Introduction and Acknowledgements
Authors: Danfeng Chen(Western University), Liqun Liu(Western University)
Contact: Danfeng Chen, firstname.lastname@example.org
GraphCutSegment is an extension of 3D Slicer for tumor segmentation using graph cut and star shape algorithm.
Segmentation can be done with only four user clicks.
The results can be saved as label map.
- Load brain volume and find the tumor.
- Add FOUR fiducials around the tumor.(It has to be four fiducials, otherwise it may not work properly.)
One on the first slice where the tumor begins to show, at the center of the tumor.
Two on the middle slice where the tumor area is the largest, at the diagonal cornors of a rectangle which
can contain the tumor. (These two fiducials have to be on the same slice)
One on the last slice where the tumor ends to show, at the center of the tumor.
Just like the example showing in the following pictures.
- Choose right input volume and segment markups. And then click Crop button.
3D Star Shape and 2D Star Shape constraints can be checked as needed.
(Mainly to avoid unconsistant and concave segments, like holes)
Cropping calls the CropVolume module to have the sub-volume only containing the tumor part, as shown in the image below.
- Click Apply button to get the segmentation result.
The result is stored as LabelMap. To choose the relevant LabelMap to let the result show in the volume, as shown in the picture.
- If not satisfy with the result, you can also use the Painter in the Editor Module to edit the LabelMap.
The label of object is 1 and the background is 0.
- After editing, go back to the extension to click Reapply button.
A second segmentation wiil be done based on the edited LabelMap.
"XXX-label_1" is the LabelMap after editing. "XXX-label_2" is the LabelMap of second GraphCutSegment.
Also, the total editing time using Painter will be shown in the widget on the left.
Tumor Sample Data(RegLib_C01_1.nrrd)
Information for Developers & Users
This extension and its code are only for research purpose only.