largedatavisualizationinitiative: results: level sets
|
This is a model of a 12-day-old mouse embryo extracted from an MR scan from the Caltech Biological Imaging Center. The red structures are its brain ventricles. The blue structure is the liver. |
| This image presents a level set segmentation of the structures contained in a series of MRI scans of a developing frog embryo. Our approach is especially suitable for segmenting multiple, sequential scans, because the output of one segmentation may be used as the initialization for the segmentation of the next scan in the time series. The blue structure is the blastocoel. The red structure is the blastoporal lip. The green structure is the archenteron. |
|
 
|
The first image is a Marching Cubes mesh of the regions of isotropic diffusion generated from a diffusion tensor scan of a human brain. The second image demonstrates that we can extract just the ventricles with our segmentation program. This image shows the spatial relationship of the ventricles and the white matter in a human brain. Both of these structures were extracted from the same dataset. 
|
| The first image is a Marching Cubes mesh of the regions of anisotropic diffusion generated from a diffusion tensor scan of a human brain. In the second image we show that applying level set models to the dataset smooths and simplifies the data, picking out major structures. |
 A second diffusion tensor dataset of the human brain is segmented using our technique with approximately the same algorithm parameters, demonstrating the generality of our approach. The first image is a Marching Cubes mesh of the anisotropic regions of the dataset. The second image contains the initialization of the segmentation produced with a flood fill. The final images presents the results of a level set smoothing. |
 A level set model segmented from an electron tomogram of a spiny dendrite. |