Hunan Reproduction vol.11 no.2 pp.3O4-310, 1996
A detailed study of the effect of videoframe rates of 25,
30 and 60 Hertz on human sperm movement
characteristics
A.RJVIorris1, J.R.T.Coutts1 and L.Robertson2'3
2
Departments of 'Obstetrics and Gynaecology and Veterinary
Anatomy, University of Glasgow, Bearsden Road, Bearsden,
Glasgow G61 1QH, Scotland, UK
^To whom correspondence should be addressed
A comparison was made of the movement characteristics
of human spermatozoa analysed at three videoframe rates
(25, 30 and 60 Hz) using two computerized motUity analysers from Hamilton-Thorn Research (the HTM-2030 and
the FVOS) operating at 25 and 30 Hz respectively. Analysis
at 30 and 60 Hz was performed on the FVOS. The use of
{incapacitated, capacitated and pentoxifylline-stimulated
spermatozoa ensured a full range of movement characteristics was analysed. The velocity parameters curvilinear
velocity and average path velocity were highly frame-rate
dependent, and mean values increased with videoframe
rate. An interaction of framing rate and time of data
collection resulted hi an increase in straight-line velocity
with framing rate. Mean lateral head displacement and
linearity were similar at 25 or 30 Hz but significantly
depressed at 60 Hz. Beat-cross frequency increased by 74%
when analysed at 60 rather than 30 Hz. The following
criteria: curvilinear velocity >100 jim/s, linearity <65%
and lateral head displacement >7.5 urn, were used to define
hyperactivated spermatozoa. Significantly more hyperactivated cells were identified at 30 Hz than 25 Hz (1-10%)
but not at 60 Hz. A different population of cells is likely
to have been identified as hyperactivated at 60 Hz due to
alterations hi component movement parameters from which
the definition of hyperactivation was derived. In conclusion,
direct comparisons should not be drawn between data
analysed at 25 and 30 Hz. Analysis at 60 Hz introduced
complex alterations which made simple comparisons with
30 Hz data invalid.
Key words: CASA/Harnilton-Thorn/human spermatozoa/hyperactivarion/videoframe rate
Introduction
Conventional semen analysis, based on visual assessment, is
subject to significant intra- and interlaboratory variation even
when standard protocols are followed (Dunphy et aL, 1989).
Furthermore, data obtained in this way are poorly correlated
with fertilizing potential (Polansky and Lamb, 1988). An
objective method of semen analysis was sought and computerassisted semen analysis (CASA) systems, which combine the
technology of videomicrography and image processing, were
304
developed (Katz and Overstreet, 1981; Holt et al, 1985;
Mortimer, 1986). These systems had the additional aim of
denning sperm movement more accurately, and it was hoped
that CASA would achieve rapid, objective and reliable analysis
of large numbers of individual spermatozoa. With experience
of using this type of equipment it has become apparent that
different results can be produced according to the system in
use and the method of setting sperm detection parameters
(Mahony et al, 1988; Aanesen and Benvold, 1990). Davies
et al. (1992a) have shown that satisfactory calibration of
machines from the same manufacturer sited at different centres
can be achieved, allowing direct comparison of data in a
multicentre trial. However, there is as yet no consensus on
calibration of equipment or parameter settings for analysis.
Sperm trajectories are calculated from the position of the
sperm head identified in successive videoframes. Mortimer
et al. (1988) demonstrated that some of the parameters used
to describe sperm movement are highly dependent on the
videoframe rate employed. Difficulties can therefore arise
when data from different studies are compared. Unfortunately,
a fundamental problem exists when comparison of data generated in Europe and North America is attempted. The European
video standard utilizes the PAL system which runs at a framing
rate of 25 images, or frames, per second (Hz) while the North
American video standard uses the NTSC system operating at
30 Hz. One manufacturer supplying CASA systems to the
European market has attempted to resolve this problem by
replacing the original model which used the 25 Hz system
with a new model designed to run at 30 Hz (Hamilton-Thorn
2030 and Hamilton-Thorn IVOS respectively; Hamilton-Thorn
Research, 181 Elliot Street, MA 019150, USA). While this is a
move towards uniformity with North America, it may make
comparison of data produced by European andrology laboratories even more difficult Indeed, some such centres already
possess both a 25 and a 30 Hz system.
This study was carried out to compare directly data for
movement characteristics produced from the analyses of
duplicate sperm samples using the 25 and 30 Hz machines
named above. Since the IVOS is also capable of analysing at
60 Hz, a further comparison was made by re-analysing the
same videotapes at that rate. Semen samples used in the
comparison were manipulated to provide spermatozoa with a
range of movement characteristics (diluted semen, washed
cells, capacitated cells and pentoxifylline-stimulated cells). In
addition to assessing die effect of different frame rates on
basic movement characteristics (velocity parameters, amplitude
of lateral head deviation, linearity and beat-cross frequency),
the effect on detection of hyperactivation was assessed.
© European Society for Human Reproduction and Embryology
Study of sperm movement analysed at 25, 30 or 60 Hz
Table I. Basic machine settings used during analyses carried out with either
of two Hamilton—Thorn computerized motility analyser (HTM-2030 and
HTM-IVOS). Spermatozoa also required average path velocity (VAP)
» 5 um/s for inclusion in the data set
Parameter
HTM-2030
HTM-IVOS
Sampling frequency (Hz)
Frames acquired
Magnification factor
Minimum contrast
Minimum cell size
Static size limits
Static intensity limits
Static elongation limits
25
13-20
2.13
6
6
0.4-1.6
0.5-2.0
NA
30 and 60
13-30
2.69
90
5
0.5-2.0
0.5-2.0
0-100
NA = not available.
Materials and methods
Semen collection and preparation
Ejaculates were collected from eight donors after a minimum of 48 h
abstinence. A maximum of 1 h was allowed for liquefaction before
assessment of volume, concentration and percentage motility, performed according to World Health Organization guidelines (WHO,
1987). All samples used were normozoospermic. To ensure a varied
range of movement patterns, four samples were prepared from each
ejaculate: (i) raw semen, containing slow moving, (incapacitated
spermatozoa; (ii) washed spermatozoa, immediately after separation
from seminal plasma; (iii) washed spermatozoa after incubation in
capacitating medium at 37°C for 75 min; (iv) washed spermatozoa
incubated with 3.6 mM pentoxifylline at 37°C for 75 min to stimulate
motility and hyperactivation.
Dilution and incubation was carried out using Medicult IVF
medium, pH 7.4, 285 mOsm/kg (DK-2100; Medi-Cult, Copenhagen,
Denmark) incubated at 37°C in 5% CO 2 in air. For each sample, an
aliquot of raw spermatozoa was diluted in culture medium (1:1 or
1:2 parts of semen to culture medium) and videotaped immediately.
Up to 2 ml of the remainder was centrifuged through a discontinuous
Percoll (Sigma Chemical Co., Poole, Dorset, UK) gradient (Kay
et al, 1993). One portion of thiswashed sample was incubated with
3.6 mM pentoxifylline for 75 min, a treatment previously shown to
maximize stimulation of hyperactivated motility (Kay et al., 1993).
For video recording, 5 |il aliquots of samples were transferred by
one operator to duplicate chambers of 32 um depth and recorded
simultaneously by two operators using the two analyser/video systems
described below. Cell chambers were prepared as previously described
(Kay et al., 1993). This process was repeated to give duplicate slides
per sample per machine. All recording was carried out at 37°C.
Multiple microscope fields were recorded for 10 s and slides discarded
within 5 min of preparation.
Movement analysis
Two Hamilton-Thorn automated motility analysers and video recorder
combinations were used for motility analyses: (i) PAL (25 Hz) system,
HTM 2030 version 7.1 with VHS video recorder (Ferguson videostar
FV 321) and (ii) NTSC (30 Hz) system, HTM IVOS version 10.4m
with VHS video recorder (Aiwa HV-M110).
Parameter settings were optimized for identification of spermatozoa
within each set-up (Table I). A minimum of 200 motile spermatozoa
with at least 13 track points and an average path velocity (VAP)
>5 u\m/s were analysed from the videotapes. Track data were collected
for the parameters described below. Track data were saved onto the
HTM database (Hdata) and from there transferred to a separate
statistics package (Minitab Inc., 3081 Enterprise Drive, State College,
PA 16801-2756, USA).
Comparison of movement parameters analysed at 25, 30 or 60 Hz
From these data two comparisons were made. A comparison was
made of data derived from duplicate samples analysed at 25 or 30 Hz.
A further comparison was made of motility parameters analysed at
30 frames/s for 1 s and at 60 frames/s for 0.5 s from the videotapes
recorded at 30 Hz.
Definitions of movement parameters
Sperm trajectories were reconstructed from the position of the sperm
head (centre of brightness) in successive videoframes for a minimum
and maximum number of videoframes defined by the user, within the
limits of the equipment
Movement parameters are derived according to the following
definitions. Curvilinear velocity (VCL, um/s) was the total distance
between each head point for a given cell during the acquisition
period, divided by the time elapsed. Straight-line velocity (VSL,
Um/s) was the straight line distance between the first and last headpoint divided by the acquisition time. Average path velocity (VAP,
|im/s): to obtain a path more representative of the sperm position
along the trajectory a smoothing algorithm was used to reduce the
effect of lateral head displacement The track was smoothed by
averaging several neighbouring positions on the track and joining the
averaged positions. At 25 and 30 Hz, five neighbouring points were
averaged and at 60 Hz nine points. This was intended to minimize
the effect of framing rate on perceived velocity. The amplitude of
lateral head displacement (ALH, (im) was intended to give a measure
of sperm head oscillation. It was calculated from the maximum sperm
head departure from the average path, in (im. Since this only
represented departure from the path in one direction, this figure was
doubled to give a full width amplitude. Beat-cross frequency (BCF,
Hz) was determined by measuring the frequency with which the
sperm track crossed the average path in either direction. Linearity
(UN, %) measured the departure from linear progression. It was
defined as VSL/VCLX100, with 100% representing an absolutely
straight track. Straightness (STR, %) gave a similar measure using
the ratio VAP/VCLX100. Hyperactivation (HA, %) was determined
according to the criteria VCL >100 |im/s, ALH >15 |im and
LIN <65%.
Results
Effect of treatment on movement characteristics
A summary of values for population mean, median and range,
organized by treatment and by framing rate of analysis, is
shown in Table II. The mean population VCL, VAP and ALH
increased significantly in washed cells when compared to
spermatozoa in seminal plasma (P < 0.01). The percentage
of hyperactivated motility detected increased in washed and
incubated populations (P < 0.01) and showed highest values
in samples stimulated with pentoxifylline (>40%). The effect
of treatment ensured that a representative selection of motility
patterns was analysed. There was a significant difference in
VSL and LIN between individual donors (P < 0.01) though
not in the other parameters.
A comparison of values derived from analysis at 25, 30 or
60 Hz
In order to determine the level of agreement between measurement of the same movement parameters at different framing
305
40.20
38.90
27.5-59.1
43.70
44.40
32.4-52.7
4.50
4.50
3.7-5.0
VSL (jim/s)
Mean
Median
Range
VAP (\urds)
Mean
Median
Range
ALH(jun)
Mean
Median
Range
64.80*
63.10
59.5-73.8
59.30
58.60
51.2-66.3
78.50
81.10
55.8-86.3
2.30
2.00
0-4.5
STR (%)
Mean
Median
Range
HA (%)
Mean
Median
Range
3.50
2.60
0-11.7
80.20
84.70
43.0-90.1
57.10*
54.50
43.3-73.5
26.6-37.3
31.90°
30.80
3.90°
3.80
3.0-4.7
62.10°
59.70
49.2-73.1
57.30°
56.80
42.7-76.3
95.10°
90.30
76.3-115.1
16.40
16.00
0.5-39
73.60
76.50
54.6-79.0
55.90
55.80
49.9-61.6
14.90
14.90
12.9-16.2
6.80
6.80
4.8-8.9
62.50
63.40
49.9-80.8
58.60
52.50
42.3-115.7
74.00
66.30
52.3-115.7
21.30"
21.10
3.2-42.8
24.50"
27.10
3.0-41.3
75.40
78.80
51.3-84.3
50.60°
50.60
45.6-56.7
59.80
60.80
47.3-67.7
77.30"
78.40
63.2-86.1
35.30°
35.90
31.4-39.0
5.90°
6.10
4.3-7.4
85.90°
85.40
70.6-99.7
80.80*
69.50
59.6-166.7
123.40°
120.80
64.7-166.7
60 Hz
19.10°
19.10
17.8-20.0
6.80
6.90
4.7-8.3
72.20*
72.40
61.8-86.6
.68.40°
61.90
53.8-125.0
87.20°
80.50
52.3-115.7
30 Hz
26.30
24.00
4.8-55.2
78.80
78.60
70.2-85.2
59.00
57.30
53.7-67.3
13.00
13.90
7.9-15.8
7.30
7.30
5.3-9.4
69.40
71.10
54.8-83.3
64.20
54.90
46.7-123.8
95.50
96.70
71.1-123.8
25 Hz
31.30
29.90
5.3-54.0
34.10
30.10
6.0-79.7
43.90
50.20
15.0-58.7
72.90
76.30
46.6-82.4
53.40°
50.30
40.6-60.5
59.00
59.60
49.8-66.6
79.20
80.90
72.2-84.2
53.60
52.80
45.6-66.3
28.7-34.6
16.7-20.3
80.90"
80.20
74.5-87.4
13.40
14.80
6.1-17.1
8.70
9.00
6.3-9.8
73.40
73.10
67.2-79.0
64.80
58.10
52.4-118.9
107.90
112.30
85.7-120.3
25 Hz
31.80°
32.10
6.50°
6.50
4.9-8.0
87.60°
87.00
71.2-99.6
82.50"
70.10
61.6-176.6
145.60*
151.60
102.0-176.6
60 Hz
53.60*
57.50
23.0-78.8
72.70
75.80
40.3-83.7
50.40"
49.70
40.3-63.4
19.50°
19.60
17.5-20.7
9.00
9.00
6.5-11.2
75.90"
75.60
70.5-85.7
68.80
61.50
54.7-129.4
121.70°
127.20
97.3-139.7
30 Hz
54.50
57.60
16.0-81.4
70.10
75.00
31.6-83.9
43.10°
42.90
31.6-56.3
26.7-35.1
29.90°
29.60
7.90°
8.00
5.3-9.8
91.80°
93.00
81.2-100.1
84.60*
71.90
59.7-190.0
170.90°
175.00
135.5-195.2
60 Hz
Incubated in presence of pentoxifylline
18.90°
19.30
7.50°
7.50
5.2-9.2
75.10"
74.00
58.5-85.0
69.60*
60.20
51.9-128.8
107.70b
113.30
76.0-128.8
30 Hz
Incubated in absence of pentoxifylline
VCL =• curvilinear velocity; VSL «= straight-line velocity; VAP = average path velocity; ALH = amplitude of lateral head displacement; BCF = beat-cross frequency; LIN = linearity; STR = straightness;
HA = hyperactivation. See text for more details.
"•^Significant differences between 25 and 30 Hz or 30 and 60 Hz; "/> < 0.05, bP < 0.01, CP < 0.001.
3.50
2.50
0-12.2
82.30*
84.80
62.7-87.7
16.70°
16.50
15.4-18.2
13.60
13.70
12.4-14.5
4.70"
4.60
3.6-5.5
50.50°
49.10
37.1-61.7
47.70°
48.40
32.9-61.6
65.80°
65.20
48.0-78.9
Mean
Median
Range
LIN (%)
Mean
Median
Range
BCF(Hz)
56.90
56.70
42.7-65.4
VCL (nm/s)
Mean
Median
Range
25 Hz
60 Hz
25 Hz
30 Hz
Washed
Semen
Treatment
Table II. Mean, median and range of values for movement characteristics of human spermatozoa in seminal plasma, after washing by centrifuga''in and after incubation in the presence or absence of
pentoxifylline, analysed at 25,30 or 60 Hz
Study of sperm movement analysed at 25, 30 or 60 Hz
70-
VCL Gun/sec)
b
• 2SD
70-,
60-
60-
5040-
•2SXI.
£
40-
S
20-
30-
20-
8
10
DHtnnn
100
0-
-10
80
100
120
-20
140
1
20
40
Average (of 30 and 60 Hz)
\
1
1
60
'
1
80
•
1
1
'
TO
1
'
120
I
'
'
140
180
140
180
Average (of 30 and 60 Hz)
•
OOmmiue
-2SD.
•
40
80
80
100
120
140
1S0
180
Average (of 25 and 30 Hz)
Figure 1. (a) Bland and Altman plot of VCL analysed at 25 and
30 Hz (lower graph). Mean difference between measurements was
11.6 |im/s, with a 95% range of differences from -0.7 to
23.9 |im/s. (b) Bland and Altman plot of VCL analysed at 30 and
60 Hz (upper graph). Mean difference between measurements was
38.1 |im/s, with a 95% range of differences from 14.3 to
61.9 nxn/s.
rates, data were manipulated as described by Bland and Altman
(1986). The difference between the values derived at 25 and
30 Hz or 30 and 60 Hz was plotted against their average
value. Mean of the differences and the 95% range of differences
were also plotted. Bland and Altman plots are shown for VCL,
VSL, ALH, BCF and HA (Figures 1-5). Mean differences
between two analysis frequencies, SD and 95% range of
differences (i.e. ± 2 SD) for all movement parameters are
summarized in Table m .
The velocity parameters VCL, VSL and VAP increased
significantly when analysed at 30 Hz rather than 25 Hz. A
mean increase of 11.6 |im/s (14%) was observed for VCL,
7 itm/s (13%) for VSL and 6.1 fim/s (11%) for VAP. A greater
increase was noted in samples analysed at 60 Hz, where the
mean difference for VCL equalled 38 \lio/s, a 40% increase
above the 30 Hz measurement. Values for the increase in
VSL and VAP were 12.5 \tm/s (15%) and 13.4 pxn/s (22%)
respectively. The frame rate dependency of VCL was particularly clear in the Bland and Altman plot (Figure lb), which
showed that, as the average of the 30 and 60 Hz measurement
increased, the difference between 30 and 60 Hz increased in
a linear manner. There was a much smaller range of values
for VSL than VCL, with most measurements falling in the range
40-80 (im/s. Three outlying measurements were contributed by
20
40
60
SO
100
120
Average (of 25 and 30 Hz)
Figure 2. (a) Bland and Altman plot of VSL analysed at 25 and
30 Hz (lower graph). Mean difference between measurements was
7.0 um/s, with a 95% range of differences from -2.8 to 16.8 |im/s.
(b) Bland and Altman plot of VSL analysed at 30 and 60 Hz
(upper graph). Mean difference between measurements was
12.5 Hm/s, with a 95% range of differences from -13.5 to
38.5 uxn/s.
the washed, incubated and pentoxifylline-treated samples from
an individual donor.
A Bland and Altman plot of ALH is shown in Figure 3.
The difference in ALH measured at 25 and 30 Hz was small,
on average 0.2 |im higher at 30 Hz, with the 95% range of
differences falling between -0.94 and 1.34 \xm. This was nonsignificant, with the exception of the measurement in semen,
which registered P < 0.05. The mean difference in ALH
measured at 30 and 60 Hz was a reduction of 0.9 (im, which
was significant within each treatment (P < 0.001).
BCF was highly frame rate dependent (Figure 4), showing
a significant increase between both 25 and 30 Hz (35%) and
30 and 60 Hz (74%) (P < 0.001).
The mean difference between LIN (ratio VSL/VCL) at 25
and 30 Hz was 0.7%. Between 30 and 60 Hz there was a
mean reduction of 8.2% (P < 0.001), as the increase in VCL
was proportionately larger than that of VSL at 60 Hz. A
similar pattern of response was observed for STR (the ratio
VAP/VCL), although the 2% reduction in STR observed
between 30 and 60 Hz was not significant
The effect of frame rate on detection of hyperactivation
A higher proportion of hyperactivation was detected when
analysed at 30 Hz rather than 25 Hz (Figure 5). Overall this
was 5.2%, ranging from -7.4 to 17.8% (Table JH). There was
307
AJtMorris, JJt-T.Coutts and LJlobertson
" "
BCF(Hz)
ALH Gun)
X
* 0 5
3 «-
s.
00
-O5-
m 10"
£
-1.0-
DUt»i«nc»
-2S.D
6
S
10
20
Average (of 30 and 60 Hz)
2S
Average (of 6 0 - 3 0 Hz)
a 25+ 2SD.
Hlraa
. ..
2
4
8
8
10
-2SD
12
Average (of 25 and 30 Hz)
15
20
25
30
Average (of 25 and 30 Hz)
Figure 3. (a) Bland and Altman plot of ALH analysed at 25 and
30 Hz Gower graph). Mean difference between measurements was
0.2 Jim, with a 95% range of differences from -0.94 to 1.34 urn.
(b) Bland and Altman plot of ALH analysed at 30 and 60 Hz
(upper graph). Mean difference between measurements was
-0.9 urn, with a 95% range of differences from -1.62 to -0.18 Jim.
a difference in increase according to treatment which ranged
from 1% for cells in seminal plasma, to 5% for washed cells,
to 10% in the pentoxifylMne-treated group, where a higher
proportion of hyperactivated cells existed. There was a smaller
difference between analysis at 30 and 60 Hz, with a mean
difference of 1.7%. This was only significant at the 5% level
for washed cells.
Discussion
Sperm motility is the result of flagellar bearing. However,
analysis of the flagellar beat is technically difficult and requires
expensive equipment. The present generation of movement
analysers derive values for movement characteristics from
tracks reconstructed using sperm head positions in sequential
video frames. To understand the manner in which the framing
rate will affect the movement parameters it is necessary to
consider how the trajectories are constructed VCL is calculated
from the distance travelled from head point to head point over
the length of the track. The higher the frame rate, the more
accurate the track reconstruction, the greater the perceived
distance covered in the time and therefore the higher the
calculated VCL. The results of this study confirm the observations of Mortimer et aL (1988), with apparent VCL increasing
as the number of frames per second increased.
In this study there was not a good agreement between VSL
308
Figure 4. (a) Bland and Altman plot of BCF analysed at 25 and
30 Hz (lower graph). Mean difference between measurements was
4.8 Hz, with a 95% range of differences from -0.2 to 9.8 Hz.
(b) Bland and Altman plot of BCF analysed at 30 and 60 Hz
(upper graph). Mean difference between measurements was
13.8 Hz, with a 95% range of differences from 6 to 21.6 Hz.
analysed at 25 and 30 Hz nor at 30 and 60 Hz. In both
comparisons, VSL increased at the higher framing rate by 13
and 15% respectively. This was not treatment-dependent since
the same effect was seen in semen and washed cells. Theoretically, the frequency of collection should not affect this movement
parameter since VSL is computed using the distance from first
to last track point. This was the finding of Mortimer et al.
(1988), who reported that VSL was unaffected by the framing
rate. HoweveT, Mortimer et aL selected 'progressively motile'
spermatozoa moving in regular sinusoidal waves, and irregular
patterns such as hyperactivation were deliberately avoided. All
trajectories were plotted for the same length of time (1 s). It
is clear, however, that VSL will be affected by the tracking
time unless the spermatozoon is swimming in a perfectly
straight line. Tracking time is not directly related to framing
rate, but due to limitations of computer memory, maximum
tracking time is different for each parameter set-up. Using the
HTM 2030, trajectories could be reconstructed from a minimum
of 13 head-points to a maximum of 20 points over 0.79 s at
25 frames/s; using the IVOS, trajectories could be reconstructed
from 13 to 30 trackpoints over 1 s at 30 frames/s or 13 to 60
over 0.5 s at 60 frames/s. For a spermatozoon showing a
sinusoidal wave motion, swimming in a straight line, the
calculated VSL would be the same under all three collection
conditions. However, different sections of a non-straight traject-
Study of sperm movement analysed at 25, 30 or 60 Hz
Table III. The mean and SD of differences between movement parameters analysed at 25 and 30 Hz or 30 and 60 Hz plus 95% range of differences (i.e.
±2SD)
30-25 Hz
VCL (um/s)
VSL (um/s)
VAP (um/s)
ALH dim)
BCF (Hz)
LIN (%)
STR(%)
HA (%)
60-30 Hz
Mean difference
SD
-2 SD
+2SD
Mean difference
SD
- 2 SD
+2SD
11.60
7.00
6.10
0.20
4.80
0.70
4.20
5.20
6.30
4.90
5.20
0.57
2.50
2.60
9.20
6.30
-0.70
-2.80
^.30
-0.94
-0.20
-4.50
-14.20
-7.40
23.90
16.80
16.50
1.34
9.80
5.90
22.60
17.80
38.10
12.50
13.40
-0.90
13.80
-8.20
-2.10
1.70
11.90
13.00
3.20
0.36
3.90
3.80
4.50
5.60
14.30
-13.50
7.00
-1.62
6.00
-15.80
-11.10
-9.50
61.90
38.50
19.80
-0.18
21.60
-0.60
6.90
12.90
For abbreviations, see Table II.
HA(%)
-2SJD.
20
40
80
Average (of 30 and 60 Hz)
-2 SO.
0
20
40
SO
80
Average (of 25 and 30 Hz)
Figure 5. (a) Bland and Altaian plot of HA analysed at 25 and
30 Hz flower graph). Mean difference between measurements was
5.2%, with a 95% range of differences from -7.4 to 17.8%.
(b) Bland and Altaian plot of HA analysed at 30 and 60 Hz (upper
graph). Mean difference between measurements was 1.7%, with a
95% range of differences from -9.5 to 12.9%.
ory could display widely different straight-line velocities. Thus
it seems that VSL is highly dependent on the time of tracking
rather than the frame rate.
VAP is a computer-generated smoothed path designed to
minimize the effects of video frequency. In this study it
increased significantly with framing rate in a manner similar
to VCL.
That LIN and ALH were largely unaffected by analysis at
30 rather than 25 Hz was reported by Zhu et al. (1994) and
confirmed here. However, both parameters were affected by an
increase in framing rate to 60 Hz. LIN decreased significantly as
VCL increased proportionately more than VSL between 30
and 60 Hz. This was not the case between 25 and 30 Hz.
Straightness was affected in a similar manner because of the
frame rate effect on VCL and VAP. ALH is computed by the
analysers using two different algorithms, the HTM-2030 using
a 5-point running averaged position and the HTM-IVOS a 9point spatially averaged system. Despite the difference in
algorithm and in frequency there was no significant difference
in calculation of ALH at 25 and 30 Hz. That ALH was
significantly reduced at 60 Hz is presumably due to the distance
of sperm head deviation from the average path being apparently
reduced because of greater sampling frequency.
BCF was the movement parameter most strongly affected
by videoframe rate. The mean population increased markedly
between 25 and 30 Hz (35%), while the increase between 30
and 60 Hz was even greater (74%). BCF is designed to give
information about the frequency of head oscillations about the
average path. It is calculated from the number of times the
curvilinear track crosses the computer-generated average path
in both directions. Davies et al (1992a,b) suggest that BCF,
with ALH and VAP, is poorly modelled using a fixed running
average, and that accurate measurement is only likely for a
spermatozoon showing sinusoidal wave progression. They
suggest that an alternative method (harmonic analysis) would
give a better measure of these parameters for irregular tracks.
The substantial increase in BCF seen here suggests that high
frequency analysis is necessary for accurate measurement of
this parameter.
The movement characteristics which define HA at 30 Hz
have been well documented (Robertson et al, 1988; Mortimer
and Mortimer, 1990; Burkman, 1991). Hyperactivated cells
are generally thought of as a discrete subpopulation within
the general sperm suspension, physiologically identified by
vigorous motion, low forward progression and wide amplitude
flagellar beat CASA systems detect this type of movement by
high VCL, low linearity of progression and wide ALH deviation. In this study, hyperactivation was defined by VCL > 100
p.m/s, with LIN <65% and ALH >7.5 \im. When applied to
samples analysed at 25 Hz, a lower proportion of cells was
categorized as hyperactivated because fewer cells reached the
threshold VCL. This was proportional to the size of the
hyperactivated subpopulation within the treatment, with a
reduction of 1% in diluted semen, 5% in washed cells and
10% in pentoxifylline-stimulated samples.
309
A.R^Monis, J.R.T.Coutts and LJtobertson
There was no significant difference between the numbers
of cells defined as hyperactivated at 30 and 60 Hz when the
same criteria were used. This was despite the fact that the
population mean VCL increased while mean ALH and LIN
were reduced at 60 Hz. This would tend to increase the number
of cells meeting the criteria for VCL and LIN but reduce the
number meeting those for ALH. Therefore, no significant
difference was noted between number of cells categorized as
hyperactivated at the two frequencies, but it was not clear
whether the movement patterns of the cells categorized at
60 Hz were the same as those observed at 30 Hz. Recent work
in which hyperactivated human spermatozoa were tracked
manually at 60 Hz suggested that a modified set of criteria
was more appropriate for identifying hyperactivation at 60 Hz
(Mortimer and Swan, 1994). Because of the equipment limitations, the same number of track points are collected at 60 Hz
for 0.5 s as at 30 Hz over 1 s. Thus, at 60 Hz a shorter
trajectory was reconstructed. In fact a more accurate measure
of the movement is likely to be achieved by constructing a
longer trajectory, especially since hyperactivated spermatozoa
are known to switch from progressive to non-progressive
motility (Burkman, 1990).
In conclusion, this study highlights that different measurements are achieved for the same sperm populations depending
on the videoframing rate of the equipment used. For samples
analysed at 30 Hz rather than at 25 Hz the main effects are
higher values for population average velocity parameters and
detection of a higher proportion of hyperactive cells. When
the frequency is doubled to 60 Hz, all movement characteristics
are affected and the criteria used to define hyperactivation at
30 Hz identify a different subpopulation of cells at 60 Hz.
Acknowledgements
The authors arc most grateful to Sharon T.Mortimer for her advice
and comments in the preparation of this paper. This work was
supported by a grant from Nuffield Hospitals Trust.
References
Aanesen, A. and Benvold, E. (1990) Studies on the requirements for trackpoints
in Cellsoft automated semen analysis. Fertil. Steril., 54, 910-916.
Bland, J.M. and AJtman, D.G. (1986) Statistical methods for assessing
agreement between two methods of clinical measurement. Lancet, L
307-310.
Burkman, L J . (1990) Hyperactivated motility of human spermatozoa during
in vitro capacitation and implications for fertility. In Gagnon, C. (ed.),
Controls of Sperm Motility: Biological and Clinical Aspects. CRC Press,
Boca Raton, FL, pp. 303-329.
Burkman, L J . (1991) Discrimination between non-hyperactivated and classical
hyperactivated motility patterns in human spermatozoa using computerized
analysis. Fertil. Steril., 55, 363-371.
Davies, R.O., Rothman, S.A. and Overstreet, J.W. (1992a) Accuracy and
precision of computer-aided sperm analysis in multicenter studies. Fertil.
SteriL, 57, 648-653.
Davies, R.O., Niswander, P.W. and Katz, D.F. (1992b) New measures of
sperm motion. I. Adaptive smoothing and harmonic analysis. J. AndrvL,
13, 139-152.
Dunphy, B.C., Kay, R., Barratt, C.L.R. and Cooke, I.D. (1989) Quality control
during the conventional analysis of semen, an essential exercise. J. AndrvL,
10, 378-385.
Holt, W.V., Moore, H.D.M. and Hillier, S.G. (1985) Computer-assisted
measurement of sperm swimming speed in human semen: correlation of
results with in vitro fertilization assays. Fertil Steril., 44, 112-119.
310
Katz, D.F. and Overstreet, J.W. (1981) Sperm motility assessment by
videomicrography. Fertil. Steril., 35, 188-193.
Kay, VJ., Coutts, J.R.T. and Robertson, L. (1993) Pentoxifylline stimulates
hyperactivation in human spermatozoa. Hum. Reprod., 8, 727-731.
Mahony, M.C., Alexander, N.J. and Swanson, RJ. (1988) Evaluation of semen
parameters by means of automated sperm motion analysers. Fertil. Steril.,
49, 876-880.
Mortimer, D. (1986) A microcomputer based semi-automated system for
human sperm movement analysis. Clin. Reprod. Fenil., 4, 283-295.
Mortimer, S.T. and Mortimer, D. (1990) Kinematics of human spermatozoa
incubated under capacitating conditions. J. Androl., 11, 195-203.
Mortimer, S.T. and Swan, M.A. (1995) Kinematics of hyperactivated human
spermatozoa analysed at 60 Hz . Hum. Reprod., 10, 873-879.
Mortimer, D., Serres, C , Mortimer, S.T. and Jouannet, P. (1988) Influence of
image sampling frequency on the perceived movement characteristics of
progressively motile human spermatozoa. Gamete Res., 20, 313-327.
Polansky, F.F. and Lamb, EJ. (1988) Do the results of semen analyses predict
future fertility—a survival analysis study. Fertil. Steril., 49, 1059-1065.
Robertson, L., Wolf, D.P. and Tash, J.S. (1988) Temporal changes in motility
parameters related to acrosomal status: identification and characterization
of populations of hyperactivated human sperm. Biol. Reprod., 39, 797-805.
World Health Organization (1987) WHO Laboratory Manual for the
Examination of Human Semen and Semen-Cervical Mucous Interaction.
Cambridge University Press, Cambridge.
Zhu, JJ., Pacey, A.A., Barrett, C.L.R. and Cooke, I.D. (1994) Computerassisted measurement of hyperactivation in human spermatozoa: differences
between European and American versions of the Hamilton-Thom motility
analyser. Hum Reprod., 9, 456-462.
Received on November 28, 1994; accepted on November 25, 1995