How to add Inception-Res block and Dense-Inception block in 3D U-Net Layers

6 次查看（过去 30 天）
mohd akmal masud
mohd akmal masud 2024-9-10
编辑： Malay Agarwal 2024-9-18
Dear All,
I have the coding below,
lgraph = layerGraph();
tempLayers = [
image3dInputLayer([128 128 128 3],"Name","ImageInputLayer")
convolution3dLayer([3 3 3],16,"Name","Encoder-Stage-1-Conv-1","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Encoder-Stage-1-BN-1")
reluLayer("Name","Encoder-Stage-1-ReLU-1")
convolution3dLayer([3 3 3],32,"Name","Encoder-Stage-1-Conv-2","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Encoder-Stage-1-BN-2")
reluLayer("Name","Encoder-Stage-1-ReLU-2")];
lgraph = addLayers(lgraph,tempLayers);
tempLayers = [
maxPooling3dLayer([2 2 2],"Name","Encoder-Stage-1-MaxPool","Stride",[2 2 2])
convolution3dLayer([3 3 3],32,"Name","Encoder-Stage-2-Conv-1","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Encoder-Stage-2-BN-1")
reluLayer("Name","Encoder-Stage-2-ReLU-1")
convolution3dLayer([3 3 3],64,"Name","Encoder-Stage-2-Conv-2","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Encoder-Stage-2-BN-2")
reluLayer("Name","Encoder-Stage-2-ReLU-2")];
lgraph = addLayers(lgraph,tempLayers);
tempLayers = [
maxPooling3dLayer([2 2 2],"Name","Encoder-Stage-2-MaxPool","Stride",[2 2 2])
convolution3dLayer([3 3 3],64,"Name","Bridge-Conv-1","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Bridge-BN-1")
reluLayer("Name","Bridge-ReLU-1")
convolution3dLayer([3 3 3],128,"Name","Bridge-Conv-2","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Bridge-BN-2")
reluLayer("Name","Bridge-ReLU-2")
transposedConv3dLayer([2 2 2],128,"Name","Decoder-Stage-1-UpConv","BiasLearnRateFactor",2,"Stride",[2 2 2],"WeightsInitializer","he")];
lgraph = addLayers(lgraph,tempLayers);
tempLayers = [
concatenationLayer(4,2,"Name","Decoder-Stage-1-Concatenation")
convolution3dLayer([3 3 3],64,"Name","Decoder-Stage-1-Conv-1","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Decoder-Stage-1-BN-1")
reluLayer("Name","Decoder-Stage-1-ReLU-1")
convolution3dLayer([3 3 3],64,"Name","Decoder-Stage-1-Conv-2","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Decoder-Stage-1-BN-2")
reluLayer("Name","Decoder-Stage-1-ReLU-2")
transposedConv3dLayer([2 2 2],64,"Name","Decoder-Stage-2-UpConv","BiasLearnRateFactor",2,"Stride",[2 2 2],"WeightsInitializer","he")];
lgraph = addLayers(lgraph,tempLayers);
tempLayers = [
concatenationLayer(4,2,"Name","Decoder-Stage-2-Concatenation")
convolution3dLayer([3 3 3],32,"Name","Decoder-Stage-2-Conv-1","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Decoder-Stage-2-BN-1")
reluLayer("Name","Decoder-Stage-2-ReLU-1")
convolution3dLayer([3 3 3],32,"Name","Decoder-Stage-2-Conv-2","Padding","same","WeightsInitializer","he")
batchNormalizationLayer("Name","Decoder-Stage-2-BN-2")
reluLayer("Name","Decoder-Stage-2-ReLU-2")
convolution3dLayer([1 1 1],5,"Name","Final-ConvolutionLayer","Padding","same","WeightsInitializer","he")
softmaxLayer("Name","Softmax-Layer")
pixelClassificationLayer("Name","Segmentation-Layer")];
lgraph = addLayers(lgraph,tempLayers);
% clean up helper variable
clear tempLayers;
lgraph = connectLayers(lgraph,"Encoder-Stage-1-ReLU-2","Encoder-Stage-1-MaxPool");
lgraph = connectLayers(lgraph,"Encoder-Stage-1-ReLU-2","Decoder-Stage-2-Concatenation/in2");
lgraph = connectLayers(lgraph,"Encoder-Stage-2-ReLU-2","Encoder-Stage-2-MaxPool");
lgraph = connectLayers(lgraph,"Encoder-Stage-2-ReLU-2","Decoder-Stage-1-Concatenation/in2");
lgraph = connectLayers(lgraph,"Decoder-Stage-1-UpConv","Decoder-Stage-1-Concatenation/in1");
lgraph = connectLayers(lgraph,"Decoder-Stage-2-UpConv","Decoder-Stage-2-Concatenation/in1");
plot(lgraph);
and the graph like below
But I want to insert the Inception-Res block and Dense-Inception block in my 3D U-Net like picture below
This is Inception-Res block propose
This is Dense-Inception block propose
Anyone can help me?

请先登录，再回答此问题。

采纳的回答

Malay Agarwal
Malay Agarwal 2024-9-18
编辑：Malay Agarwal 2024-9-18
Hi @mohd akmal masud,
You can implement the Inception-Res and Dense Inception-Res blocks by implementing a custom nested deep learning layer.
You can refer to the following resource, which shows how to create a residual block using nested layers: https://www.mathworks.com/help/releases/R2023a/deeplearning/ug/define-nested-deep-learning-layer.html.
I have attached an example implementation of the Inception-Res block to the answer. If you'd like to plot the internal network, you can use the following code:
layer = Inception_Res();
inputSize = [224 224 3];
layout = networkDataLayout(inputSize, "SSC");
layer = initialize(layer, layout);
plot(layer.Network)
Note that the code contains a commented line (line 72) in the initialize() method. I have commented this out since I am not sure about the dimensions of the convolutional operations and initializing the internal network is leading to an error. You'll need to modify the parameters of the convolutional operations to make sure all the dimensions work out.
The Dense Inception-Res block can be created similarly.
Hope this helps!

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Deep Learning Toolbox 的更多信息

标签

产品


版本

R2023a

Translated by