Embedded Files
In Fiji-ImageJ we have seen how to conduct quick operations in order to extract meaningful visual structures. However, what happens when you have not one or two images, but hundreds or even thousands of images! Of course, you can write scripts in Fiji (New > Script > Choose Language) but unfortunately Fiji does not have a great Python support. You are still stuck to Python 2, and the current features do not allow you to integrate with other Python goodies (e.g., numpy, scipy, scikit-learn). Surely, devoid of a good Python support Fiji does not look like a scalable application for our purpose.
It gets lot easier if one can automate this process to a reasonably large number of images. And it may be further attractive if you can do this automation without coding (depending on your expertise). This is where CellProfiler comes in.
CellProfiler allows you to construct pipelines by letting you specify your image analysis operations sequentially. For example, you can construct a pipeline like this:
- Load all images with 'tiff' extension
- Group them in "R", "T" and "G" by reading the filenames (the names may indicate specific channels like nucleus, cellular membranes, etc.)
- Apply Otsu thresholding
- Apply refining segmentation
- Extract blobs and compute a series of features
- Save the output images
- Save the statistics in a CSV file and export
The first step is to install CellProfiler on your machine. The most straightforward way to getting started with CellProfiler is to download an existing pipeline (note the .cppline file corresponding to Fruitfly cells) from their tutorial website and import it as follows.
Load a Downloaded Pipeline
Load a Downloaded Pipeline
I recommend this basic pipeline comprising fruitfly cells (Drosophila Kc167). There are many excellent examples of both basic and advanced pipelines on CellProfiler site.
Drag and Drop Input Images
Drag and Drop Input Images
Note the images have specific sets of characters in their filenames, e.g., "D", "F", and "R". They indicate specific channels. We shall use Names and Types Feature to load them appropriately.
Load the Channels
Load the Channels
Click on Update to load the channels.
Pipeline and Running Analysis
Pipeline and Running Analysis
On left you can click on the components of the pipeline (shown in red dotted rectangle) to explore their respective parameter settings. When you are done, click on "View Output Settings" on left-bottom to check/change output directories. Lastly, click on "Analyze Images" to run the analysis.
Initially it is recommended that you understand a template pipeline well and change it to suit your needs. However, it requires some practice to construct a pipeline from scratch. A very good tutorial is here and you can download the data from here. This tutorial is albeit old, and formats not necessarily match, but you will get the big picture of constructing a pipeline from the scratch.
A point worth mentioning is that when you decide to construct a pipeline of your own you can click on the + sign on bottom left to start adding pipelines (see the screenshot above). When you click, a separate window appears where you can choose various tasks and sub-tasks as shown. Once done with this, the last thing that requires fixing is the set of parameters that will appear in the main window.
Page updated
Google Sites
Report abuse