Shallow Fully-Connected Neural Networks for Ischemic Stroke-Lesion Segmentation in MRI.
Christian LucasOskar MaierMattias P. HeinrichPublished in: Bildverarbeitung für die Medizin (2017)
Keyphrases
- fully connected
- lesion segmentation
- neural network
- magnetic resonance imaging
- multiple sclerosis
- activation function
- magnetic resonance images
- ischemic stroke
- mr images
- mri data
- medical images
- conditional random fields
- scale free
- white matter
- brain mri
- feed forward
- artificial neural networks
- manual segmentation
- information extraction
- magnetic resonance
- clinical trials
- neural network model
- medical imaging
- back propagation
- fuzzy logic
- power law
- neural nets
- hidden layer
- multi channel
- automatic segmentation
- multilayer perceptron
- high resolution
- basis functions
- image segmentation
- feature selection