Deep neural networks trained for segmentation are sensitive to brightness changes: preliminary results.
Zhe ZhuMustafa R. BashirMaciej A. MazurowskiPublished in: Medical Imaging: Computer-Aided Diagnosis (2021)
Keyphrases
- neural network
- multilayer perceptron
- segmentation algorithm
- trained neural networks
- image segmentation
- region growing
- segmentation method
- level set
- training process
- medical images
- pattern recognition
- shape prior
- gray level
- multi layer perceptron
- trained neural network
- genetic algorithm
- segmented images
- segmentation accuracy
- auto associative
- multi layer
- automatic segmentation
- back propagation
- fuzzy logic
- artificial neural networks
- training set
- motion segmentation
- object segmentation
- recurrent neural networks
- feed forward
- fault diagnosis
- ground truth
- pixel intensities
- word segmentation
- deep learning
- segmentation errors
- optical flow