Interactive Volume Segmentation with Threshold Field Painting
summary
An interactive system paints a spatially varying per-voxel threshold field so different places use different isosurface cutoffs, with tools like flood fill, brush tool and eraser plus a smoothing operator and a ghost isosurface detector that flags gradient/normal mismatches. Implementation projects 2D strokes into 3D and progressively grows surfaces. A user study with four neurosurgeons on repeated tasks shows conventional workflows takes ~200 s while this method takes ~50–100 s and preserves segmentation quality.
AI-written summary · about
chapters (11)
- 0:00Introduction
- 1:08Automating
- 2:16Related Work
- 3:52Current Practice
- 4:51Demo
- 11:49Threshold
- 12:38More explanation
- 12:55Technical details
- 14:29Ghost ice surface
- 15:45User study
- 17:06Conclusion
related talks
45mTech Tutorial: Image segmentation
27mOn challenges and considerations for real time AI processing
06m#ProjectDeepFill
16mEye-Trace: Segmentation of Volumetric Microscopy Images with Eyegaze
17mYuZu: Neural-Enhanced Volumetric Video Streaming
49mGraph Based Processing of Big Images
22mSAM 2: Segment Anything in Images & Video
14mAutocomplete Animated Sculpting