Skip to content

Phase Contrast Cell Analysis Tool (Trainable WEKA Segmentation)

Volker edited this page Sep 10, 2020 · 2 revisions

The tool allows to segment cells in non fluorescent microscopy images using the trainable WEKA segmentation. It allows to run a preprocessing that crops and converts images, to apply a classifier created with the Trainable Weka Segmentation plugin to a folder containing images and to open the images in a folder as a stack in the "Trainable Weka Segmentation plugin" to create a classifier.

You can find some example images here.

You can find an example classifier here 1_echantillon_quantif.model. You need to save the classifier into a folder models in your FIJI base folder.

Getting started

To install the tool, drag the link MRI - Phase Contrast Cell Analysis Toolto the FIJI launcher window, save it under macros/toolsets in the FIJI installation.

Select the "MRI - Phase Contrast Cell Analysis Tool" toolset from the >> button of the FIJI launcher.

weka-tooset.png

  • The first button (the one with the image) opens this help page
  • The p-button runs the pre-processing (crop and convert the image type)
  • The s-button runs the segmentation, applying a classifier from trainable the weka-segmentation
  • The c-button opens the trainable weka plugin in order to create a classifier using your training images

Preprocessing

The pre-processing allows to crop the images in a folder and to convert them to 8-bit. Press the p-button to run the pre-processing. A dialog asking for the folder containing the images will be opened. The result images will be written to a subfolder "cropped" of the input folder.

Options

pre-processing-options.png

  • x, y, width and height define the rectangle to which the image will be cropped
  • select convert to 8bit if you want to convert the images to 8-bit.

Segmentation

Press the s-button to run the segmentation. A dialog that allows to select the folder containing the input images will be opened. Select the pre-processed images. A selection will be saved on each input image.

Options

select_cells_options.png

  • Select the classifier you want to use from the list. The classifiers are taken from the folder models in the FIJI base folder
  • exclude small objects - if selected objects smaller than a minimum size will be excluded from the result
  • min. size - if exclude small objects is selected objects smaller than min. size are excluded from the result
  • fill holes - if selected holes in detected objects are filled.

Create Classifier

The c-button allows to open the weka-plugin to create a new classifier. A dialog will be opened. Select the images that you want to use for the training of the classifier. They will be opened in the weka-plugin as a stack. When finished, save the classifier in the folder models under the FIJI base folder.

Results

result01.png result02.png

Publications citing the tool

  1. Le Borgne-Rochet, M., Angevin, L., Bazellières, E., Ordas, L., Comunale, F., Denisov, E.V., Tashireva, L.A., Perelmuter, V.M., Bièche, I., Vacher, S., et al. (2019). [P-cadherin-induced decorin secretion is required for collagen fiber alignment and directional collective cell migration](https://jcs.biologists.org/content/132/21/jcs233189. J Cell Sci 132, jcs233189).

  2. Planchon, D., Rios Morris, E., Genest, M., Comunale, F., Vacher, S., Bièche, I., Denisov, E.V., Tashireva, L.A., Perelmuter, V.M., Linder, S., et al. (2018). MT1-MMP targeting to endolysosomes is mediated by upregulation of flotillins. Journal of Cell Science 131, jcs218925.

See also

  1. Analyze Spheroid Cell Invasion In 3D Matrix

Other solutions (third party) for the problem

  1. PHANTAST-FIJI-plugin
Clone this wiki locally