Create Mask from blurred ROI

[MattLee]

Hello,

I'm looking to create distinct ROI's from the attached image using
intensity. From the image it is clear there are 3 cells in the centre
however when I create a Mask it combines these into a single ROI. Altering
the threshold doesn't really help either.

Is there a way to outline each cell using the intensity drop of the pixels
as a sort of Watershed?  The Watershed process requires a binary image, but
converting this image to binary combines them, so is not very helpful.

Thanks.

ROI


Mask

[Michael Schmid]

Hi Matt,

you could try 'Find Maxima', which can be limited to values above the
lower threshold and can do a basic watershed segmentation.
(it is a 'seeded' watershed, starting at the maxima found)

For your thumbnail image, the following seems to work:

   setThreshold(67, 255);
   run("Find Maxima...", "noise=15 output=[Segmented Particles] exclude
above");

For other images, you might need to adjust the noise threshold of 'Find
Maxima' - use the preview function to see how many particles it finds.

HTH,

Michael
________________________________________________________________
On 2016-11-18 17:02, MattLee wrote:

> Hello,
>
> I'm looking to create distinct ROI's from the attached image using
> intensity. From the image it is clear there are 3 cells in the centre
> however when I create a Mask it combines these into a single ROI. Altering
> the threshold doesn't really help either.
>
> Is there a way to outline each cell using the intensity drop of the pixels
> as a sort of Watershed?  The Watershed process requires a binary image, but
> converting this image to binary combines them, so is not very helpful.
>
> Thanks.
>
> ROI
> <http://imagej.1557.x6.nabble.com/file/n5017598/ROI.jpg>
>
> Mask
> <http://imagej.1557.x6.nabble.com/file/n5017598/Mask.jpg>
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html