ROI color merge

[Zachary Freyberg]

Dear all,
I am trying to analyze ROIs in a field and determine, how many  
potentially overlap. To date, I have been using the Properties tab in  
ROI Manager to distinguish specific ROI subsets by color. If ROIs  
overlap, unfortunately, the ROI colors do not merge. For example, if  
ROI#1 is blue and ROI#2 is yellow, their region of overlap does not  
appear green.  While I can see which ROIs overlap, in a crowded field  
of hundreds of ROIs it becomes increasingly difficult to do so.  
Therefore, I also want to depict this visually using color merging,  
which I cannot currently do.

Best,
Zach

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

[Gabriel Landini]

On Tuesday 12 Feb 2013 18:05:15 you wrote:
> I am trying to analyze ROIs in a field and determine, how many
> potentially overlap. To date, I have been using the Properties tab in
> ROI Manager to distinguish specific ROI subsets by color. If ROIs
> overlap, unfortunately, the ROI colors do not merge. For example, if
> ROI#1 is blue and ROI#2 is yellow, their region of overlap does not
> appear green.  While I can see which ROIs overlap, in a crowded field
> of hundreds of ROIs it becomes increasingly difficult to do so.
> Therefore, I also want to depict this visually using color merging,
> which I cannot currently do.

I guess that you are expecting too much from the ROIs as these are not RGB
images, but labelled objects.

I think that what you are looking for could be done programmatically using the
region connection calculus:
http://www.dentistry.bham.ac.uk/landinig/software/spatial/rcc8d.html
if you use binary regions instead of ROIs (e.g. fill the ROIs in empty
images).

I do not know how complex the problem is, but I hope that this might give you
some alternative ideas.

Regards

Gabriel

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