OpenCASA Window Size for high fps, sharing experience and asking for discussion/advice

[Felipe Martínez-Pastor]

Dear all,

We are analyzing bull semen and we have noticed a progressive motility of nearly 0. Tracks are not much linear, but it should be higher, and using the commercial software ISAS we are getting much higher numbers.

I noticed that the average STR is very low and deduced that we could have a problem with the Window Size parameter (I believe it is for calculating some kind of moving average) and our current fps, which is pretty high (240). We are using a value of 10, the same than our commercial CASA running at 53 fps.

Once we increased our Window Size, we noticed that VAP decreased (as predicted, it should affect the trajectory smoothing), ALHs increased (slightly, though), and BCF decreased. STR increased and WOB decreased (again, expected), and progressivity increased from almost 0 to 16 (maybe we should revise our thresholds, especially for VAP).

We increased Window Size to 50 (higher values seems to freeze the analysis, but we must check). It looks very high comparing to the default, but this is a similar ratio with fps comparing to the commercial CASA (50/240 vs. 10/53).

Do you believe this is a correct strategy to increase Window Size when you acquire at higher fps? We are considering to stick with something around 45 and maybe compare with the commercial CASA.

Best wishes,

Felipe

[f.martin...@gmail.com]

Hi,
Responding to myself, in case it is helpful for others or spurring
dicussion. I realised I was talking about that with Jesús Yañiz some
time ago and he gave me some helpful hints (selecting a Window Size).
Briefly, if you visualize the tracks, you can get an estimation of the
"wavelength," that is, images between two peaks in the waving
trajectory, making that your Window Size.
In any case, we found that for bull sperm a Window Size of 20 for 240
fps (this is important, camera speed should affect your decision) is
more convenient. However, for boar, a Window Size of 10 seems to be
fine with the same fps, possibly because of the way sperm moves. Bull
sperm were more "wavy."
Window Size affects track smoothing, therefore VAP and ALH
calculations, and indirectly all the derived parameters, especially
STR. With only a Window Size of 10, VAP was too high for bull, STR
dropped, and progressive motility was extremelly low (I mean, <5%)
even when the videos showed fast sperm with mostly straight paths.
Using 20, VAP seems to be calculated better in this species, with STR
like we are used to see in other CASA and more "realistic" progressive
results (we use STR≥70% and VAP≥20 µm/s in both cases).
Just a warning that, even if some parameters seems to be well
justified theoretically you have to play with real videos and adjust
your parameters accordingly.
Best wishes,
Felipe

El mar, 18 feb 2025 a las 13:58, Felipe Martínez-Pastor
(<f.martin...@gmail.com>) escribió:

> --
> You received this message because you are subscribed to the Google Groups "OpenCASA Mailing List" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to opencasa_mailing...@googlegroups.com.
> To view this discussion, visit https://groups.google.com/d/msgid/opencasa_mailinglist/a8bb00b1-9aba-4e7b-aba8-294405aee64fn%40googlegroups.com.