Posted by
Xavier Draye-2 on
Mar 29, 2006; 8:51pm
URL: http://imagej.273.s1.nabble.com/Root-tp3703253p3703254.html
Hallo Pedro,
All depends on the kind of images that you have or are planning to have. If
you have been able to extract the intact root system from the growing
medium, have access to a transparency scanner to get very well contrasted
images and have enough time to prepare the root material on the scanner to
get rid of most of the ennoying overlappings of your roots, then the
solutions suggested by Michael and Robert will work really well and will
provide you with global statistics on root system morphology. This is
exactly what is implemented in commercial software like WinRhizo, and you
can get similar data using ImageJ.
If you are interested in length distribution among diameter classes, I
would suggest getting a mask of the root system (Threshold), then get a
skeleton and also get a distance map. You can then AND the skeleton and the
distance map to get the skeleton with each pixel value being the radius of
the root at that location. I would then try to discretize the radius in
radius classes, successively select the sub-skeleton that belong to each
class and, for each class, calculate the cumulated perimeter of the
sub-skeleton.
However, if you are more interested in quantifying specific features on
specific roots (which assumes that you have some a priori knowledge on root
morphology and you are therefore able to sample your root system image and
focus on a limited number of roots) and especially if your images contains
a lot of "parasit" objects or have a structured background (like when you
grow plants on agar or you get pictures through a petry dish that get a lot
of scratches...) of if you have a lot of root overlapping, then one
solution is an interactive software that allows you to pick specific roots,
to fine tune the tracing of that root, to collect information and to send
them to some kind of database system (because this way of working will
supply you with a lot of information). SmartRoot is an ImageJ plugin that
does exactly that (in addition to being a root annotation program). We have
used SmartRoot with scanner images of barley, lupin, maize, and arabidopsis
(if you have sufficient resolution - a root should be minimum 5 pixels
thick if you want to get the best results).
I will probably be slow in answering your messages because I am quite
overloaded for the moment, but let me know if you are interested. Probably
what you could do is send me a typical of your images and explain what
measurements you would like to get and I will tell you if SmartRoot would
be helpfull or not.
Cheers,
Xavier.
At 11:39 PM 3/28/2006 +0100, you wrote:
>Dear all,
> Do you know if I can use ImageJ to analyse plant root systems
>morphology (Area, volume and diameter).
>
>Thanks
>
>Pedro
>
>
>
>
>
>_______________________________________________________ Yahoo! doce lar.
>Faça do Yahoo! sua homepage.
http://br.yahoo.com/homepageset.html