clinicadl.networks.nn

ClinicaDL neural networks.

Build your own neural network

MLP	Simple fully-connected neural network (or Multi-Layer Perceptron) with linear, normalization, activation and dropout layers.
ConvEncoder	Fully convolutional encoder network with convolutional, pooling, normalization, activation and dropout layers.
ConvDecoder	Fully convolutional decoder network with transposed convolutions, unpooling, normalization, activation and dropout layers.
CNN	A regressor/classifier with first convolutional layers and then fully connected layers.
Generator	A generator with first fully-connected layers and then convolutional layers.
AutoEncoder	An AutoEncoder with convolutional and fully connected layers.
VAE	A Variational AutoEncoder with convolutional and fully connected layers.

Common neural networks

UNet	UNet, based on U-Net: Convolutional Networks for Biomedical Image Segmentation.
AttentionUNet	Attention-UNet, based on Attention U-Net: Learning Where to Look for the Pancreas.
DenseNet	DenseNet, based on Densely Connected Convolutional Networks.
ResNet	ResNet, based on Deep Residual Learning for Image Recognition.
SEResNet	Squeeze-and-Excitation ResNet, based on Squeeze-and-Excitation Networks.
ViT	Vision Transformer, based on An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale.

From literature

DenseNets

DenseNet121	DenseNet-121, from Densely Connected Convolutional Networks.
DenseNet161	DenseNet-161, from Densely Connected Convolutional Networks.
DenseNet169	DenseNet-169, from Densely Connected Convolutional Networks.
DenseNet201	DenseNet-201, from Densely Connected Convolutional Networks.

ResNets

ResNet18	ResNet-18, from Deep Residual Learning for Image Recognition.
ResNet34	ResNet-34, from Deep Residual Learning for Image Recognition.
ResNet50	ResNet-50, from Deep Residual Learning for Image Recognition.
ResNet101	ResNet-101, from Deep Residual Learning for Image Recognition.
ResNet152	ResNet-152, from Deep Residual Learning for Image Recognition.

Squeeze-and-Excitation ResNets

SEResNet50	SEResNet-50, from Squeeze-and-Excitation Networks.
SEResNet101	SEResNet-101, from Squeeze-and-Excitation Networks.
SEResNet152	SEResNet-152, from Squeeze-and-Excitation Networks.

Vision Transformers

ViTB16	ViT-B/16, from An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale.
ViTB32	ViT-B/32, from An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale.
ViTL16	ViT-L/16, from An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale.
ViTL32	ViT-L/32, from An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale.