Posted by
ctrueden on
Nov 12, 2013; 7:02pm
URL: http://imagej.273.s1.nabble.com/Thresholding-cell-blob-tp5005540p5005542.html
Hi Vanessa,
The sample image you sent is very challenging. The cell membranes are low
resolution, with lots of noise both inside and outside the membranes (at
least, what I *think* are the membranes to my untrained eye!).
It is unlikely there will be a series of preprocessing steps and plugins
that can clean up the image to the point where it can be thresholded [1].
So I threw Fiji's "big hammer" at it: the Trainable Weka Segmentation
machine learning plugin. But even that plugin, which is usually pretty
impressive IMO, did a terrible job with my initial attempt:
http://curtis.imagej.net/2013-11-12-trainable-seg.pngSorry for the bad news, but I think you will need to improve your
acquisition protocol if you hope to analyze these structures in the way you
seek. Either that, or draw all ROIs (nearly) 100% manually, which is highly
error-prone and irreproducible from the perspective of quantitative science.
Regards,
Curtis
[1] For details on Fiji's usual approaches to segmentation, see:
http://fiji.sc/SegmentationOn Tue, Nov 12, 2013 at 11:43 AM, Vanessa O'Donnell <
[hidden email]> wrote:
> Hello ImageJ Listserv,
>
> Thank you in advance for your help. I am a new user and am using ImageJ to
> quantify colocalization. I am using Fiji and CoLoc_2 plugin, (I have the
> ROI manager plug-in as well) and am aware of the Bio-format plug-in for
> .lif files. I can check my images for direct information loss and choose
> multiple ROI's in manager.
> I have cells that have bacteria colocalizing with the cell's lysosomes. So
> where I don't know:
>
> 1) threshold out the cell borders. determine ROI (each cell) using tools
> like Blob, threshold, analyze particles etc. This is important because
> bacteria within a certain distance of the cell will be used in study and
> those father away will not be analyzed for colocalization with lysosomes.
>
> Once ROI is determined then I believe I will run analysis on each bacteria
> with in that distance from the cell (blob) to determine what percentage of
> bacteria are colocalizing.after I have the procedure down I would like to
> do this using the Bioformat plug in to use on Z-stacks.
>
> Thanks for any input on this and how to threshold out my cells. I have
> attached a sample image. Thank you again.
>
> Vanessa
>
>
>
>
> --
> ImageJ mailing list:
http://imagej.nih.gov/ij/list.html>
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html