Track the liquid flow in an open channel

Hi,
we search for ‏particular geometry for an open channel such that liquid droplets would move forward and not backward (‘liquid diode’).

we have samples ‏within the range of 10 to100 microns width open channels ‏and a liquid is dropped on them so the capillary forces drive the liquid in them.

we need to calculate ‏the front edge velocity and distance.

A link to a movie: https://drive.google.com/open?id=1NRzUtYiJmRSK__0I0PFwzd2ywu1bIcJY

Can you help us find a way to automatically track the liquid flow in this complex geometry?

Thank you,

Hagit and Yonadav

Hello @hagitgilon,

welcome to the ImageJ forum.
Here are a few thoughts:

  • how are the images acquired? With what type of illumination and optics? As much as possible, you should avoid reflections and try to get an even illumination. A LED light ring can help, or, if you’re under a microscope, center your specimen.
  • If you can increase the magnification, do that, you will get more precise results.
  • Try also to align the channels with the horizontal direction (it’s almost good in the movie you posted), your results will be easier to process.
  • Are you allowed to change the color of the liquid? The darker, the better.

Then, processing the movie:

  • when you talk about velocity, how do you define it? I see some kinds of triangular cells of different sizes that get progressively filled, and liquid that propagates along channels between them. Do you need to “just” measure how many cells are filled in a given amount of time, or the instantaneous displacement of the front at each image? And how do you deal when the filling is uncomplete?
  • What I first would try is to process the movie in a manner that compares the first image to each of the following images.
  • Another, yet close approach would be to compare successive images, or images at fixed intervals.
    In both cases, the idea is to exhibit differences relative to a reference state, so that they can be easily segmented. The second method may be more robust.
    And also, some people here do mouse tracking with DeepLabCut. Could that apply to your case?

Many thanks for your prompt and vast reply. we will consider each of the things you mentioned and will reply soon.
Thank you!
Hagit.