Combining Heterogeneously Labeled Datasets For Training Segmentation Networks.
Jana KemnitzChristian F. BaumgartnerWolfgang WirthFelix EcksteinSebastian K. EderEnder KonukogluPublished in: MLMI@MICCAI (2018)
Keyphrases
- supervised learning
- training set
- training dataset
- segmentation algorithm
- manual labeling
- level set
- image segmentation
- multiscale
- segmentation method
- social networks
- recurrent networks
- medical images
- fully automatic
- training samples
- fully labeled
- image analysis
- benchmark datasets
- ground truth labels
- fully unsupervised
- labeled training data
- shape prior
- region growing
- medical imaging
- complex networks
- synthetic datasets
- network structure
- segmentation accuracy
- linear svm
- neural network
- training and test data
- labeled images
- training data
- labeling process
- feature selection
- echo state networks
- edge detection
- test set