Segmentation without center maximum

C1-S1R1 untreated mpk9-12 DCF (4.3 uM 15 min)_Linear unmixing_Maximum intensity projection.tif (88.6 KB)

I am interested in segmenting out individual chloroplasts from this image. The problem I am finding is that the intensity is brightest around the edges of the structure, instead of some center point.

They are too close to do segmentation by intensity or hk-means. And the way the intensity is arranged seems to interfere with watershed functions. Is there a segmentation algorithm that I’m missing?

I’ve tried what I could find in both Fiji and Icy

What about this?
grafik

// Open your image C1-S1R1 untreated mpk9-12 DCF (4.3 uM 15 min)_Linear unmixing_Maximum intensity projection.tif");
run("Grays");
//optional:
//run("Gaussian Blur...", "sigma=1");
setAutoThreshold("Minimum dark");
setOption("BlackBackground", true);
run("Convert to Mask");
run("Watershed");
1 Like

Hi Heather,

You can segment your image using the Marker Controled Watershed of MorpholibJ:

  1. Threshold your image and use the Analyse particles to get rid of small objects
  2. Duplicate this image and rename “Mask”
  3. Duplicate again and run “ultimate points”, rename as points.
  4. Go to Plugins>MorpholibJ >Segmentation>Marker controlled watershed and use the images appropriately. You’ll get a label image like this:
    image

You can then extract the labels to ROIs if needed using BIOP’s extractor: Plugins>BIOP>Image Analysis>ROIs>Label Image to ROIs

Sincerely,

Matthieu

1 Like

Thank you @phaub & @MatthieuV !

Using the threshold and then converting to a mask helps immensely!