2011 Summer Project Week GBM Masks
GBM Segmentation Result achieved with GrowCut under Slicer
GBM Segmentation Result: Slicer (green) and pure manual (blue)
Project Title: Obtaining and evaluation of Glioblastoma multiforme (GBM) segmentation masks
Key Investigators
- BWH: Andriy Fedorov, Rivka Colen, Tina Kapur, Alex Golby, Ron Kikinis
- University Hospital of Marburg: Jan Egger, Christopher Nimsky
Objective
This project aims to obtain Glioblastoma multiforme (GBM) segmentation masks that have been generated by a physician using GrowCut and semi-automated segmentation tools under the medical platform 3D Slicer. Afterwards, the segmentation results are compared via the Dice Similarity Coefficient (DSC) to pure manual slice by slice segmentations.
Approach, Plan
GBM segmentation with Slicer by a trained physician. Measuring segmentation time.
Progress
A physician has been trained in segmenting GBMs with GrowCut and the Editor of Slicer. The trained physician has segmented 10 GBMs with Slicer and the segmentation times have been measured. A first comparison with the pure manual segmentation times has already shown that the segmentation process can be speed up using Slicer – depending on the tumor up to several minutes. The next step will be to compare the Slicer segmentations with the manual slice by slice segmentations via the Dice Similarity Coefficient and write a paper that presents the results in detail for all data sets.
References
- 3D Slicer (http://www.slicer.org/)
- GrowCutSegmentation (https://www.slicer.org/wiki/Modules:GrowCutSegmentation-Documentation-3.6)
- Neurosurgery Department of Marburg (http://www.ukgm.de/ugm_2/deu/umr_nch/umr_nch_team.php?id=767)
- Collaboration Project (http://wiki.na-mic.org/Wiki/index.php/Collaboration:Marburg)
Delivery Mechanism
The results will be provided as a publication.
