How to compute 3D surface area from 2D CT images

Hi, I work with 2D slices of CT images (Dicom images). I segmented tumor from all these 20 slices using some segmentation algorithms.

Now, I have 20 slices of segmented CT scans lets say slice1, slice2, ...slice20

Now I have to combine this 20 slices somehow and extract surface area.

I found this link:

https://www.mathworks.com/matlabcentral/answers/184234-how-do-i-determine-the-surface-area-of-a-2-d-surface-in-a-3-d-space

I don't understand how is x-cord,y-cord, and z-cord computed here.

And I think some of this code is repeated and wrong.

Can someone please help me compute the 3D surface area by combing all the individual 2D slices.

Any help is greatly appreciated.

Thanks,

Gopi

Answer 1

[Image Analyst]

Simply call bwperim() on your segmented tumor images, which you say you already have. Then sum up the image

totalSurfaceArea = 0;
for slice = 1 : numSlices
    thisBinaryImage = .......whatever, your segmented image.
    thisPerimeter = bwperim(thisBinaryImage)
    voxelsInThisSlice = sum(thisPerimeter(:));
    totalSurfaceArea = totalSurfaceArea + voxelsInThisSlice;
end

Essentially you're counting up all voxels that are on the outer edge of your tumor and touch other tissue or air or whatever is not tumor.

[Gopichandh Danala]

Thank you for Your answer Image Analyst..I used your code and have a small problem. I will try and explain : I attached a copy of montage image of 8 thisperimeter images. From my understanding of your explanation, it counts voxels on boundaries and takes it as surface area. But it is only counting the voxels on edges of tumor ..as my images won't converge to form a sphere or something I am missing the total surface around these tumor images and only getting the edge voxels. Can I compute the surface voxels of first and last tumor images and add it to totalSurfaceArea to make it like a whole surface/... Please give your suggestions Modify code and help me understand better Thanks Gopi

[Gopichandh Danala]

More Details on this: Here I have pixel spacing so, I computed pixel Area and used it along with number of voxels to get surface area and for edge slices, I considered all pixels in slice and for all in between pixels I used the boundary pixels Below is the code:

slice_thickness = info.SliceThickness;

PixelSpacing = info.PixelSpacing;

pixelArea = PixelSpacing(1, 1) * PixelSpacing(2, 1);

totalSurfaceArea = 0;

for slice = 1 : numel(region)

    thisBinaryImage = region{slice};
    if (slice == 1 || slice == numel(region)) % if end slice
        bwimage = logical(region{slice});
        voxelsInThisSlice = sum(bwimage(:));
    else                                      % else if between slice
       thisBinaryImage = region{slice};
       thisPerimeter = bwperim(thisBinaryImage);
       %permeterimg{slice} = thisPerimeter;
       voxelsInThisSlice = sum(thisPerimeter(:));
    end
    voxelsInThisSliceArea = voxelsInThisSlice * pixelArea;
    totalSurfaceArea = totalSurfaceArea + voxelsInThisSliceArea;
end

Does this make sense to compute the surface area, I was not able to use slice thickness.

Check this and suggest is it is valid and suggest edits and can i make use of slice thickness to compute the surface area.

Thanks Gopi

Answer 2

[Koray Ertan]

I think the above approach may be wrong because you do not account for the slope of the surface when you just sum the voxelized area.

boundary function may be a better fit for this task.

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