I am converting a 3D image to a set of 2D images(slices), but the images I receive are not of good quality, so there is a loss of information. Is there a solution? please help
5 次查看(过去 30 天)
显示 更早的评论
clear all;
close all;
i=niftiinfo('picture.nii');
Va = niftiread(i);
Vmax = max(Va);
Vmin = min(Va);
Va_prime = ((Va-abs(Vmin))./(Vmax-Vmin))*255;%Normalisation -> valeurs entre 0 et 1
dim = size(Va_prime);%Dimensions du volume
volumeViewer(Va)
[sliceX sliceY sliceZ]=size(Va);
frontal = Va_prime(:,:,sliceZ);
sagittal = reshape(Va_prime(sliceX,:,:),[dim(2) dim(3)]);
horizontal = reshape(Va_prime(:,sliceY,:),[dim(1) dim(3)]);
folder='direction file';
for i=1:sliceY
frontal=uint8(Va(:,:,i));
frontal=imrotate(frontal,90);
basefilename=sprintf('image %d.png',i);
fullfilemane=fullfile(folder,basefilename);
imshow(frontal)
imwrite(frontal, fullfilemane);
end
0 个评论
采纳的回答
DGM
2021-6-4
Well here goes. I'm pretty sure the main problems were in the normalization, but there were other issues that could be simplified anyway. I left the unused stuff in there with notes.
% i have no idea what your image is
ni=niftiinfo('brain.nii');
Va = niftiread(ni);
% if Va is 3D, min(Va) will return a 2D array
% since it only minimizes along dim1. is that desired?
%Vmax = max(Va);
%Vmin = min(Va);
% for the global minimum instead (see below why neither are used)
%Vmax = max(Va(:));
%Vmin = min(Va(:));
%Va_prime = ((Va-abs(Vmin))./(Vmax-Vmin))*255; % that's not normalization
Va_prime = mat2gray(Va); % this normalizes independent of image class
%dim = size(Va_prime); % this contains the same information as sliceX/Y/Z
%[sliceX sliceY sliceZ]=size(Va); % so just use them instead
% again, don't really need to use either though
% you're picking only one slice each (the last one)... why?
%frontal = Va_prime(:,:,sliceZ);
%sagittal = reshape(Va_prime(sliceX,:,:),[sliceY sliceZ]);
%horizontal = reshape(Va_prime(:,sliceY,:),[sliceX sliceZ]);
% i don't know the orientation of your image, so i'm guessing which side is which
% but using permute() avoids needing to know array geometry
frontal = Va_prime;
sagittal = permute(Va_prime,[2 3 1]);
horizontal = permute(Va_prime,[1 3 2]);
% just pick which image to output
outpict = frontal;
outpict = rot90(outpict,1); % just rotate the whole thing if necessary
folder='direction file';
for p=1:size(outpict,3) % automatically picks the right number of slices
outslice = outpict(:,:,p);
% picking filenames and stuff
basefilename=sprintf('image %d.png',i);
fullfilemane=fullfile(folder,basefilename);
imwrite(outslice,fullfilemane);
end
% you could put this in the loop, but it slows things way down
% and you'll only see the last slice anyway
imshow(outslice)
You'll have to check the orientation of the saggital/horizontal slicing with respect to your image. You can just swap them if they're flipped.
更多回答(0 个)
另请参阅
类别
在 Help Center 和 File Exchange 中查找有关 Characters and Strings 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!