How to use imwrite from dicom to png

Hi all, I have 4D images. I want to exctract it to png every slice. Below my coding. But Got error. Anyone can help me?

I = dicomread('SPECT BENTUK LIVER.dcm');
I = squeeze(I);
I  = uint8(255 * mat2gray(  I)); 
imwrite(I,'SPECTbnetukLiver.png');

ERROR

Error using writepng (line 23)

PNG image data must be either MxN or MxNx3.

Error in imwrite (line 566)

feval(fmt_s.write, data, map, filename, paramPairs{:});

Answer 1

[Simon Chan]

It may contains more than one image and hence you need to save it one by one:

Following code for your reference:

[I, cmap] = dicomread('SPECT BENTUK LIVER.dcm');
J = squeeze(I);
[~,~,Nz] = size(J);                                                 % Check number of images
for k = 1:Nz
    imwrite(J(:,:,k),cmap,sprintf('SPECTbnetukLiver_%02d.png',k))   % Save image one by one
end

[mohd akmal masud]

I've try, got error

Error using imwrite>parse_inputs (line 628)

The colormap should have three columns.

Error in imwrite (line 440)

[data, map, filename, format, paramPairs] = parse_inputs(varargin{:});

[mohd akmal masud]

my image have 128x128x90 uint16

[Simon Chan]

If they are uint16, the size of the colormap will be 65535x3 and hence not accepted by imwrite.

You may downsample the colormap as follows:

[I, cmap] = dicomread('SPECT BENTUK LIVER.dcm');
J = squeeze(I);
idx = 1:256:65535;
newmap = cmap(idx,:);                                               % Downsample the colormap
[~,~,Nz] = size(J);                                                 % Check number of images
for k = 1:Nz
    imwrite(J(:,:,k),newmap,sprintf('SPECTbnetukLiver_%02d.png',k))   % Save image one by one
end

[mohd akmal masud]

Index in position 1 exceeds array bounds.

[Walter Roberson]

After you do imread(), what is size(cmap) ?

[Simon Chan]

Could you descript the size and format of variables I and J.

I am suppose the third dimension of variable J is storing the number of frames and hence the image contains 2D data and a colormap only.

As the colormap downsample from size of 65536 to 256, I guess the image would also needs to convert from uint16 to uint8. However, the color of the output image may be different from what you expect.

 imwrite(uint8(J(:,:,k)),newmap,sprintf('SPECTbnetukLiver_%02d.png',k))

Moreover, per Walter Roberson mentioned, you may check via imread and size of the variables to give more information about your situation.

[mohd akmal masud]

Sir, this is size I and J.

>> I = dicomread('I10');
>> size(I)

ans =

   258   258     1   176

>> J = squeeze(I);
>> size(J)

ans =

   258   258   176

[mohd akmal masud]

Im wrote my own but have error

clc
clear all

[I, cmap] = dicomread('I10');
dcmImagei = squeeze(I);                                               % Check number of images
for k = 1:176
    dcmImagei(:,:,k)  = uint8(255 * mat2gray(  dcmImagei));
    imwrite(dcmImagei(:,:,k), sprintf('qqqmyGray%d.png', k));
end

ERROR

Unable to perform assignment because the size of the left side is 258-by-258 and the size of

the right side is 258-by-258-by-176.

[Walter Roberson]

    dcmImagei(:,:,k)  = uint8(255 * mat2gray(  dcmImagei(:,:,k)));

[mohd akmal masud]

wow!!..its work..tq so much

[mohd akmal masud]

Bit I have one problem that the picture is black. as below

Should be the image like this

I have try this one

clc
clear all

[I, cmap] = dicomread('I10');
dcmImagei = squeeze(I); % Check number of images
% imshow3D(dcmImagei); %look for the WL andd best picture
idx = 1:256:65535;
newmap = cmap(idx,:);   
for k = 103
       dcmImagei(:,:,k)  = uint8(255 * mat2gray(  dcmImagei(:,:,k)));
    imwrite(dcmImagei(:,:,k), newmap, sprintf('%d.png', k));
end

But got error

Index in position 1 exceeds array bounds.

[Walter Roberson]

Do not combine using mat2gray() with using colormaps. mat2gray() is only to be used when the values are continuous, representing proportional values, but colormaps are for the case where values are arbitrary and discrete, where a value of (say) 48723 might have a completely different color than is associated with value 48724.

"my image have 128x128x90 uint16"

Then use

I = dicomread('I10');
I8 = im2uint8(I);
for k = 1 : size(I8,3)
   imwrite(I8(:,:,[k k k]), sprintf('%d.png', k));
end

The repeated [k k k] effectively converts from 2D uint8 grayscale int 3D uint8 RGB that has the same color.

[mohd akmal masud]

I got to convert the images in dcm. But why the counts in write images is not same as original?

clc
clear all
[I, cmap] = dicomread('I10');
I  = squeeze(I);
imshow3D(I)
dcmImagei = squeeze(I);% Check number of images
 dcmImagei  =  uint8(255*mat2gray(  dcmImagei));

for k = 103

    dicomwrite(dcmImagei(:,:,k), sprintf('%d.dcm', k));
end

this is the original image in 4D images, the counts is 156 in coordinate (151, 123).

But after write image, the count at the same coordinate is changed to 243. Please help me sir

[Walter Roberson]

The fix for that is to use the code that I already posted. Your code is doing the wrong thing.

[mohd akmal masud]

but you coding also give the black picture

[mohd akmal masud]

in your code,

for k = 1 : size(I8,3)
end

what is 3?

[Walter Roberson]

https://www.mathworks.com/help/matlab/ref/size.html

szdim = size(A,dim) returns the length of dimension dim when dim is a positive integer scalar. Starting in R2019b, you can also specify dim as a vector of positive integers to query multiple dimension lengths at a time. For example, size(A,[2 3]) returns the lengths of the second and third dimensions of A in the 1-by-2 row vector szdim.

In other words, size(I8,3) is the number of "pages" (third dimension) that I8 has. You said "my image have 128x128x90 uint16" so the third dimension of that woud be 90 .