where can I get pretrained unet, pixelnet, segnet deconvnet matlab codes for CT lung cancer segmentation?

Question

Abel Belay 2024-3-4

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2090291-where-can-i-get-pretrained-unet-pixelnet-segnet-deconvnet-matlab-codes-for-ct-lung-cancer-segmenta

评论： Abel Belay 2024-3-5

This code fragment causes an error in the decoder and output layer. It provides a derror in the transponse layers of decoder and output layer, concatenation layers of the decoder, relu layerS of decoder and outputlayer and additional input missing and no input connection errors correspondigly.

% Define the UNet architecture
% Define input image size
inputSize = [120 150];
% Number of filters in the first layer
numFilters = 64;
% Number of segmentation classes
numClasses = 2;
% Define encoder layers
encoder_layers = [
    imageInputLayer(inputSize)
    
    convolution2dLayer(3, numFilters, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, numFilters, 'Padding', 'same')
    reluLayer()
    maxPooling2dLayer(2, 'Stride', 2)
    
    convolution2dLayer(3, 2*numFilters, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, 2*numFilters, 'Padding', 'same')
    reluLayer()
    maxPooling2dLayer(2, 'Stride', 2)
    
    convolution2dLayer(3, 4*numFilters, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, 4*numFilters, 'Padding', 'same')
    reluLayer()
    maxPooling2dLayer(2, 'Stride', 2)
    convolution2dLayer(1, numClasses, 'Name', 'output')
    softmaxLayer('Name', 'softmax')
    pixelClassificationLayer('Name', 'pixel_class')
    ];
% Define decoder layers
decoder_layers = [
    transposedConv2dLayer(2, 2*numFilters, 'Stride', 2)
    reluLayer()
    concatenationLayer(3, 2, 'Name', 'concat_1'); % Connect to maxPooling2dLayer in encoder
    convolution2dLayer(3, 2*numFilters, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, 2*numFilters, 'Padding', 'same')
    reluLayer()
    
    transposedConv2dLayer(2, numFilters, 'Stride', 2)
    reluLayer()
    concatenationLayer(3, 2, 'Name', 'concat_2'); % Connect to maxPooling2dLayer in encoder
    convolution2dLayer(3, numFilters, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, numFilters, 'Padding', 'same')
    reluLayer()
    
    transposedConv2dLayer(2, numFilters/2, 'Stride', 2)
    reluLayer()
    concatenationLayer(3, 2, 'Name', 'concat_3'); % Connect to maxPooling2dLayer in encoder
    convolution2dLayer(3, numFilters/2, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, numFilters/2, 'Padding', 'same')
    reluLayer()
    ];
outputLayer = [
    concatenationLayer(3, 2, 'Name', 'concat_3') % Connect to maxPooling2dLayer in encoder
    transposedConv2dLayer(2, numFilters/2, 'Stride', 2)
    reluLayer()
    convolution2dLayer(3, numFilters/2, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, numFilters/2, 'Padding', 'same')
    reluLayer()
    
    concatenationLayer(3, 2, 'Name', 'concat_2') % Connect to maxPooling2dLayer in encoder
    transposedConv2dLayer(2, numFilters, 'Stride', 2)
    reluLayer()
    convolution2dLayer(3, numFilters, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, numFilters, 'Padding', 'same')
    reluLayer()
    
    concatenationLayer(3, 2, 'Name', 'concat_1') % Connect to maxPooling2dLayer in encoder
    transposedConv2dLayer(2, 2*numFilters, 'Stride', 2)
    reluLayer()
    convolution2dLayer(3, 2*numFilters, 'Padding', 'same')
    reluLayer()
    convolution2dLayer(3, 2*numFilters, 'Padding', 'same')
    reluLayer()
    
    convolution2dLayer(1, numClasses, 'Name', 'output')
    softmaxLayer('Name', 'softmax')
    pixelClassificationLayer('Name', 'pixel_class')
    ];
output_network = layerGraph(outputLayer);
% Analyze the network to check output sizes
analyzeNetwork(output_network);
Error using analyzeNetwork
This functionality is not available on remote platforms.

2 个评论
显示无隐藏无

Walter Roberson 2024-3-4

You do not use encoder_layers or decoder_layers after you declare them.

It is correct that inside what you have labeled outputLayer that you do not have any output layer.

Abel Belay 2024-3-5

在 MATLAB Online 中打开

this was a code fragment only specifying the layers, I did use them.

layers = [
    encoder_layers
    decoder_layers
%     convolution2dLayer(1, numClasses)
%     softmaxLayer()
 pixelClassificationLayer('Name', 'pixel_class')
];
pixellabel = [0 255];
categories = {'Lung', 'Tumor'};
imdsTrain = imageDatastore('D:\trainingimages');
pxdsTrain = pixelLabelDatastore('D:\masks',categories, pixellabel);
imdsValidation = imageDatastore('D:\validation');
pxdsValidation = pixelLabelDatastore('D:\validationlabel',categories, pixellabel);
trainData = pixelLabelImageDatastore(imdsTrain, pxdsTrain);
valData = pixelLabelImageDatastore(imdsValidation, pxdsValidation);
options = trainingOptions('adam', ...
    'MaxEpochs', 50, ...
    'MiniBatchSize', 16, ...
    'Shuffle', 'every-epoch', ...
    'InitialLearnRate', 1e-3, ...
    'Verbose', true, ...
    'Plots', 'training-progress', ...
    'ValidationData', {imdsValidation, pxdsValidation});
net = trainNetwork(trainData, layers, options);

but the trainNetwork command has caused "unconnected input) in several connections.

请先登录，再进行评论。

请先登录，再回答此问题。

where can I get pretrained unet, pixelnet, segnet deconvnet matlab codes for CT lung cancer segmentation?

2 个评论
显示无隐藏无

回答（0 个）

另请参阅

类别

标签

产品

Community Treasure Hunt

where can I get pretrained unet, pixelnet, segnet deconvnet matlab codes for CT lung cancer segmentation?

2 个评论 显示 无隐藏 无

回答（0 个）

另请参阅

类别

标签

产品

Community Treasure Hunt

2 个评论
显示无隐藏无