Use 2D ROI slices to create 3D surface?

I have 15 MR image slices of my heart (left-ventricle) that I have contoured using drawfreehand. I would like to line them up as they would be in my body, and then create a surface model of the ventricle. I've already experimented with several ways of doing this:

I tried to plot them all as polygons with fill3, which looks good, but I want to connect these contours together:

I've also tried reorganizing the data and plotting it as lines using plot3:

It makes a 'groovy' picture, but also not what I wanted. I have also of course tried surf:

but, I keep getting this garbage. I can see why I'm getting that, but I would like help getting something that looks more like a left ventricle (kind of a cone/volcano looking shape). I tried using meshgrid, but I'm not very good at 3D plotting, so I could just be doing something wrong. I've also tried triangulation and a 3d surface fit, which both gave me a mess and connected points from opposite sides of the volume. Can anyone help me get on the right track?

Answer 1

[Suvansh Arora]

One way to do this is the following example. However instead of using ‘imread’ with ‘me.jpg’, the data can be used.

I = imread('me.jpg'); % load in image 
imshow(I) 

% Figure 1 is the original image 
f = figure 
level = graythresh(I); 
BW = im2bw(I,level); 
imshow(BW) 

% Figure 2 is the threshold image 
% Create the stacked image by stacking the binary image repeatedly 
f = figure; 
binaryFile = [] 
for i = 1:200 
    binaryFile(:,:,i) = BW; 
end 
% Sample the binary file 
binaryFile = binaryFile(1:5:end, 1:5:end, 1:5:end) 

% Smooth the image 
binaryFile = smooth3(binaryFile); 
[x,y,z] = meshgrid(1:size(binaryFile,2),1:size(binaryFile,1),1:size(binaryFile,3)); 
[faces,verts] = isosurface(x,y,z,binaryFile,0.5); 

% Creates a surface between 1's and 0's. 
p = patch('Vertices', verts, 'Faces', faces, ...  
    'FaceColor','interp', ...  
    'edgecolor', 'interp'); 

% Remove the color 
p.FaceColor = 'none'; 

% Leave color on the edges 
p.EdgeColor = 'red'; 

% Figure 3 is figure 2 stacked repeatedly 
colormap jet 
axis equal 
view(3) 

Kindly go through the following links for reference:

• Registering Multimodal 3d medical images: https://www.mathworks.com/help/images/registering-multimodal-3-d-medical-images.html
• Converting 2d Image to 3D: https://www.mathworks.com/matlabcentral/answers/154714-how-can-i-convert-2d-images-to-a-3d-image
• Model Construction from CT slices: https://www.mathworks.com/matlabcentral/answers/182511-model-reconstruction-from-ct-slices
• ROI Based processing: ROI-Based Processing - MATLAB & Simulink

[Alexander Moody]

Thank you! This is a thorough answer, and it's much appreciated.