SegTools

TODO

• Update all menu paths once moving to scf update site
• Update the update site instructions once moving to the scf update site

About

SegTools is a collection of small utility plugins for 3D segmentation in ImageJ/Fiji. They provide functionality for semi-manual segmentation, conversion between ROIs and binary masks and visualization of segmentations.

Authors

The plugins are being developed at the Scientific Computing Facility of the MPI-CBG Dresden.

Installation

• SegTools is distributed via an Update site in Fiji. To add it, open Fiji, then click the Menu Help > Update ..., then XXXX TODO. Then restart Fiji.
• Some of the plugins depend on the MorpholibJ library.
  • Note that MorpholibJ is distributed under the GNU Lesser General Public License (LGPL).
  • To add this library to your Fiji: Click Help > Update ... > select IJPB-plugins > then select Close > Apply Changes. Then restart Fiji.
• For general instructions on how to follow an update site, check here.

Plugin Descriptions

Semi-manual Segmentation (3D)

Menu path: Plugins > SegTools > Semi-manual Segmentation (3D)

This plugin is a thin wrapper around existing Fiji functionality to do semi-manual segmentation of a single 3D object. The user manually draws ROIs into several image slices. The ROIs are interpolated (built-in Fiji functionality). A 3D binary mask of the object is created and the volume of the region is then computed (using MorpholibJ).

Example use case: You want to measure the volume of a 3-dimensional region which cannot be (easily) segmented automatically.

Usage

Open a 3D image and start the plugin. A window with instructions will appear (don't press OK just yet!).

GUI options:

• Add overlay: Adds an overlay of the created segmentation mask to the original image.
• Compute statistics (MorpholibJ). Computes the (calibrated) volume and other statistics, using the Analyze Regions 3D plugin of MorpholibJ.

Follow the instructions given in the GUI:

• Choose for example the Freehand tool, then draw a ROI into a slice, add it to the ROI manager. Repeat this for multiple slices. Make sure to draw ROIs into the first and last slice of an object and every time that the objects shape changed quite a bit from the previous ROI.
• It can be a good idea to save the ROIs before continuing (ROIManager > More > Save...).
• When done with drawing, right click into the ROI list, and select Interpolate ROIs.
• Verify that the outline of the shape is good in all slices, otherwise correct it (for example with the Brush tool. Then don't forget to Update the ROI in the manager).

Press OK to finish, and the plugin creates a 3D binary mask and, if chosen, computes statistics.

Mask (3D) to ROI Manager ROIs

Menu path: Plugins > SegTools > Mask (3D) to ROI Manager ROIs

This plugin converts a binary 3D (or 2D) mask into a list of ROIs in the ROI manager. The ROIs are associated with slices. The ROI manager is cleared before adding the new ROIs.

ROI Manager ROIs to Mask (3D)

Menu path: Plugins > SegTools > ROI Manager ROIs to Mask (3D)

This plugin creates a 3D binary mask from all ROIs stored in the ROIManager. The plugin works also in 2D but is then equivalent to Edit > Selection > Create Mask.

Usage

Create one or more ROIs and add to the ROI Manager. In 3D, ROIs are typically associated with specific slices. Then start the plugin.

GUI options:

• Target dimensions: Size of the mask that will be created. Should be the same size as the grayscale image to which the mask belongs. If an image is open, the default size is taken from the active image.
• Use calibration from active image: If checked, then copies the calibration information from the current active image.
• Associate ROIs to slices: If checked, then each ROI is drawn into its associated slice, otherwise it is drawn into all slices.
  Examples:
  • Keep checked if the ROIs correspond to an actual 3D shape.
  • Uncheck if you want to repeat a single ROI in all slices.

Create Overlay of Segmentation (3D)

Menu path: Plugins > SegTools > Create Overlay of Segmentation (3D)

Creates an overlay of a 3D (or 2D) segmentation image onto another grayscale image. This is useful to visualize the segmentation.

Usage

Open a segmentation image and the corresponding raw grayscale image (multi-channel raw image works as well). The segmentation image can be either binary (foreground/background) or contain labelled regions (with region values 1,2,3,.....; this can for example be obtained with the Connected Component Labeling plugin in MorpholibJ). Then start the plugin.

GUI options:

• segmentation image: Either binary or labelled regions (see above).
• grayscale image: The corresponding raw image onto which the overlay is added. Must have same width, height and slices as the segmentation image. Can have multiple channels.
• color mode: In which color to draw the overlay. If the segmentation image is a label image, then individual regions can optionally be drawn in different colors ("multicolor" mode). Otherwise, the overlay is always drawn in a single color.

Macros and Scripting

Macros

The SegTool plugins are macro recordable. Recorded example macro code:

Convert ROIs to a binary 3D mask:

run("Mask (3D) to ROI Manager ROIs");

Convert binary 3D mask to ROIs:

run("ROI Manager ROIs to Mask (3D)", "width=256 height=256 slices=129 use associate");

Create an overlay of a segmentation image:

run("Create Overlay of Segmentation (3D)", "segimp=labelimage.tif grayimp=mri-stack.tif colorstr=multicolor");

Note: Semi-manual segmentation is not macro-recordable but this plugin is interactive by its nature.

Scripting languages

For scripting languages (like jython etc. ) the plugins can be recorded similar to the macro language. It can however also be useful to directly make use of the Conversions class (a collection of static utility functions) to convert between ROIs and binary masks, since the results are then returned and not displayed.

Examples (jython):

Create segmentation overlay (recorded script):

from ij import IJ 

# labelledregions.tif and t1-head.tif are open images

IJ.run("Create Overlay of Segmentation (3D)", "segimp=labelledregions.tif grayimp=t1-head.tif colorstr=multicolor");

Convert binary mask to an array of ROIs with Conversions class:

from ij import IJ 
from de.mpicbg.scf.segtools import Conversions 

mask=IJ.getImage(); # binary mask

roiArray=Conversions.RoisFromBinaryMask(mask) # array of rois

Convert ROIs (of the ROI manager) to binary 3D mask with Conversions class:

from ij import IJ 
from ij.plugin.frame import RoiManager
from de.mpicbg.scf.segtools import Conversions 

# grab the ROI manager. it should be filled with ROIs
rm = RoiManager.getRoiManager();

# target size
width=100
height=100
nslices=20

mask=Conversions.BinaryMaskFromRois(rm,  width, height, nslices, True) # True: associate to slices

Note: Alternative function that takes the ROIs in the shape of an array instead exists.