A persistent homology-based topological loss for CNN-based multi-class segmentation of CMR.
Nick ByrneJames R. CloughIsra ValverdeGiovanni MontanaAndrew P. KingPublished in: CoRR (2021)
Keyphrases
- multi class
- persistent homology
- morse theory
- topological features
- multi class classification
- computational geometry
- binary classification
- support vector machine
- segmentation algorithm
- object detection
- level set
- region growing
- image segmentation
- cost sensitive
- multiscale
- pairwise
- base classifiers
- multi class boosting
- feature selection
- multiple classes
- medical images
- image analysis
- binary classifiers
- error correcting output codes
- multi class svm
- bounding box
- object segmentation
- shape prior
- mathematical morphology
- training data
- machine learning