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: MIDL (2021)
Keyphrases
- segmentation method
- segmentation algorithm
- image analysis
- test images
- fully automatic
- adaptive thresholding
- image regions
- accurate segmentation
- grey level
- segmented images
- segmentation errors
- edge detection
- low level image segmentation
- microscopic images
- image database
- image data
- unsupervised segmentation
- segmentation accuracy
- medical images
- ground truth
- input image
- image retrieval
- tubular structures
- region growing
- multiple objects
- pixel wise
- intensity images
- enhanced image
- pixel level
- image segments
- object recognition
- contrast enhancement
- perceptual information
- image features
- image segmentation
- piece wise
- microscopy images
- low depth of field
- image segmentation algorithms
- gray value
- low contrast
- bounding box
- automatic segmentation
- image enhancement
- energy function
- graph cuts
- image registration
- multiscale
- three dimensional