matlab code for liver segmentation

Hello evryone can anybody share the matlab code for liver segmentation on CT images.

so that it is very useful to my research

[Rik]

Have a read here and here. It will greatly improve your chances of getting an answer.

[Manjunath R V]

went through that but i have not got any information

[Rik]

Well, I doubt people are going to do your project for you. The links I provided will help you ask a better question. As my response was the only you got in 3 months, you must realize this strategy is not working.

[Star Strider]

The human liver is already partially segmented into lobes! (Anatomy joke! Sorry — I just couldn’t resist!)

[Manjunath R V]

@Rik sure thank you for your response

[Image Analyst]

Did you try my solution below? If my simplistic code below in my Answer didn't work for you, then you can try the more sophisticated methods here:

http://www.visionbib.com/bibliography/medical867l1.html#TT181607

For example, this one looks promising:

Comparison and Evaluation of Methods for Liver Segmentation From CT Datasets,

MedImg(28), No. 8, August 2009, pp. 1251-1265.

IEEE DOI 0909

Survey, Liver Segmentation. Evaluation, Liver Segmentation. BibRef

[Manjunath R V]

@image analyst need to try what u provided i will update once i complete that

i have went through number of papers related to liver segmentation but what exactly i need now is dataset of CT images and liver segmentation matlab code

Answer 1

[Image Analyst]

See attached demos for finding tumors and skulls. Adapt as needed.

Also see my Image Segmentation Tutorial in my file Exchange

Image Analyst's File Exchange

[Manjunath R V]

after executing this i will post my reply thank you for u r response

[Manjunath R V]

i went through executing skull.demo program it is working

[Image Analyst]

Thanks. If it solved your question, could you "Accept this Answer"? If it didn't, post your code and your image.

[Manjunath R V]

it will work oly for the image you provided other images it is not giving result

if u have any code related to liver segmentation let me know sir it will be helpful

[Rik]

  If it didn't, post your code and your image.

[Image Analyst]

We don't have "a "dataset of CT images" so you're on your own for that. Surely your project sponsor has some images. I mean, why else would they hire you to do this project if they didn't have any images?

Secondly you said you read several papers that discussed liver segmentation in the link I gave you. There are hundreds of papers there and I don't have the code for any of them. I imagine that you did what I would have done, and that is to contact the authors and see if there is a way you can obtain their code (like buy it). Assuming you did that and either did not get a response, or the response was that they are not willing to help you, then you're left to write it on your own based on whatever paper you thought was best. Or you could invent your own new and improved method. But again, there is no Liver Segmentation Toolbox in MATLAB, and none of us volunteers have any code for that. If you want, the Mathworks would be delighted to write it for you. Just click on the "Consulting" link on their home page.

Answer 2

[Manjunath R V]

Hello sir this is the Unet code am using to train the 256x256 size CT images.

when i tested the code after training, output image will be same as that of input image. it is unable to segment the liver from the CT image. Here am attaching the code as well as an input and output image please do the needfull.

clc

clear all

close all

load gTruth.mat

% currentPathPixels = "C:\RVM\code_with_image_labl\Unet\PixelLabelData";

% newPathPixels = fullfile('C:\RVM\code_with_image_labl\Unet\liverPixelLabels');

% alternativePaths = {[currentPathPixels newPathPixels]};

% unresolvedPaths = changeFilePaths(gTruth,alternativePaths)

% Load training images and pixel labels into the workspace.

dataSetDir = fullfile('C:\RVM\code_with_image_labl\Unet'); %%Change the the path before run the code

imDir = fullfile(dataSetDir,'Inp_256');

pxDir = fullfile(dataSetDir,'lab256');

% Create an imageDatastore object to store the training images.

Original_imageData = imageDatastore(imDir);

% Define the class names and their associated label IData_set.

classNames = ["background","Abdomen","Liver"];

labelIData_set = [0 1 2];

% labelIData_set = cell(3,1);

% labelIData_set{1,1} = [2;0];

% labelIData_set{2,1} = 1;

% labelIData_set{3,1} = 3;

% Create a pixelLabelDatastore object to store the ground truth pixel labels for the training images.

Segmented_imageData = pixelLabelDatastore(pxDir,classNames,labelIData_set);

% Create the U-Net network.

imageSize = [256 256 1];

numClasses = 3;

Unet_Strct = unetStracture(imageSize, numClasses);

% Create a datastore for training the network for training a semantic segmentation network using deep learning

Data_set = pixelLabelImageDatastore(Original_imageData,Segmented_imageData);

% Set training options.

% % options = trainingOptions('sgdm', ...

% % 'InitialLearnRate',1e-3, ...

% % 'MaxEpochs',10, ...

% % 'VerboseFrequency',10);

options = trainingOptions('sgdm', ...

'Momentum',0.9, ...

'InitialLearnRate',8e-4, ...

'L2Regularization',0.0005, ...

'MaxEpochs',50, ...

'MiniBatchSize',3, ...

'Shuffle','every-epoch', ...

'CheckpointPath', tempdir, ...

'VerboseFrequency',2, ...

'Plots','training-progress');

% Train the network.

newcode_net = trainNetwork(Data_set,Unet_Strct,options)

%Trained data will be saved

save newcode_net newcode_net;

Answer 3

[Manjunath R V]

in the above code if i add encoder depth in the Unet_Strct line like as mentioned below

Unet_Strct = unetStracture(imageSize, numClasses,Encoder depth,'1');

it is segmenting the liver but still improvement is required for reference am attaching input and output images moreover here

it is displaying 22 layers but i want 53 layers unet so needed clarification on this also.

any information provided related to the above is appreciable

[Rik]

Neither of these answers are actually answers. Please post them as comments in the appropriate section. Please also use the formatting tools to make your code more readable.

[Manjunath R V]

thank you for u r suggestion sir