using threshold on 3D images to creat mask for training Unet

Question

Aisha Al Ghurabi 2024-4-21

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2109801-using-threshold-on-3d-images-to-creat-mask-for-training-unet

评论： Walter Roberson 2024-4-22

SampleA_128x3_#1.png

I am using this code for 128x128x128 3D mirco CT image

% Open input and output files
fid3DOCT = fopen('SampleA_128x3_#1.raw','r');
fid3DOCTout = fopen('SampleA_128x3_#1_segmented.raw','w');
% Define image dimensions
nx = 128;
ny = 128;
nz = 128;
 
% Read input image
A = fread(fid3DOCT,nx*ny*nz, 'uint8');
A = uint8(reshape(A,nx,ny,nz));
 
 %Define thresholds for segmentation
threshold1 = 48; % Threshold for class 1
threshold2 = 58; % Threshold for class 2
% You can adjust these thresholds as needed
 
% Apply segmentation
outputImage = zeros(size(A)); % Initialize segmented image
outputImage(A <= threshold1) = 1; % Class 1
outputImage(A > threshold1 & A <= threshold2) = 2; % Class 2
outputImage(A > threshold2) = 3; % Class 3
 
% Write segmented image to output file
for i = 1:nz
fwrite(fid3DOCTout, outputImage(:,:,i), 'uint8');
end

but what I get is a black image

I attached the image I am working on it is actually .raw but I saved as .png so I can attach it

Can you please see why the result is black

2 个评论
显示无隐藏无

Walter Roberson 2024-4-21

You might want to switch to using multibandread for the reading section.

Walter Roberson 2024-4-21

You will get out a file that is uint8 1, 2 or 3. That is going to look black unless you use imagesc()

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Thomas 2024-4-21

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2109801-using-threshold-on-3d-images-to-creat-mask-for-training-unet#answer_1445291

在 MATLAB Online 中打开

Your segmentation routine is correct. However, the issue is that the visualization of a uint8 or double composed of index values of [1, 2, 3] will 'appear' black without scaling. For example using your raster image:

im = imread('SampleA_128x3_#1.png');
%Define thresholds for segmentation
threshold1 = 48; % Threshold for class 1
threshold2 = 58; % Threshold for class 2
% Apply segmentation
outputImage = zeros(size(im)); % Initialize segmented image
outputImage(im <= threshold1) = 1; % Class 1
outputImage(im > threshold1 & im <= threshold2) = 2; % Class 2
outputImage(im > threshold2) = 3; % Class 3
% display thresholded image with scaling
imagesc(outputImage);