Problem of increased 3D volume after reconstruction from 2D CT slides

Dear Everyone,

I have a problem of accuracy after reconstructing CT slides to 3D volume.

In this example, the width of the patient's head is about 140mm as seen from the CT slide. After reconstructing to 3D volume, it was found that the width has increased to about 300mm after outputting to STL file.

Here is the sample of 2D slide and an image of the 3D volume after reconstruction :

Here is the code :

for i = 30 : 160 
    info = dicominfo(dirOutput(i).name,'UseDictionaryVR',true);
    imgTemp = dicomread(info);
    imgCropped = imcrop(imgTemp,croppedArea);    
    imgCroppedBin = imbinarize(imgCropped,0.51728); %Binarize the image
    I(:,:,i) = imgCroppedBin;
end

Is = smooth3(I,'gaussian'); 
fv2 = isosurface(Is,0.5);   
fv2 = smoothpatch(fv2,0,3,1,1);  %"Smooth Triangulated Mesh" by Dirk-Jan Kroon
patch(fv2,...
   'FaceColor',[1,.75,.65],...
   'EdgeColor','none');view(35,30); axis off, daspect([1,1,1]);lightangle(45,30);lighting gouraud;

%Create filename according to current timestamp and output 3D model to .STL file
t = datetime('now');
timestamp = datestr(t,30);
fname = '../Desktop/MatLab Demo/TestResult/3Dreconst_';
fext = '.stl';
outfname = strcat(fname,timestamp,fext);
stlwrite(outfname,fv2);

Grateful if anyone can help. Thanks, Ringo

[Robert Roberts]

Are you still trying to solve this issue?

[Ringo Cheung]

Hi Robert, Thanks for your comment and finally I used affine3D and imwarp to resolve this issue: First dig out two pieces of information “PixelSpacing” and “SliceThickness” from the DICOM file, then use this information to create an affine3d object. Then define a transform “tform” using this object:

Sx = info.PixelSpacing(1);
Sy = info.PixelSpacing(1);
Sz = info.SliceThickness;
tform = affine3d([Sx 0 0 0; 0 Sy 0 0; 0 0 Sz 0; 0 0 0 1]);

Finally, we complete the transform by applying MATLAB® imwarp function:

I = imwarp(I,tform);

where I is the 3D volume data previously obtained from the 2D slices.

The problem is resolved.

Thanks, Ringo