Difference between revisions of "2009 Summer Project Week Lupus Lesion Segmentation"
From NAMIC Wiki
(Added lesion model image) |
|||
Line 5: | Line 5: | ||
Image:lesionHeatMap-PW2009.png|Predicted Heat Map. | Image:lesionHeatMap-PW2009.png|Predicted Heat Map. | ||
Image:lesionThreshold-PW2009.png|Heat Map after thresholding. | Image:lesionThreshold-PW2009.png|Heat Map after thresholding. | ||
+ | Image:LesionInBrain-PW2009.png|Lesion model inside brain. | ||
</gallery> | </gallery> | ||
Revision as of 13:49, 26 June 2009
Home < 2009 Summer Project Week Lupus Lesion SegmentationKey Investigators
- MRN: H. Jeremy Bockholt, Mark Scully
- Isomics: Steve Pieper
Objective
To create an end-to-end Lupus lesion segmentation and analysis pipeline that can work in both an automated and semi-supervised fashion.
See our Roadmap for more details.
Approach, Plan
Our approach involves the combination of morphometric features with a non-linear classifier in order to capture more of the gestalt properties a human rater uses when segmenting Lupus lesions.
Our plan for the project week is to:
- Work with Steve Pieper to implement the wizard interface to the lesion analysis module to make the segmentation process easier.
- Refactor the lesion analysis code and integrate it with the Slicer code repository so it is built with Slicer.
- Meet with the white matter lesion segmentation group to discuss overlap.
Progress
- Charted the logic of the wizard interface and Steve Pieper produced a prototype.
- Made the lesion segmentation module available as a downloadable extension. This obviates the need to add the source code to the Slicer repository.
- Met with the white matter lesion segmentation project group to exchange data and discuss methods.
- Updated tutorial reflecting feedback from dissemination core.
- Improved testing coverage of lesion segmentation tools.