Re: Looking for tissue segmentation plugin

Posted by TimFeinstein on Jul 05, 2016; 2:25pm
URL: http://imagej.273.s1.nabble.com/Looking-for-tissue-segmentation-plugin-tp5016812p5016816.html

Hi Ard,

Segmenting tissue can be tricky in practice; simple segmentation tools
often have a hard time giving a satisfactory result.  It sounds like your
ideal solution would be a trainable algorithm that uses machine learning
to identify types of tissue.  For example I had an excellent experience
with the Inform software that PE packages with their Vectra pathology
system (no commercial interest).   I believe that you can do something
similar using Trainable Weka Segmentation with Fiji.

Best,


Tim

Timothy Feinstein, Ph.D.
Research Scientist
University of Pittsburgh Department of Developmental Biology





On 7/5/16, 7:34 AM, "ImageJ Interest Group on behalf of A. Jonker"
<[hidden email] on behalf of [hidden email]> wrote:

>In Fiji we have the Color Inspector 3D, the Cell Counter plugin, Color
>Deconvolution and the Color Segmentation option.
>I'm looking for a combo of these; to interactively decide that pixels
>belong to one group of tissue or to another.
>The Color Segmentation is not fine-grained enough as the tissue has
>non-adjacent colour properties, the Color Deconvolution only allows one
>continuous ROI to be analysed (yes, I know, I should have one color in
>the image to find new vectors).
>
>The effect I'm after is a bit like the magic wand selector of Photoshop,
>where one can shift-click a color or alt-click a colour to respectively
>add or remove pixels in an image that have certain colour ranges.
>The colour ranges I'm thinking of are alreay available in the Display
>Mode: Histogram of the Color Inspector. Clicking the balls in the 3D cube
>should ideally be converted to a tissue type selection.
>
>Is such a utility available or could someone give some pointers in this
>field?
>https://na01.safelinks.protection.outlook.com/?url=http%3a%2f%2fwww.f4.fht
>w-berlin.de&data=01%7c01%7ctnf8%40pitt.edu%7ce291464c08384a956c0c08d3a4ca8
>7b9%7c9ef9f489e0a04eeb87cc3a526112fd0d%7c1&sdata=rdhBXr8zKNf6eOp71PvUkDR51
>uIaDav6J%2bd%2fG9hjiq0%3d seems to be off line currently.
>
>Ard
>________________________________
>
>AMC Disclaimer :
>https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2fwww.amc.n
>l%2fdisclaimer&data=01%7c01%7ctnf8%40pitt.edu%7ce291464c08384a956c0c08d3a4
>ca87b9%7c9ef9f489e0a04eeb87cc3a526112fd0d%7c1&sdata=IzociploLNmlxgQBDZpicU
>rdibk3oJjDc8T8w0Z%2b2x8%3d
>
>________________________________
>
>--
>ImageJ mailing list:
>https://na01.safelinks.protection.outlook.com/?url=http%3a%2f%2fimagej.nih
>.gov%2fij%2flist.html&data=01%7c01%7ctnf8%40pitt.edu%7ce291464c08384a956c0
>c08d3a4ca87b9%7c9ef9f489e0a04eeb87cc3a526112fd0d%7c1&sdata=qENvpuGph0WWSEo
>86l76OyH9N3k0UIfPGtHyMRqXlcM%3d

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