Using CycleGANs for effectively reducing image variability across OCT devices and improving retinal fluid segmentation.
Philipp SeeböckDavid Romo-BucheliSebastian M. WaldsteinHrvoje BogunovicJosé Ignacio OrlandoBianca S. GerendasGeorg LangsUrsula Schmidt-ErfurthPublished in: CoRR (2019)
Keyphrases
- image analysis
- segmentation algorithm
- image segmentation
- segmentation method
- grey level
- optical coherence tomography
- multiscale
- image data
- image regions
- segmented images
- image pixels
- single image
- edge detection
- gray value
- test images
- pixel level
- fundus images
- input image
- oct images
- region segmentation
- segmentation accuracy
- watershed algorithm
- image segmentation algorithm
- image segments
- optimal segmentation
- image content
- energy functional
- berkeley segmentation dataset
- region growing
- image retrieval
- gray level images
- homogeneous regions
- energy function
- multiple objects
- adaptive thresholding
- image features
- watershed segmentation
- level set
- foreground and background
- pixel wise
- vector field
- watershed transformation
- segmentation errors
- brain mr images
- high resolution
- region of interest
- retinal fundus images
- optic disc
- retinal images
- contrast enhancement
- automatic segmentation
- image structure
- gray level