I was given neutron tomograms of plant roots. I don't want to analyze them manually, whether they are different or not. For example, counting the number of nodes myself would be problematic. I saw that there is a program ImageJ, but as far as I understand, you need to manually select pieces of roots (branching points, stem thickness and other parameters). Having these data sets, you can probably use some kind of statistical test? Does it make sense to train a neural network that will give me the number of branching points itself? I saw articles on computer vision, but I do not have a goal to invent something of my own. My goal is to use the most effective and fastest way to compare the topology and quantitative characteristics of two tomograms. It is also assumed that we will have about 8 tomograms per root in order to save time. Initially, I thought that if these are 3D images, then we can use some standard parameters for comparing two pictures, but 3D is not an option
@Ahmad Bakir
Thank you! I think it can help me.
I've found these:
1. Augustin, M., Haxhimusa, Y., Busch, W. et al. A framework for the extraction of quantitative traits from 2D images
of mature Arabidopsis thaliana . Machine Vision and Applications 27, 647–661 (2016). https://doi.org/10.1007/s00138-015-0720-z
2. Pagès, Loïc. (2019). Analysis and Modeling of the Variations of Root Branching Density Within Individual
Plants and Among Species. Frontiers in Plant Science. 10.10.3389/fpls.2019.01020.
3. Loïc Pagès, Branching patterns of root systems: quantitative analysis of the diversity among dicotyledonous species,
Annals of Botany, Volume 114, Issue 3, September 2014, Pages 591–598, https://doi.org/10.1093/aob/mcu145
The first article is about a computer vision algorithm, as far as I understand. The last two use ImageJ. In fact, I've seen a few more articles about ImageJ, so it seems that this is indeed the most convenient method.
I was also thinking about teaching a neural network to draw graphs based on plant tomograms, but I don't know how to work with these graphs later. In the sense that I will have a picture of the graph, but I will still need to extract the most necessary parameters from the graph. But at least it is easier for the eye to look at a graph than at a slightly blurry tomogram with artifacts.
- Badges
- Science topic