Using openCASA on mouse spermatozoa

[Erick J R Silva]

Dear Carlos and colleagues,

I am working on standardizing openCASA settings to run mouse sperm motility analysis. 

To do that, I use *.avi files converted to openCASA specifications using ffmpeg. Files are recorded at 60 frames/sec, 1.5 sec using a HT CEROS II CASA system connected to an Olympus CX-41 microscope with a 4x dark field objective.

Based on my first tries, I changed the openCASA settings to 

Min cell size: 5.0

Max cell size: 100

Progressive motility (STR>%): 50

Progressive motility (VAP>): 50

Minimum vlc (um/s): 10

vcl lower threshold (um/s): 30

vcl maximum threshold (um/s): 180

Frame rate (frames/s): 60

Minimum track lenght (frames): 40

Maximum displacement (um): 27

Window size(frames): 10

I analyzed a video using these tracks (video available on the link below). 

https://drive.google.com/open?id=1v4koXpQce1mDiz53EpvJQvpdvuY9hhUx

Motile and non-motile tracks are being detected pretty well. However, kinematics parameters are quite low than what I get using CEROS II, specially VSL, VAP, VCL, BCF and ALH. 

Of course, that affected trajectories. They were either slow or moderate. 

See average kinematics data obtained:

Motile trajectories  25            

VSL Mean (um/s)  19.569

VCL Mean (um/s)  43.6

VAP Mean (um/s)  23.287

LIN Mean  50.073

WOB Mean 58.991

STR Mean  80.765 

ALH_Mean Mean (um)  0.948

ALH_Max Mean (um)  2.155

BCF Mean (Hz) 15.136

 

Motility: 62.5%

Progressive motility: 0

What could be changed on settings to improve kinematics?

Any help is very much appreciated!

Kind regards,

Erick

[Carlos Alquezar]

Hi Erick!

I have asked you for access the drive file. I will take a look and see how can I help.

[Erick J R Silva]

Hi Carlos!

I noticed the link was not open for download without previous authorization.

I've changed that. Now everyone with the link will be able to download the test file.

Here is the new link

https://drive.google.com/file/d/1v4koXpQce1mDiz53EpvJQvpdvuY9hhUx/view?usp=sharing

Thanks!

Erick

[Carlos Alquezar]

Thanks Erick!

I could download the video.Let me play with it and I will try to figure out why the results are lower than expected. Could you give me some average of motility parameters in mouse, in order to estimate how close/far the results are?

Also, I do not know if you missed it, but I will need to know the microns per pixel rate of your setup. If you have not updated it and you are using the default value, maybe this is the reason why the motility analysis is given you low values. This rate translates distances from pixels to microns, and to know it you can use a micrometer and divide the width/height of a field in microns by same distance in pixels.

Other question, can you record videos with a higher magnification?

[Erick J R Silva]

Dear Carlos,

Thank you for your reply.

Answering your questions: 

1) Average kinematics rates from the experiment we recorded that video were: VAP: 121 um/s; VSL: 90 um/s; VCL: 271 um/s; ALH: 19 um; BCF: 35 Hz

2) Regarding the micron per pixel rate, I will get this information soon (I need to go to the lab to check that). Just if I understood correctly, we need to take a picture of a micrometer ruler using the same objective and then use that image to get a scale in pixels, correct? (We use the same rationale to get a scale bar for a microscopic image). But I am not sure how can I tell the distance in pixels. Could you please give me more info on how to do that?

3) Rodent (mice and rats) sperm are ~120 and 180 um long. So, the typical magnification to record sperm movement for motility analysis is the one shown in the video (4X). We have a 10X objective coupled to our CASA system, but if we use it we can only record few cells (1-4) cells per video. So that's not very helpful for motility analysis. Would you like to try it anyway?

By the way, here is another video with a higher illumination (so sperm cells are way more clear than the previous one).

https://drive.google.com/file/d/1G68YvFhNQ_LjD1TGXDq6PfDEk6wL89eT/view?usp=sharing

Kind regards,

Erick

[Carlos Alquezar]

Hi Erick!

Thank you for the average values.

In order to get the microns per pixel rate, you can follow these steps: http://microscopy.berkeley.edu/courses/dib/sections/04IPIII/IJsetscale.html. In the penultimate screenshot, you will see that, after setting the corresponding distance, you will obtain a "Scale" value in pixels/um (in this case, 2.84 pixels/um). The parameter you need to set in OpenCASA is the inverse (um/pixel), so you have to calculate, using this example, 1/2.84=0.352 pixels/um.  As you mention, the objective should be same you use to obtain the motility videos.

About the 10X objective, thanks for the info. It is not directly related to your problem but talking with other member, I think that using lower magnifications could affect the quality of the motility results in some way. For the moment I am just thinking about it but maybe in the future I could need your help just to record a sample and test some hyphotesis. For now it is ok.

Let me know if you can calculate the ratio and with that measure I will analyse your videos.

Best!

[Carlos Alquezar]

sorry, I would like to say " 1/2.84=0.352 um/pixel"

[Erick J R Silva]

Dear Carlos,

I am sorry my delay to reply your message. It has been quite a busy week.

I figured out how to get the um/pixel ratio based on your link and Katherine Hamil's tips.

However, something is weird. I used a CEROS II system video of a hemocytometer as ruler (video below).

https://drive.google.com/file/d/1Uf93yuIU5vgEsOytKfcY6nsWrTAO96Is/view?usp=sharing

I keep getting a pixel/um ratio= 0.475, which translates to a um/pixel ratio= 2.105.

Isn't that too high?

If I use the um/pixel ratio=2.105 in the OpenCasa settings, I am not getting any track detection or a bunch of tracks in the lower corner of the video (right on top of the blue canvas; if you are using teste2.avi as a template).

So, I am wondering what am I doing wrong. 

Thanks a lot for you help.

Kind regards,

Erick

PS: let me know if you need any mouse sperm video taken with a 10X objective. I can provide them to you.

[Carlos Alquezar]

Hi Erick!

Thank you for the info. And thanks also to Katherine :)

Ok, here are my preliminary results:

The microns per pixel rate seems to be correct, but it should not be a matter of perception, but an assessment. I used the video you sent to me and, assuming that the big square has the size of 1mm, it is correct that the value you get is in the order of the one you mention.

Taking this, I analyzed the teste2 video and the results, as you mention, were a little bit weird. So, thinking about it, and making some tests, I realized that the range of minimun and maximun head size was not set properly. These values are about the area, not the length, and I notice that the range of 120-180 um you mention was not about area but length. The area is measured in microns^2. Making some tests I found that it works with a range between 200-600 um^2, but in the future I think you will be able to adjust it (you can use a higher magnification to get images of fixed cells, and use the OpenCASA's Morphometry module to get a better range).

I attach my results, and also the parameters I used in the settings:

Microns per pixel: 2.105

Min cell size: 200

Max cell size: 600

Progressive motility (STR>%): 50

Progressive motility (VAP>): 50

Minimum vlc (um/s): 10

vcl lower threshold (um/s): 30

vcl maximum threshold (um/s): 180

Frame rate (frames/s): 60

Minimum track lenght (frames): 20

Maximum displacement (um): 20

Window size(frames): 4

The only problem I notice is in the ALH parameter. It seems that it is not in the range you suggested, but first let me know if you agree with the others. For the momnt, try with this setting and let me know if it works now for vsl, vcl, etc. The morphometry of these spermatozoa seems to be very different of the ram sperm, and maybe the algorithm I implemented is not working properly. But it is for the moment only an hypothesis.

Best

[Erick J R Silva]

Dear Carlos,

Thank you for your feedback.

I also reached a similar settings than the one you sent, except by a smaller min head size (150).

I will check the sperm kinematics from this video that we obtained using ceros II system and compare them with openCASA results.

Both ALH and BCF seem the most different parameters, indeed. I must say that ALH is a parameter with a huge variation in the literature (especially when it is used to hyperactivation motility identification). So, I am not surprised it changed.

If you allow me, I would like to make one suggestion on the average motility report: it would be nice to have the total number of trajectories (both motile and immotile) included in the report. Also the %progressive should be taken considering the total number of trajectories (motile and immotile), so it stays lower than %motility.

I will keep you updated about our tests and data. 

Kind regards,

Erick