Unable to perform assignment

Question

Eisa Alyaqoub 2020-2-28

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/508021-unable-to-perform-assignment

评论： Eisa Alyaqoub 2020-2-28

采纳的回答： Walter Roberson

在 MATLAB Online 中打开

Hello Helpers,

I am not seeing the error that I have. Could any one please help.

close all;
clc; 
% locate the file
my_path = uigetdir();
files = dir(fullfile(my_path, '*.dcm'));
f_names = {files.name};
info = dicominfo(fullfile(my_path, f_names{1}));
%% read all slices and write them to 3D matrix
ct_scans = zeros(512, 512, length(f_names)); % here where I got the error
for scan = 1:length(f_names)
    slices = fullfile(my_path, f_names{scan});
    ct_scans(:,:,scan) = uint16(dicomread(slices));
end

The eroor:

Unable to perform assignment because the size of the left side is 512-by-512 and the size of the right side is

917-by-888.

I tride to change the inizilation, but still get the same error

ct_scans = zeros(917, 888, length(f_names));

The error:

Unable to perform assignment because the size of the left side is 917-by-888 and the size of the right side is

512-by-512.

1 个评论
显示 -1更早的评论隐藏 -1更早的评论

Eisa Alyaqoub 2020-2-28

I figured out the issue.

The issue is that the first 3 files are 917x888, but the rest files that in the same folder that I am trying to read are 512x512.

Can any one suggest a solution for this problem ?

Thank you,

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Walter Roberson 2020-2-28

2
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/508021-unable-to-perform-assignment#answer_417704

在 MATLAB Online 中打开

   this_scan = uint16(dicomread(slices));
   ct_scans(1:size(this_scan,1), 1:size(this_scan,2), scan) = this_scan;
   

This will put the 512 x 512 into the upper left corner, zero padded. If you happen to encounter a file that is larger than the original size, then the ct_scans array would be extended with zero padding.

Some people would instead imresize() to a consistent size. However, that can distort aspect ratios, and can blur edges.

It is questionable as to whether the smaller scans belong at the upper left of the larger scans. imresize() is not always the right answer either. Sometimes the right answer is to do image registration, to find regions of correspondance even though there might be rescaling or rotation or localized distoration.

3 个评论
显示 1更早的评论隐藏 1更早的评论

Walter Roberson 2020-2-28

在 MATLAB Online 中打开

Before the loop:

ctr = size(ct_scans,1);
ctc = size(ct_scans,2);

In the loop:

   this_scan = uint16(dicomread(slices));
   tr = size(this_scan,1);
   tc = size(this_scan,2);
   roff = floor((ctr - tr)/2);
   coff = floor((ctc - tc)/2);
   ct_scans(roff:roff+tr-1, coff:coff+tc-1, scan) = this_scan;

This would fail if one of the later images was larger than the original.

Eisa Alyaqoub 2020-2-28