Displace images from a txt file

Hi. I have a stack of images (4D images). I have a txt file that has 2 columns and 15 rows, same number as the images. Now for every row that corresponds to an image, there should be 2 values (x,y) where x denotes the horizontal position of the image and y denotes the vertical one.

All images have the same x value. Means they are properly aligned horizontally with each other. But, they slightly differ on their y vertical position. Some slices are upper or lower than others by a few mm.

Any code idea of how to do that?

I tried this one,but, the translation is applied to all images and the whole stack is shifted. I want to be able to individually translate slices and not the whole stack in one time.

data = dlmread('imgpositions.txt');
[row, col] = size(data);
x=data(1:row, 1);
y=data(1:row, 2);

translation_vector=[row,col,0,0];
IM2=imtranslate(im2(:,:,:,1),translation_vector(1:3));
for n=2:size(im2,4)
    IM2(:,:,:,n)=imtranslate(im2(:,:,:,n),translation_vector(1:3));
end
if translation_vector(4)~=0
    IM2b=imtranslate(squeeze(IM2(1,:,:,:)),[row col translation_vector(4)]);
    IM2b=permute(IM2b,[4 1 2 3]);
    for n=2:size(IM2,1)
        IM2b(n,:,:,:)=imtranslate(squeeze(IM2(n,:,:,:)),[row col translation_vector(4)]);
    end
    IM2=IM2b;
end

[Stelios Fanourakis]

I though of another solution. When I use result = F(g) I apply interpolation again to g which are the original set points (image) multiplied by translation_vector.

That means that the original image has already been shifted plus a new interpolation coming on the way (F(g)). Does it make it somehow to be smushed or squeezed?

If I visualize with only g without F(g) I get errors later on to the uicontrol of the slider (e.g. Index Exceeds Matrix Dimensions).

Would it be correct to visualize only with g?

[Jan]

Even after reading your pile of other related thread, I do not have an idea what this means:

 When I use result = F(g) I apply interpolation again to g which
 are the original set points (image) multiplied by
 translation_vector.

Think twice. "Would it be correct to visualize only with g?" is not clear enough to be answered also.

Answer 1

[Jan]

Use the debugger to find out, that your translation_vector is exactly the same in each iteration, because it contains the dimension of the imported data only. But I guess, you want to translate by x and y:

data = dlmread('imgpositions.txt');
x    = data(:, 1);
y    = data(:, 2);

for k = 1:size(im2,4)
    IM2(:,:,:,k) = imtranslate(im2(:,:,:,k), [x(k), y(k)]);
end

The part in if translation_vector(4)~=0 is meaningless, because you et this element explicitly to zero. But again translating by:

[row col translation_vector(4)]

cannot be useful, when row and col are the dimension of the imported data, but not the data itself.

Use the debugger to find such problems directly by your own.

[Stelios Fanourakis]

@Jan. F is the griddedInterpolant of the original (non translated) image im2. Right?

g = F.GridVectors

g{i}= g{i}*translation_vector(i) Is the original interpolated image translated by translation_vector.

When at the end, I apply again F(g). It means I apply griddedInterpolation to the translated image (g) once more. And as I said previously, if once the image has been translated and again in the end I apply for a second time the F (griddedInterpolation) doesn't it mean that the image may smush or shifted to the left somehow and won't be perfectly visible? I hope you got my thorough explanation. I am sorry but I cannot make it any easier.

Although, your suggestion gave me a nice color of the grayscale DICOM still they are not translated (see attached)

[Stelios Fanourakis]

@Jan. Any idea of how to do it using cat(3) function and slice()??

[Jan]

@Stelios: "your suggestion gave me a nice color of the grayscale DICOM" - no, my code does not display anything. It is a modification of your code to apply a translation with the values taken from the file. Do you see the problem in your code now?

"still they are not translated (see attached)". I have no idea, why you assume this. After this line of code:

IM2(:,:,:,k) = imtranslate(im2(:,:,:,k), [x(k), y(k)]);

the images are translated. I've tried it with a small test data set.

"F is the griddedInterpolant of the original (non translated) image im2. Right?" - I do not have any information in this thread to decide this.

g{i} = g{i} * translation_vector(i)

Whatever g{i} is, a multiplication with the scalar translation_vector(i) will most likely not perform a translation.

I do not understand any part of the explanations concerning the griddedInterpolation.

[Stelios Fanourakis]

the images are translated. I've tried it with a small test data set.

Well try it, with the images I sent you to see the difference.

As I have explained numerous times. With that code you gave me

IM2(:,:,:,k) = imtranslate(im2(:,:,:,k), [x(k), y(k)]);

The whole stack IM2 is shifted to a certain degree. NOT single slices. To test this out, just try to change the values of the second column in the txt file. Nothing will happen. They remain the same, where, it should change the translation by changing the values of the txt file.

g{i} = g{i} * translation_vector(i)
result = F(g)

That code (see above) gives me the translation result I want to but still struggling to find out why images appear binary.

If only you could give me code examples of how to concatenate 4D images, like merging the translated to the original ones. Or how to use imfuse for 4D images.

That's all I need Jan. Please help.

[Stelios Fanourakis]

@Jan. I promise, you'll get a small surprise if you solve my problem

[Jan]

 The whole stack IM2 is shifted to a certain degree. NOT single slices.

No, I cannot confirm this. After this line:

IM2(:,:,:,k) = imtranslate(im2(:,:,:,k), [x(k), y(k)]);

The 3D slice at index k is translated by x(k) and y(k). If you observe something else, the method of observing is wrong.

im2 = rand(20, 20, 3, 5);
x(k) = 1:10;
y(k) = 10:-1:1;
for k = 1:5
  IM2(:,:,:,k) = imtranslate(im2(:,:,:,k), [x(k), y(k)]);
end

subplot(2,2,1);
image(im2(:, :, :, 1));
subplot(2,2,2);
image(IM2(:, :, :, 1));

subplot(2,2,3);
image(im2(:, :, :, 2));
subplot(2,2,4);
image(IM2(:, :, :, 2));

Works. If you are still convinced, that the code for translating the slices has any problem, please post a piece of code, which reproduces your observation.

 try to change the values of the second column in the txt file.
 Nothing will happen.

Then your code to read the file is wrong - perhaps you are reading a file with the same name, but in a different folder. Or you do not run the code I have posted. Again: you can reveal this simply by using the debugger.

You did not mention, if you understand what the problem of your code was:

translation_vector = [row,col,0,0];

Of course this translates all slices be the number of lines and columns of the data file. But my code considers the values found in this file. Therefore I consider the original question as solved.

 That code (see above) gives me the translation result I want to
 but still struggling to find out why images appear binary.

This becomes a slapstick comedy. You did not answer a pile of questions for clarifications and ignored many explanations and proofs, that the output is not binary. You did not show any evidence for your claim.

Post a simple and running code. Concentrate on the basics, so better use some rand commands to avoid, that a bug in the import of the images causes too much confusion when you think, that the problem is in the processing.

I do not want to get a surprise. It would be fine enough, if your problem is solved. Until now you did not provide a useful description of your problem.

[Stelios Fanourakis]

Jan. First of all. I'd like to make it clear to those who read those comments (if not just you and me), that I have done anything I could to make it as clearer as possible, for someone to understand with a thorough description of my problem and have provided all resources for assessment.

After implementing your example you gave me above (see attached), it is what I say. If you had the figure to scroll the slider, you would see that the image won't go up or down based upon the txt translation. It is remain to a fixed translation, without moving. I want images to move up and down.

I didn't ignore any of the explanations and proof and said that I agree that my image is not converted to binary YET, no one has given me a solution why images do not appear normal (the translation part of GriddedInterpolant is correct though).

This is the most integral code part that translates the images correctly but I or noone here, can solve why images appear black and white.

t=im2double(im2)
F=griddedInterpolant(t)
g=F.GridVectors

for i=1:numel(g)
    g{i} = g{i}*translation_vector(i)
end

result = F(g)
r=uint8(255*result)

[Stelios Fanourakis]

Sorry, this is your image for above example. Ignore the previous one.

[Stelios Fanourakis]

You gave up Jan?