Quantifying Placental Vascular Growth

Hello everybody,

I am a medical PhD student at the department of Obstetrics and I investigate identical twin pregnancies. Identical twins share one placenta and their blood circulation is therefore connected to each other via small vessels on the placental surface. A part of job is injecting placentas with colored dye to make the vessels more visible and to check whether there are connecting vessels between the twins. (The picture below is an injected placenta of a identical twin pregenancy with normal vascular distribution)


In some identical twin pregnancies there is a disproportional distribution of placental mass. This factor may result in poor nourishment of one of the twins, resulting in subsequent poor overall fetal growth. (Below a picture of a placenta with a disproportionate vascular distribution)

For my study I want to quantify this disproportion by measuring the length, amount and complexity of the branches for each twin in these pictures.

Do any of you guys know if there is a way to do this? A specific imageJ plugin?
I have only a little experience in ImageJ. I came across some imageJ plugins like AnalyzeSkeleton, but I am not sure whether it really fits my purpose.

It would be great if anyone of you would have an advice for me. Thank you in advance.

Kind regards,

Lisanne

1 Like

Welcome to the forum, @Lisanne!

You’re on the right track with Analyze Skeleton, that’s what you’ll need in the end to measure the total length of objects in your image. The more challenging step however is the segmentation of the image to separate the four classes of stained vessels from each other and from the background.

One possible way would be:

  • Plugins > Segmentation > Trainable Weka Segmentation
    • Define five classes: Red, Blue, Green, Yellow and Background
    • Draw examples for each class and add them to the labels
    • Train classifier
    • Get probabilty

Here’s a montage of the probability maps that I got when quickly trying with your first image:

  • On the resulting stack of probability map, you can process each slice:
    • Set a threshold (e.g. 0.5) using Image > Adjust > Threshold
    • Analyze > Analyze Particles with a size filter to get rid of small particles
    • Plugins > Skeleton > Skeletonize (2D/3D)
    • Finally, Analzye > Skeleton > Analyze Skeleton (2D/3D) and sum up the lengths of all branches

Hope that helps getting you started.


Alternatively, you can directly try to set different thresholds on the Hue channel after converting your RGB image via Image > Type > HSB Stack. In any case, you should not work with jpg images for your color analysis because the jpg compression will produce artifacts that are most pronounced in the Hue channel.

1 Like

Dear Lissane
acctually i had developed a method to do that, so if you are interested, plese send me and email and we will collaborate.
miroslaba.villalobos@ccadet.unam.mx or famivica@hotmail.com
Fabiola Villalobos

Hi Lissane,
These vessels are beautiful!! May I ask what dyes you are injecting?
My email is mwallingford at tuftsmedicalcenter.org if email communication is easier.
Mary