RE: [openpiv-users] "Failed to process image pair(s)" error
Nepomnyashchikh, Ivan
Oh my God!
The more I get to know about what you guys do at your biology departments, the less sensitive to the cruelty in this world I become! I am sure I will have nightmares tonight.
By all means, we need to have your code to say what’s wrong with it.
But I feel confident guessing that your main mistakes are the following.
1) It is very clear that the image becomes saturated towards the end of your video. It becomes so bright that there is no way one can distinguish particles there. You must maintain the same brightness of your particles throughout the video taking time. And do not saturate the image. If you see you’ve achieved saturation, either put a neutral density filter on the camera (and I do assume you already use band pass filter for your fluorescent particles), or reduce the aperture or change the sensitivity of your camera if your camera has an intensifier.
2) You cannot use images for PIV when there is flow only at the portion of the image. Of course, theoretically you can do it. But it involves a lot of additional adjustments. In your case, those additional adjustments may be too complicated. So, what you should better do is zoom in your camera on a particular blood vessel. And do it for all the vessels. If that doesn’t work for you (e.g., you want to know the flow fields simultaneously in all the vessels), then take your images as they are now and cut off all the vessels from them. Then do PIV for each separate vessel.
3) I noticed there is a lot of noise in your video. Towards the end of it, some “dirt” appears all over the image. Also, the blood vessels themselves are not absolutely transparent and add their noise. So, first take background images. Then take PIV images. Then subtract backgrounds from your PIV images. And only then do PIV analysis.
4) Before you start doing any imaging, you need to do the calculations. You need to calculate the desired time separation between your PIV frames and the size of the particles to name the most important parameters. Then you do practice experiments (not on the living organisms!) to see if your seeding density is appropriate (you must have about 8 particles per interrogation window), if your particle size is appropriate (a particle must occupy about 2 pixels, not more then 4) and if your time of the frame separation is appropriate (a particle must move a quarter of the interrogation window between the two frames).
4a) I assume you do what’s called microPIV (when you have a confocal microscope and the entire blood flow width is within your depth of focus). Since I am an expert on microPIV, you need to read the German PIV book Raffel et al “Particle image velocimetry” where they talk about microPIV. See if you are missing something.
5) And please, or please, for the love of God, do not torture mice! Go practice doing PIV on something else which is not living. Once you are confident in your PIV skills, you can move to poor mice. I am even ready to offer my help at every step of your PIV journey so that you didn’t torture mice. Just go step by step, send me your code, I will review it, offer my suggestions and so on.
From: 'Cooper Gray' via openpiv-users <openpi...@googlegroups.com>
Sent: Monday, April 28, 2025 3:37 PM
To: openpiv-users <openpi...@googlegroups.com>
Subject: [openpiv-users] "Failed to process image pair(s)" error
[This email originated from outside of OSU. Use caution with links and attachments.]
Hello everyone,
I have been imaging flow of cerebrospinal fluid at the surface of the mouse brain by injecting fluorescent particles into the cerebrospinal fluid. In the attached video, fluorescent particles can be seen flowing into the mouse brain along the brain's blood vasculature via a path called the glymphatic system.
Our imaging is done at 1 Hz over the course of 1-2 hours.
I have 2 questions:
1. I have attached a raw video of the dataset. Does this look at all viable for giving reliable PIV analysis?
2. I have tried processing several sequential tifs from this data set and I always get the same error regardless of how I set up the PIV parameters: "Failed to process image pair(s)". The images that I am collecting are 16 bit images which
I think might be part of the problem? I have attempted to use an openpiv function from the tool module called convert_16bit_tifs but ran into more errors where the function could not process the 2048x2048 tifs.
Does anyone know how to attempt to solve these errors? Any help and insights would be much appreciated
--
You received this message because you are subscribed to the Google Groups "openpiv-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
openpiv-user...@googlegroups.com.
To view this discussion, visit
https://groups.google.com/d/msgid/openpiv-users/24c3b91b-3b2b-4d04-90ba-7e383983fa78n%40googlegroups.com.
