Re: Images that can not be Analysed

Posted by Monique Vasseur on
URL: http://imagej.273.s1.nabble.com/Analyze-Plugin-tp3703706p3703714.html

I have DIC neurone images with whom I can not get a good threshold to measure automatically the length of the dendrites.

Here is the small image problem example: a DIC transmitted light image taken on a fluorescence microscope.  Very often, it is easy to do threshold on fluorescence images (bright signal over a black background and no "edge shadowing") but on a DIC (differential Interference Contrast also called Nomarski) image, the "3D look" of the image mimics a kind of side-illumination that makes thin structures visible (dendrites in our case) but makes it very hard to threshold to get automatic measurements on it.  We need to measure the length of curved prolongations (dendrites) which don't have the same "edge shadowing" all along the way - so then the threshold never gives good result: we are always on the side of the track and/or the track is not always continuous - so we have to do everything manually with the mouse which is not easy.

So if someone can overpass this threshold problem of DIC images it would permit us to automate the skeleton detection and the length measurement of the dendrites on the DIC image; and then our final goal measure the distance run by dim fluorescent points along that dendrite taking account of the path followed (not straight distance between beginning and arrival, but distance following the curved path from nucleus to the loci on the dendrite).

If this kind of image interests you, I will send you some.

Many thanks in advance.

Monique Vasseur
Microscopie et imagerie
Département de biochimie
Université de Montréal
tél. [phone] poste 5148

-----Message d'origine-----
De : ImageJ Interest Group [mailto:[hidden email]] De la part de Daniel Mauch
Envoyé : 8 février 2006 15:29
À : [hidden email]
Objet : Images that can not be Analysed

Hi All

I am doing my PhD in medical image processing at the TU- Munich in Germany.
I am searching for people who have images that cannot be analysed.

If you know somebody, it would be very helpful if you could send me the
contact information's or the images.

With kind regards
Daniel Mauch
[hidden email]