Get Cells
Segment the junction staining to detect the cells apical area.
Choose Get cells
in the main interface to run it. (1)
- Nb: in previous version of FishFeats, this step is called
Get junctions
in the interface
Segmentation process
1/ 3D->2D: The segmentation of the cellular junctions is performed in 2D. First the junction staining will be projected in a 2D plane by taking a local average of the slices around the maximum intensity signal. If the nuclei staining is in the same channel, the signals will be separated first. In that case, when you click on Get junctions
the interface to separate junctions and nuclei will appear and you must execute it before to be able to do the segmentation. A layer called 2DJunctions
will appears when the projection will be finished/loaded.
2/ 2D segmentation: The plugin proposes several options to perform the segmentation from the 2D images of junctions:
- Epyseg: tool for epithelial segmentation from Aigouy et al. 2020.
- CellPose: tool for cellular segmentation from Stringer et al. 2021.
- Load an already segmented file (should be a file of labelled cell). If you choose this option, the plugin will look for the labelled file of cell, named imagename
_cells2D.tif
in the results folder, but you can select another file.
3/ Manual correction: When the computation of the segmentation is finished, fishfeats
will show you the results in a label
layer called Junctions
. You can then perform manual correction if needed before to save the results.
4/ 2D->3D: At the end of the process, when you click on Junctions done
, the pipeline creates the cells from the segmentation. The shape will be the label shape and the cell position in Z will be back-projected into the junction staining. Thus this step can take a few minutes to calculate the back-projection.
2D projection
The junction staining will be segmented in 2D. If you have already calculated the projection previously or with another pipeline, you can load here the image of the projection and directly use it. Else, the pipeline will calculate the projection by looking at local maximum intensities. Click on do projection
to launch the calculation or load the selected file.
The pipeline will by default save the calculated projection in the results
folder, except if you don't check the save projection
option. When this step is done, you can now performs the segmentation of this projected image.
Manual correction
The result of the segmentation is saved and displayed as labelled apical cells: each cell is assigned a unique number (the label) that is put in all the pixels inside the cell surface. The colors reflect the values (labels) of each cell. 0 indicates background pixels (no cell). In fishfeats
the label 1
is reserved to indicate unassigned elements, so should not be used to label cells.
To correct eventual segmentation errors, on the left top panel, you have the napari tools to edit labels and on the right panel additional fishfeats
editing options.
Shortcut/options
See Napari Label layer documentation for more information on the label edition tools available by default in Napari (and accessible in the top left panel of the interface)
F1,F2,F3... | Show/Hide the first, 2nd, 3th.. layer (ordered in the bottom left panel from bottom to top) |
contour |
Labels (cells) can be displayed as filled areas (put contour to 0) or only with the contour lines (contour >0). |
show selected |
Display only the current label (the pixels which ahve the value that is currently acitve in the label field). |
L | Show/Hide all the labels value as a text overlay. Or check/uncheck show cellnames in the right panel |
5 | Switch between zoom/moving mode |
V | Show/hide the current layer (the Junctions layer) |
Ctrl+c/Ctrl+d | Increase/Decrease the labels contour width (will be filled if reaches 0) |
2 | Select drawing modes (or click on ). When it's active, it will draw with the current brush stroke size |
3 | Select filling mode: it replaces a whole label at one by the current value () |
Select the label under the mouse. Or 4 to switch to picker mode, then click on it | |
Ctrl+ | Delete the whole label (cell) below the click |
Ctrl+ | Merge two neighboring labels into one cell. You should keep the mouse button clicked when dragging from one to another. |
Relabel |
Reorder the cell names (labels) with consecutive values from 2 to the number of cells |
M | Draw a new cell: select unused value and switch to drawing mode |
preserve labels |
If checked, other labels than the currently active one cannot be modified when you are drawing (so even if you touch them they will not be edited) |
Save corrected results
To save the manual corrections, click on save junctions
. It will save a file called imagename_cells2D.tif
in the results
folder that contained the labelled cells.
Measures
You can display a table of measurements of the cells position, area and label. Click on the button Show measures
to perform the measurement. A new window containing the table of all the cells and their area will appear.
This table will be automatically saved when the cell segmentation is saved, in the imagename_results.csv
output file and will be completed during the pipeline by other measurements (cytoplasmic measures, nuclei measure, RNA counts...)
Junctions analysis finished
When you have finished the segmentation and manual correction steps, click on Junctions done
to quit this step and go back to the main step choices.
If save when done
is checked, the segmentation will be saved as a .tif
file that can reloaded later, as well as the table of the cell coordinates and area in the imagename_results.csv
output file.