HAD-Net: A Hierarchical Adversarial Knowledge Distillation Network for Improved Enhanced Tumour Segmentation Without Post-Contrast Images.
Saverio VadacchinoRaghav MehtaNazanin Mohammadi SepahvandBrennan NichyporukJames J. ClarkTal ArbelPublished in: CoRR (2021)
Keyphrases
- segmentation method
- segmentation algorithm
- test images
- low level image segmentation
- image analysis
- edge detection
- image regions
- fully automatic
- image database
- segmentation errors
- segmented images
- segmentation accuracy
- image segmentation
- enhanced image
- medical images
- adaptive thresholding
- piece wise
- input image
- image slices
- accurate segmentation
- image data
- microscopy images
- microscopic images
- ground truth
- multiple objects
- three dimensional
- unsupervised segmentation
- grey level
- pixel level
- image retrieval
- image features
- image registration
- pixel wise
- image segments
- level set
- image enhancement
- region growing
- color features
- skin lesion
- segmented regions
- graph cuts
- automatic segmentation
- contrast enhanced
- contrast enhancement
- markov random field
- tubular structures
- gray value
- automatically segmented
- perceptual information
- foreground and background
- low depth of field