/s; maximum
burst speed, 400 µm/s; distance scale factor, 0.96; ALH pathsmoothing factor, 5 points; centroid3search neighbourhood (singleedge mode), 2 pixels; cell size, min–max, 1–10 pixels; path maximum
interpolation, zero frames and path prediction percentage, 0; for
selected sperm samples, frame rate, 60 Hz; duration of data capture,
30 frames; minimum motile speed, 8 µm/s; maximum burst speed,
700 µm/s; distance scale factor, 0.96; ALH path-smoothing factor,
15 points; centroid3search neighbourhood (single-edge mode), 2
612
pixels; cell size, min–max, 1–10 pixels; path maximum interpolation,
one frame, and path prediction percentage, 0.
The effect of the depth of preparation on sperm kinematics
was tested on sperm preparations of 12, 20 (disposable microcell
chambers) and 50 µm depth (rectangular capillary tubes; VitroDynamics, Rockaway, NJ, USA) for seminal and selected sperm
samples. Selected sperm samples were prepared as described above,
and seminal and selected sperm samples were videotaped at 37°C
and the subsequent CTS analyses were done on at least 200 motile
spermatozoa per sample. Moreover, for these samples, the sperm
concentration and percentage of motile spermatozoa were also
measured at the same time using the CTS instrument. The two setups were tested at all the temperatures mentioned above.
We investigated the effect of varying the manual detection threshold
on seminal sperm samples and selected sperm samples videotaped at
37°C in a 20 µm disposable microcell chamber. The grey level
threshold for the detection of spermatozoa (i.e. the optimal grey level
threshold set to detect all the spermatozoa present in the microscopic
field, and no debris) was optimized after determination of the best
illumination conditions (i.e. the illumination that maximized the
contrast of the sperm cells to the background) by ensuring that the
concentration measured by the CTS was similar to the concentration
measured by haemocytometry (Davis et al., 1992a). Then we studied
the effect on kinematic parameters of small changes in the detection
threshold (650 grey levels for a total of 2000 grey levels). Each
analysis was carried out from the same videotaped microscopic field
using the counter of the video recorder. The two set-ups described
for the temperature experiments were used. Sperm concentrations
and percentages of motile spermatozoa were measured with the CTS.
The effects on kinematic parameters of varying the maximum burst
speed and the ALH path-smoothing factor were studied using the setup reported above for the other parameters.
The maximum burst speed was set at 200, 400 and 700 µm/s for
seminal spermatozoa analysed at 30 Hz and at 400, 500, 600 and
700 µm/s for selected sperm samples analysed at 60 Hz. The same
fields for each sample tested were studied using the counter of the
recorder. For each experiment, the paths were reconstructed by the
pathfinder algorithm of the CTS software and printed. As the video
recorder did not have a time/base generator to automate this operation,
there was sometimes a shift of one or two frames for the starting
point of the sperm path.
APSF in the range 5–15 points and 5–30 points were used
respectively for seminal spermatozoa analysed at 30 Hz and selected
sperm samples analysed at 60 Hz. The effect of smoothing of the
curvilinear path to obtain an average path was studied on the same
trajectories with the aid of the ExpertVision software facility (Motion
Analysis Corp., Santa Rosa, CA, USA).
Statistical analysis
Mean, SD and distribution frequency of the kinematic parameters
were calculated and presented in tables. Differences in the distribution
profiles of the kinematic parameters for the various experimental
conditions were tested using the non-parametric Mann–Whitney test.
Results
Temperature
For seminal spermatozoa, the values of all kinematic characteristics increased significantly with increasing temperature
whereas for seminal spermatozoa, only VCL and VSL increased
significantly (Figure 1). On average, VCL and VSL of the
semen samples increased 1.5-fold and LIN and ALH increased
Factors influencing CASA
Table I. Typical effect of the depth of the preparation on sperm
concentration (CON), percentage of motile spermatozoa (MOT) and mean,
SD and distribution frequency (%) of values for kinematic characteristics of
a sperm sample selected on a Percoll gradient
12 µm
(microcell)
Figure 1. Effects of various temperatures on the mean values of
kinematic characteristics for one semen sample 1 h after collection
(s–s) and for the same sample selected on a Percoll gradient and
resuspended for 30 min in a capacitating medium (u–u).
Significant increases in the means of all characteristics with
increasing temperature were found for the semen sample. For the
selected sperm sample, only curvilinear velocity (VCL) and straight
line velocity (VSL) were found to increase significantly with a rise
in temperature. *, ** and *** denote significant differences from
value at a lower temperature at P , 0.05, P , 0.01 and P , 0.001
respectively. At any given temperature, VCL, VSL and amplitude
of lateral head displacement (ALH) values were significantly higher
for the selected sperm sample than for the initial semen sample
(P , 0.001 for all points). Linearity (LIN) was significantly higher
in the initial sperm sample than in the selected sperm sample only
at 24°C and 28°C (P , 0.001 and P , 0.05 respectively).
1.25-fold between 24 and 37°C. For the selected sperm
samples, VCL and VSL increased, on average, 1.3 and 1.8fold respectively between 24 and 37°C. At any given temperature, VCL, VSL and ALH values were significantly higher
for the selected sperm sample than for the initial semen sample
(P , 0.001 for all pairs; Figure 1). LIN was significantly
different for selected spermatozoa and seminal spermatozoa
only at 24 and 28°C, where the initial semen sample gave
higher values than the selected sperm sample (P , 0.001 and
P , 0.05 respectively).
Depth of preparation
All kinematic characteristics were not significantly different
for seminal sperm samples analysed in 12 and 20 µm microcells
and in the 50 µm-deep rectangular capillary tube at mid
distance from the walls (data not shown). For selected sperm
samples, the distribution of VSL was found to be similar in
12 and 20 µm microcell chambers, whereas VCL and ALH
values were significantly higher in 12 µm chambers (and LIN
values were significantly lower; all, P , 0.001; Table I). It
was not possible to analyse the complete population of selected
spermatozoa using a capillary tube 50 µm deep. A simple
microscopic examination indicated that the spermatozoa did
not swim in a unique plane and our CTS could not analyse
CON (3106/ml)
34
MOT (%)
81
VCL (µm/s) mean
108 (33)
(SD)
,40
3
40–80
22
80–120
33
.120
42
VSL (µm/s) mean
51 (31)
(SD)
,40
42
40–80
33
.80
25
LIN (%) mean (SD) 45 (20)
,20
18
20–60
56
.60
26
ALH (µm) mean (SD) 6.5 (2.5)
,3.0
4
3.0–6.0
43
.6.0
53
20 µm
(microcell)
50 µm (capillary tube)
1*
2*
26
84
99 (32)a
9
97
119 (29)
9
76
94 (28)a
3
26
45
26
55 (25)
0
11
40
49
54 (19)
0
29
50
18
47 (19)a
33
50
18
54 (17)a
2
57
40
5.6 (2.5)a
11
54
35
22
71
7
46 (15)
6
75
19
7.5 (2.8)
2
37
61
39
58
3
50 (15)a
5
69
26
6.3 (2.9)a
8
52
40
*As the spermatozoa did not swim in a unique plane in a rectangular
capillary tube 50 µm deep, the analysis was carried out in two different
planes of focus, at mid-distance from the walls (1) and near the upper wall
(2). aIn column denotes significant difference from preceding column at
P , 0.001 using non-parametric statistics (Mann–Whitney test). Kinematic
characteristics were measured at 37°C, on at least 200 motile spermatozoa.
VCL 5 curvilinear velocity; VSL 5 straight line velocity; LIN 5 linearity;
ALH 5 amplitude of lateral head displacement.
the motion of the entire population. Therefore, we studied two
different planes of focus, half-way down the chamber and near
the upper wall to see if the two subpopulations of motile
spermatozoa had similar kinematics. We found that spermatozoa swimming near the upper wall had significantly lower
VCL, VSL and ALH values and a significantly higher value
of LIN (all P , 0.001; Table I). By contrast, the highest
proportion of spermatozoa with high VCL (.120 µm/s) and
ALH (.6 µm) values were recorded for the spermatozoa
swimming half-way down the chamber (Table I).
Detection threshold
Use of a detection threshold 50 grey levels lower than
the optimal threshold (overdetection) gave spurious sperm
trajectories (related to electronic noise in background). This
resulted in higher recorded sperm concentration and percentage
of motile spermatozoa and significantly lower values of VCL,
VSL and LIN for seminal spermatozoa and significantly lower
VCL and LIN values for selected sperm samples (Table II).
Use of a detection threshold 50 grey levels higher than the
optimal threshold (underdetection) resulted in some of the
spermatozoa present in the microscopic field not being detected.
This gave lower recorded sperm concentrations but it had little
effect on the recorded values of kinematic parameters (only
VCL values in seminal samples were found to be significantly
different; P , 0.05).
613
M.Kraemer et al.
Table II. Typical effect of the detection threshold (grey level, GL) on sperm concentration (CON),
percentage of motile spermatozoa (MOT) and mean, SD and distribution frequency (%) of values for
kinematic characteristics for a semen sample (A) and spermatozoa from the same sample selected on a
Percoll gradient (B)
Detection threshold
CON (3106/ml)
MOT (%)
VCL (µm/s) mean (SD)
,40
40–80
80–120
.120
VSL (µm/s) mean (SD)
,40
40–80
.80
LIN (%) mean (SD)
,20
20–60
.60
ALH (µm) mean (SD)
,3.0
3.0–6.0
.6.0
A
B
250 GL
adjusted
150 GL
250 GL
adjusted
150 GL
40
61
65 (24)b
19
49
32
0
30 (14)c
71
29
0
47 (16)a
3
74
23
4.6 (1.7)a
15
65
19
31
49
71 (22)
10
57
33
0
35 (14)
61
39
0
50 (16)
2
71
27
4.3 (1.5)
19
68
13
24
48
66 (23)a
15
55
30
0
33 (13)
66
34
0
51 (17)
3
67
30
4.1 (1.4)
24
67
9
60
88
86 (30)b
3
44
40
13
41 (24)
56
36
8
45 (18)a
7
70
23
4.7 (1.9)b
17
61
22
51
87
81 (31)
6
48
34
12
42 (27)
56
34
10
48 (18)
5
66
29
4.3 (1.7)
23
61
16
39
81
80 (28)
5
53
32
10
40 (24)
56
37
7
48 (18)
6
66
28
4.2 (1.7)
27
59
14
For the two thresholds A and B, superscripts in columns 1 and 3 denote significant differences from columns
2 at aP , 0.05, bP , 0.01 and cP , 0.001 using the Mann–Whitney test. Kinematic characteristics were
measured at 37°C in a 20 µm microcell on at least 200 motile spermatozoa. Each experiment on the whole
and selected sample was carried out on the same microscopic field using the counter of the video recorder.
For abbreviations, see Table I.
Frame rate and maximum burst speed
For seminal spermatozoa, with a maximum burst speed of
200 µm/s, numerous trajectories were not detected (Figure
2A) and only 69% of the trajectories of motile spermatozoa were considered in the kinematic analysis. With a
maximum burst speed of 700 µm/s, false trajectories were
identified (Figure 2C). With a maximum burst speed set at
400 µm/s, most of the paths were correctly defined (Figure
2B) and .90% of the motile spermatozoa in the preparation
were tracked. Increasing the values of the maximum burst
speed from 200 to 700 µm/s resulted in higher values of
VCL and ALH, lower values of LIN and no change of
VSL (data not shown). For selected sperm samples analysed
at 60 Hz, if the maximum burst speed value was lower
than 700 µm/s (400, 500 or 600 µm/s), numerous trajectories
were not detected (Table III). These trajectories were
produced by the fastest spermatozoa (Figure 3A, B, C).
Consequently, the values of kinematic parameters were not
representative of the complete selected sperm population:
VCL and ALH values were significantly lower than with
the higher maximum burst speed value (700 µm/s) (Table
III). The optimal maximum burst speed value for selected
spermatozoa was 700 µm/s (Figure 3D and Table III).
Frame rate and ALH path-smoothing factor (APSF)
Changing the APSF in the range 5–11 points had little effect
on the smoothed trajectories and ALH values of seminal
spermatozoa analysed at 30 Hz (Figure 4A and B). Higher
614
values of APSF resulted in average path not projected onto
the curvilinear path of spermatozoa with no straight progression (data not shown). For selected sperm samples
analysed at 60 Hz, the value of ALH was more dependent on
the ALH path smoothing factor setting: the higher the APSF,
the smoother the mean path (Figure 4) and the higher the ALH
value (Figures 4 and 5). Setting the APSF too low resulted in
insufficiently smoothed paths and inadequate ALH values (see
Figure 4C and D, APSF 5 5), whereas setting the APSF factor
too high (30 points, for example) resulted in overly straight
trajectories which were not projected onto the curvilinear
trajectories (data not shown) and these smoothed trajectories were not representative of the ‘true’ average path
(consequently the corresponding ALH values were too high).
We found that, for selected sperm samples analysed at 60 Hz,
an APSF of 15 points gave meaningful results, based on the
appearance of the average paths and the calculated values of
ALH (Figure 5). Similar results were observed for APSF in
the range 15–20 points (Figure 4).
Discussion
We have demonstrated the importance of defining operational
conditions in the study of human sperm kinematics using
CASA in clinical situations. We have also shown the value of
empirical experiments to optimize the critical parameters of
the CTS machine for reliable track reconstructions and derived
kinematic measurements.
Factors influencing CASA
Figure 2. Effect of the parameter ‘maximum burst speed’ set at 200 (a), 400 (b) and 700 µm/s (c), at 30 Hz on path reconstruction of
seminal spermatozoa. Analyses were performed on the same video field with the same set-up as for the other parameters (see text). With a
maximum burst speed set at 200 µm/s many trajectories were not identified (a), for 700 µm/s, false trajectories were drawn, e.g.
spermatozoa 1 and 3, the system connecting some data points that did not belong to the same spermatozoa (c), while for 400 µm/s, most of
the paths were correctly identified (b).
Table III. Typical effect of burst speed threshold on mean, SD and
distribution frequency (%) of values for kinematic characteristics of
spermatozoa selected on Percoll gradient and analysed at 60 Hz
Maximum burst speed
(µm/s)
400
VCL (µm/s) mean (SD) 110 (41)a,c
,40
3
40–80
23
80–120
34
.120
40
VSL (µm/s) mean (SD)
59 (29)
,40
29
40–80
52
.80
19
LIN (%) mean (SD)
53 (17)
,20
7
20–60
54
.60
39
ALH (µm)
5.7
mean (SD)
(2.5)a,c,d
,3.0
12
3.0–6.0
48
.6.0
40
No. of paths analysed (%) 52
500
600
700
119 (39)
2
13
39
46
61 (26)
23
52
25
52 (17)
5
58
37
6.4
(2.5)a,b
5
49
46
63
124 (44)a
2
17
32
49
61 (27)
24
55
21
50 (17)
3
64
33
7.0
(2.9)c
5
38
57
69
127 (44)c
2
17
33
48
59 (26)
28
53
19
48 (19)
8
62
30
7.2
(3.3)b,d
10
38
52
77
Superscripts denote significant differences at a,bP , 0.05, cP , 0.01 and
dP , 0.001 using non-parametric statistics (Mann–Whitney test). Kinematic
characteristics were measured at 37°C in a 20 µm microcell on at least 200
motile spermatozoa.
For abbreviations, see Table I.
Temperature
Our results demonstrate the temperature dependence of human
sperm motion in the seminal plasma as well as in capacitating
medium as previously reported (Appell and Evans, 1977;
Makler et al., 1981; Auger et al., 1990). As significant
differences in sperm motion were observed at 24 and 37°C,
only one temperature should be used for routine analysis.
CASA analyses at 37°C are obviously preferable as they may
be more physiologically relevant.
Depth of the preparation
It has been reported that a 10 µm deep chamber, like the
Makler chamber, may be used for the analysis of seminal
samples as the low z-component of sperm motion in seminal
plasma allowed free movement in such a depth (Le Lannou
et al., 1992). In the present study, 12, 20 and 50 µm deep
chambers gave similar results. We found that recorded head
movements of selected spermatozoa were larger in the 12 µm
microcell chamber. Consequently, this artefactually increased
the values of VCL and ALH and this chamber depth should
not be used for capacitating spermatozoa, as previously reported
(Ginsburg and Armant, 1990; Le Lannou et al., 1992; Mortimer
and Swan, 1995). We were unable to analyse the entire sperm
population using a 50 µm deep capillary tube. In such a depth,
the selected spermatozoa swam in different planes, those
swimming near the upper wall having significantly lower mean
615
M.Kraemer et al.
Figure 3. Effect of the parameter ‘maximum burst speed’ set at 400 (a), 500 (b), 600 (c) and 700 µm/s (d), at 60 Hz on path reconstruction
of spermatozoa selected on Percoll gradients and resuspended for 30 min in capacitating medium at 37°C. Analyses were performed on the
same video field with the same set-up as for the other parameters (see text). For a maximum burst speed set at 400 (a) or 500 (b) µm/s,
several trajectories were not identified and these were produced by the fastest spermatozoa. Even with a burst speed set at 600 µm/s (c), a
trajectory of a typically rapid spermatozoon was omitted by the system. A maximum burst speed set at 700 µm/s gave optimal results (d).
velocities and lateral head amplitudes than those swimming at
some distance from the chamber walls. Although our CTS
machine could not detect all the spermatozoa in a 50 µm
depth, this was not a problem for kinematic analysis because
only the free-swimming spermatozoa should be the ones chosen
for analysis. When comparing 20 and 50 µm, the higher
proportion of spermatozoa with high VCL (.120 µm/s) were
found in the 50 µm capillary tube at mid-distance from the
walls. Our results suggest that 50 µm deep chambers are
useful for analysing capacitating spermatozoa and probably
necessary for analysing spermatozoa with higher displacements
in the z-direction (hyperactivated spermatozoa). Similarly,
some subpopulations of seminal spermatozoa sometimes have
a large displacement in the z-direction. For this reason, it
might be better to analyse seminal spermatozoa in 20 µm deep
chambers rather than in chambers with a lower depth.
Detection threshold
The manual control of the detection threshold is a critical step
under the visual control of the CTS machine operator. It is
vital to use appropriate optics (positive phase-contrast) and
the best illumination of the preparation to enhance the contrast
of the sperm heads which in turn facilitates the manual
thresholding step. We found significant differences in kinematic
parameters and some non-significant changes in the percentage
of motile spermatozoa and the sperm concentration at small
616
variations in the threshold level. The CTS machine allows the
visual adjustment of the detection threshold by projecting the
detected objects onto the live image. Thus, it is possible to
detect only sperm heads and to ensure that all sperm heads
are detected. Davis et al. (1992a) suggested a precalibration
of the machine for determining the optimal threshold by
verifying that the machine provides similar counts of
spermatozoa to the counts determined visually from videotapes. This recommendation is of value in the use of the CTS
machine. However, in most routine situations, the use of this
procedure alone is not always efficient as the microscopic
fields analysed are not homogeneous in terms of the number
of spermatozoa analysed, the presence of other cells and debris,
or the illumination of the background. Such problems may
lead to readjustment of the grey level threshold during the
analysis. Automatic thresholding methods developed in various
fields of image analysis and processing might be useful for
improving this critical step.
Frame rate and maximum burst speed
There are marked differences in the motion of human spermatozoa in seminal plasma and in capacitating medium (see
Figure 1 and Table II). Thus, different calibrations of the CASA
machine should be performed to reconstruct adequately sperm
trajectories and calculate sperm kinematics for both situations.
The acquisition frequency is important as the reconstructed
Factors influencing CASA
Figure 4. Aspect of the average path and corresponding amplitude
of lateral head displacement (ALH) value according to different
values of the ‘ALH path-smoothing factor’ (APSF) for two seminal
spermatozoa analysed at 30 Hz (A, B) and two selected
spermatozoa analysed at 60 Hz (C, D). APSF in the range 5–11
points had only a weak effect on sperm trajectories with a thin
amplitude of the head (A). The effect was more pronounced for
sperm trajectories with a larger amplitude (B). Using APSF in the
range 5–10 points, trajectories of selected spermatozoa were
insufficiently smoothed (C, D), while 15 points gave adequate
average trajectories. APSF in the range 15–20 gave similar aspects
of the average trajectory (data not shown). Higher values of APSF
were inadequate for all the spermatozoa with no linear motion,
since for such values and such spermatozoa, the average trajectory
was not projected onto the curvilinear trajectory.
Figure 5. Effect of the parameter ‘amplitude of lateral head
displacement (ALH) path-smoothing factor’ at 60 Hz on ALH
values for spermatozoa selected on Percoll gradients and
resuspended for 30 min in capacitating medium at 37°C. Analyses
were performed on at least 200 motile spermatozoa with the same
set-up as for the other parameters (see text). *P , 0.05 and
***P , 0.001 in comparison with preceding value of ‘ALH pathsmoothing factor’ using non-parametric statistics (Mann–Whitney
test).
trajectory is influenced by the frame rate (Mortimer et al.,
1988) and several kinematic parameters are frame rate
dependent (Boyers et al., 1989; Morris et al., 1996). It is
currently accepted that human sperm motion in a capacitating
medium should be analysed at 50 Hz or more (Mortimer et al.,
1995). Using the CTS instrument, we found that 60 Hz was
better than 30 Hz for such conditions. The optimal frequency
of image acquisition for studying human spermatozoa in
seminal plasma is still under discussion. Owen and Katz (1993)
studied the sampling factors influencing accuracy of sperm
kinematic analysis. They suggested that 60 frames/s was
suitable for most CASA applications to human and other
mammalian spermatozoa. Using the CTS machine, we found
no significant changes in the kinematics parameters of spermatozoa in seminal plasma at 30 and 60 Hz (pilot study). We
recommend analysis of semen samples, using the CTS machine
at 30 Hz rather than at 60 Hz as non-detection of spermatozoa
is less likely at 30 Hz (i.e. the probability of fragmented
trajectories due to a failure to detect the spermatozoa in some
frames is lower). The acquisition frequency is not the only
crucial parameter for reliable kinematic measures. If maximum
burst speed is too low, trajectories may not be detected,
whereas false trajectories may result from the connection by
the CTS algorithm of points belonging to different spermatozoa.
This problem occurs because sperm trajectories cross and this
increases with sperm concentration (Davis and Katz, 1993).
False trajectories may also occur if maximum burst speed is
set too high. According to the CTS reference manual, this
parameter ‘should be set to be slightly higher than the maximum
expected speed of the sperm cells to be tracked’. However, in
clinical applications of CASA, the value of this parameter
should account for: i) the maximum expected instantaneous
velocity of the spermatozoa, which may vary strongly from
cell to cell, ii) the micro-environment of the spermatozoa, which
influences their velocity (i.e. seminal plasma or capacitating
medium) and iii) the frame rate (i.e. for a given cell, higher
frame rates give higher maximum ‘instantaneous’ velocities).
We were able to show, by using video tapes and iterative
analyses of the same spermatozoa with different values of
these parameters, that for the CTS machine, most of the paths
for a semen sample were correctly identified at 30 Hz with a
maximum burst speed of 400 µm/s. We found that numerous
false trajectories were generated when seminal spermatozoa
were analysed at 60 Hz, with a maximum burst speed threshold
set at 600 µm/s (data not shown), the experimental conditions
reported in the CTS study of Paston et al. (1994). We found
that the optimal maximum burst speed value was 700 µm/s
for spermatozoa selected on a Percoll gradient, incubated in a
capacitating medium and analysed at 60 Hz with the CTS
machine.
ALH path-smoothing factor
It is important to obtain accurate measures of the ALH, as
this is one of the kinematic parameters affecting the outcome
of IVF (Jeulin et al., 1986; Chan et al., 1989; Barlow et al.,
1991) and the ability of human spermatozoa to penetrate
cervical mucus and fuse with the oocyte (Aitken et al., 1985,
1992, 1994; Feneux et al., 1985; Mortimer et al., 1986).
This parameter has biological importance probably because it
indicates the vigour of flagellar beating together with the
frequency of cell rotation (David et al., 1981; Serres et al.,
1984), which are probably important determinants of the
617
M.Kraemer et al.
Table IV. General recommendations for kinematics analysis of seminal samples and selected sperm samples
using the CTS machine
Use an analysis chamber designed for spermatozoa [typically, 20 µm depth, for seminal spermatozoa, and
deeper for selected (capacitated/hyperactivated) spermatozoa] and mix the sample well before its introduction
into the chamber
Use positive phase-contrast optics for semen samples and selected sperm samples, and a 310 objective (with
a 36.7 photo lens giving a 367 final magnification)
Optimize the illumination of the preparation to obtain the best contrast of sperm heads
Analyse all specimens at 37°C (heated stage)
Videotape semen samples and selected sperm samples at 37°C which should be kept for quality control,
training, diagnostic and forensic medicine purposes
Adjust the detection threshold carefully (concentration by CASA µ manual concentration)
Set the frame rate at 30 Hz (seminal spermatozoa) or 60 Hz (selected spermatozoa)
Set the duration of data capture at 15 points (seminal spermatozoa) or 30 points (selected spermatozoa)
Set the min. threshold velocity at 10 µm/s (seminal spermatozoa) or 8 µm/s* (selected spermatozoa)
Set the maximum burst speed threshold at 400 (seminal spermatozoa) or 700 µm/s (selected spermatozoa)
Set the ALH path-smoothing factor at 5–10 points (seminal spermatozoa) or 15–20 points (selected
spermatozoa)
Analyse at least 200 motile spermatozoa in seminal and selected sperm samples
Do not analyse specimens with lots of round cells or debris
Do not analyse, without suitable dilution, specimens with a concentration .100 3 106 spermatozoa/ml as
the tracks cross too frequently
Since there is no automatic selection of the analysed microscopic fields, use a standardized method for the
manual scanning of the preparation to avoid sample bias
Examine the distribution profiles of the kinematic measurements and use non-parametric statistics for
comparisons of kinematic measurements
*With the CTS machine, we determined empirically that higher values of this parameter excluded some
typical ‘star-spin’ tracks of hyperactivated spermatozoa (data not presented).
progression of spermatozoa into the cervical mucus and the
peri-oocyte envelopes. Using the CTS machine, the value of
ALH is derived from the calculation of the average path after
smoothing using a fixed-length, weighted, running average
(Boyers et al., 1989). It is vital to have enough track points
to derive appropriate average paths and consistent ALH values.
In general, acquisitions of at least 0.5 s (i.e. 15 data points at
30 Hz and 30 data points at 60 Hz) are required to identify
adequately derived, average trajectories (Mortimer et al., 1995).
Using visual controls of the reconstructed curvilinear and
derived average trajectories of different quality semen samples,
we found that the fixed-length, weighted, running average
method used by the CTS does not allow an accurate determination of the average paths of all spermatozoa and, therefore, a
valid ALH value cannot be given for every motile spermatozoon analysed. In each microscopic field, some of the trajectories are very tortuous and irregular whereas others have a
regular pseudo-sinusoidal pattern. Such differences result from
the amount and the regularity of rotation of the cell, the
curvilinear velocity and from the amplitude of the proximal
flagellar wave (Serres et al., 1984). This continuously fluctuates
and may differ strongly from one cell to another reflecting the
enormous heterogeneity of human sperm samples. Thus, the
choice of an appropriate ALH path-smoothing factor must be
based on a subjective assessment of which values of this
parameter provide valid trajectories and ALH values for most
motile spermatozoa. We showed empirically that the standard
path-smoothing factor calibration proposed by the manufacturer
(7–9 points for 30 or 60 Hz) was adequate for semen samples
but not for selected sperm samples of various quality, tested
in the present study and comparable to the selected sperm
samples we analyse routinely. We showed that for selected
sperm samples analysed at 60 Hz, an ALH path-smoothing
618
factor of 15–20 points gave relevant results for most of the
motile spermatozoa (according to the visual appearance of all
the average paths and the calculated mean values of ALH).
Optimized operational conditions for analysing semen
samples and selected sperm samples with the CTS CASA
machine
Table IV summarizes our guidelines for use of the CTS
machine for kinematic measurements. These guidelines combine previously reported recommendations for all CASA systems (for reviews, see Boyers et al., 1989; Mortimer et al.,
1995; ESHRE Andrology Special Interest Group, 1996) and
specific rules for the CTS machine used in the present study.
It should be noted, however, that each laboratory must verify
all of the parameters for its version of the CASA instrument
used.
In conclusion, multi-purpose set-ups, such as that proposed
by CASA manufacturers, cannot guarantee reliable kinematic
measurements of human spermatozoa for the vast range of
clinical applications of an andrology/IVF laboratory. The effect
of each calibration parameter on track reconstruction and
derived kinematic measurements must be assessed empirically
to optimize the set-up for each specific clinical situation.
Moreover, the use of standardized procedures is not enough
to guarantee reliable kinematic measurements. A basic understanding of human sperm motion and how the CASA machine
works, intensive training and the use of quality control techniques are all fundamental prerequisites for clinical studies to
determine the diagnostic and prognostic value of kinematic
measurements.
References
Aitken, R.J., Sutton, M., Warner, P. and Richardson, D.W. (1985) Relationships
between the movement characteristics of human spermatozoa and their
Factors influencing CASA
ability to penetrate cervical mucus and zona-free hamster oocytes. J. Reprod.
Fertil., 73, 441–449.
Aitken, R.J., Bowie, H., Harkiss, D. et al. (1992) Sperm penetration into a
hyaluronic acid polymer as a means of monitoring functional competence.
J. Androl., 13, 44–54.
Aitken, R.J., Buckingham, D. and Harkiss, D. (1994) Analysis of the extent
to which sperm movement can predict the results of ionophore-enhanced
functional assays of the acrosome reaction and sperm–oocyte fusion. Hum.
Reprod., 9, 1867–1874.
Appell, R.A. and Evans, P.R. (1977) The effect of temperature on sperm
motility and viability. Fertil. Steril., 28, 1329–1332.
Auger, J., Serres, C. and Feneux, D. (1990) Motion of individual human
spermatozoa, both normal and lacking the outer dynein arms, during a
continuous temperature rise. Cell Motil. Cytoskel., 16, 22–32.
Auger, J., Léonce, S., Jouannet, P. and Ronot, X. (1993) Flow cytometric
sorting of living, highly motile human spermatozoa based on evaluation of
their mitochondrial activity. J. Histochem. Cytochem., 41, 1247–1251.
Barratt, C.L.R., Tomlison, M.J. and Cooke, I.D. (1993) Prognostic significance
of computerized motility analysis for in vivo fertility. Fertil. Steril., 60,
520–525.
Barlow, P., Delvigne, A., Van Dromme, J. et al. (1991) Predictive value of
classical and automated sperm analysis for in vitro fertilization. Hum.
Reprod., 6, 1119–1124.
Boyers, S.P., Davis, R.O. and Katz, D.F. (1989) Automated semen analysis.
Curr. Probl. Obstet. Gynecol. Fertil., 12, 165–200.
Chan, S., Wang, C., Chan S., Ho, P. and So, W. (1989) Predictive value of
sperm morphology and movement characteristics in the outcome of in vitro
fertilization of human oocytes. J. In Vitro Fert. Embryo Transf., 6, 142–148.
Comhaire, F. H., Huisse, D., Hinting, A. et al. (1992) Objective semen
analysis: has the target been reached? Hum. Reprod., 7, 237–241.
David, G., Serres, C. and Jouannet, P. (1981) Kinematics of human
spermatozoa. Gamete Res., 4, 83–95.
Davis, R.O. and Katz, D.F. (1993) Operational standards for CASA instruments.
J. Androl., 14, 385–394.
Davis, R.O., Rothmann, S. and Overstreet, J. (1992a) Accuracy and precision
of computer-aided sperm analysis in multicentre studies. Fertil. Steril., 57,
648–653.
Davis, R.O., Bain, D.E., Siemers, R.J. et al. (1992b) Accuracy and precision
of the CellForm-human automated sperm morphometry instrument. Fertil.
Steril., 58, 763–769.
ESHRE Andrology Special Interest Group. (1996) Consensus workshop on
advanced diagnostic andrology techniques. Hum. Reprod., 11, 1463–1479.
Feneux, D., Serres, C. and Jouannet, P. (1985) Sliding spermatozoa: a
dyskinesia responsible for human infertility? Fertil. Steril., 44, 508–511.
Gandini, L., Lombardo, F., Lenzi, A. et al. (1997) The in-vitro effects of
nicotine and cotinine on sperm motility. Hum. Reprod., 12, 727–733.
Ginsburg, K.A. and Armant, D.R. (1990) The influence of chamber
characteristics on the reliability of sperm concentration and movement
measurements obtained by manual and videomicrographic analysis. Fertil.
Steril., 53, 882–887.
Holt, W., Watson, P., Curry, M. and Holt, C. (1994) Reproducibility of
computer-aided semen analysis: comparison of five different systems used
in a practical workshop. Fertil. Steril., 62, 277–1282.
Hurowitz, E.H., Leung, A. and Wang, C. (1995) Evaluation of the CellTrak
computer-assisted sperm analysis system in comparison to the Cellsoft
system to measure human sperm hyperactivation. Fertil. Steril., 64, 427–432.
Jeulin, C., Feneux, D., Serres, C. et al. (1986) Sperm factors related to failure
of human in vitro fertilization. J. Reprod. Fertil., 6, 735–744.
Katz, D.F. and Davis, R.O. (1987) Automatic analysis of human sperm motion.
J. Androl., 8, 170–182.
Katz, D.F., Davis, R.O., Delandmeter, B.A. and Overstreet, J.W. (1985) Realtime analysis of sperm motion using video image digitization. Comput.
Methods Programs Biomed., 21, 173–182.
Le Lannou, D., Griveau, J.F., Le Pichon, J.P. and Quero, J.C. (1992) Effects
of chamber depth on the motion pattern of human spermatozoa in semen
or in capacitating medium. Hum. Reprod., 7, 1417–1421.
MacLeod, I.C. and Irvine, D.S. (1995) The predictive value of computerassisted semen analysis in the context of a donor insemination programme.
Hum. Reprod., 10, 580–586.
Mahadevan, M.M., Miller, M.M. and Moutos, D.M. (1997) Absence of glucose
decreases human fertilization and sperm movement characteristics in vitro.
Hum. Reprod., 12, 119–123.
Makler, A., Deutch, M., Vilensky, A. and Palti, Y. (1981) Factors affecting
sperm motility. VIII. Velocity and survival of human spermatozoa as related
to temperature above zero. Int. J. Androl., 4, 559–569.
Moohan, J., Winston, R. and Lindsay, K. (1993) Variability of human sperm
response to immediate and prolonged exposure to pentoxifylline. Hum.
Reprod., 8, 1696–1700.
Morris, A.R., Coutts, J.R.T. and Robertson, L. (1996) A detailed study of the
effect of videoframe rates of 25, 30 and 60 Hertz on human sperm movement
characteristics. Hum. Reprod., 11, 304–310.
Mortimer, D., Pandya, I.J. and Sawers, R.S. (1986) Relationship between
human sperm motility characteristics and sperm penetration into cervical
mucus in vitro. J. Reprod. Fertil., 78, 93–102.
Mortimer, D., Serres, C., Mortimer, S.T. and Jouannet, P. (1988) Influence of
image sampling frequency on the perceived movement characteristics of
progressive motile human spermatozoa. Gamete Res., 20, 313–327.
Mortimer, D., Aitken, R.J., Mortimer, S.T. and Pacey, A.A. (1995) Workshop
report: clinical CASA – the quest for a consensus. Reprod. Fertil. Dev., 7,
951–959.
Mortimer, S.T. and Swan, M.A. (1995) Kinematics of hyperactivated human
spermatozoa analysed at 60 Hz. Hum. Reprod., 10, 873–879.
Owen, D.H. and Katz, D.F. (1993) Sampling factors influencing accuracy of
sperm kinematic analysis. J. Androl., 14, 210–221.
Paston, M., Sarkar, R. S., Oates, R. and Badawy, S. (1994) Computer-aided
semen analysis variables as predictors of male fertility potential. Arch.
Androl., 33, 93–99.
Serres, C., Feneux, D., Jouannet, P. and David, G. (1984) Influence of the
flagellar wave development and propagation on the human sperm movement
in seminal plasma. Gamete Res., 9, 183–93.
Skiba-Lahiani, M., Auger, J., Terribile, J. et al. (1995) Stimulation of
movement and acrosome reaction of human spermatozoa by PC12 liposomes
encapsulating ATP. Int. J. Androl., 18, 287–294.
World Health Organization (1992) WHO Laboratory Manual for the
Examination of Human Semen and Semen–Cervical Mucus Interaction.
Cambridge University Press, Cambridge, UK.
Received on May 13, 1997; accepted on November 26, 1997
619