An ImageJ Command plugin to calculate the pair-wise distances of nodules in 2-d nodule images. An extension of my other project, Nodule Segmentation. This plugin requires three inputs described below.
A tif file that has ImageOverlay data attached to an image is required. The overlay should contain the locations and outlines of the nodules (or other objects of interest) on the root system. Another image of the same root system that is amenable to segmentation is also required (see the examples folder). Lastly, a model file to use for segmentation. Instructions on making a model file can be found here.
For documentation of the source code, visit the documentation website.
This plugin requires that you have the .tif file that is an output to my previous plugin, Nodule Segmentation.
Additionally, it requires that you can segment the root system using the same image you segmented your nodules from or that you have two images with different lighting that allows you to segment each one, but such that the images overlap.(i.e. the plant isn't shifted over from image to image). This is because the plugin takes the tif file from my previous plugin and overlays the positional location of the nodules onto the root system in this plugin, so the locations need to align from image to image.
First install fiji. Then put all of the jar files in the jar_files folder into the plugins folder of fiji. The plugin is under plugins->Nodule Analysis->Nodule Distances.
Before you can run this plugin, you must generate a .model file for your dataset using Weka's color clustering plugin, and you must run my previous plugin, Nodule Segmentation to get it's output. Note that the Nodule Segmentation github page also has a tutorial explaining how to create a .model file. You'll be following the same procedure for this plugin, but instead attempting to segment the root system instead of the nodules. I have two model files in the model_files folder which can be used on the examples in the Examples folder to try this for yourself. the model files I have may or may not work on a different data set.
-
Create your model file for your data set(instructions above).
-
Install the plugin to your installation of Fiji(ImageJ) (instructions above).
-
Run the plugin at plugins -> Nodule Data.
-
Fill in any required information for the program to run. This includes a tif file, root file, model file, etc.
-
You can assign different characteristics to your red and green nodules if you wish; this was necessary for our research. Otherwise you can simply put "red" and "green" for the red and green attributes.
-
Let the program run.
This Fiji(ImageJ) plugin uses two dependencies that are not found in the public maven repo.
The K-Shortest-Paths dependency is from Yan-Qi found here, with a fork from github user TomasJohansson that allows the user to create graphs programmatically.
The skeletonization dependency is from github user LingDong found here.