Detect boudary of volume

Hi all,

I am trying to create a 3D ROI/Area of the outer boundary from a MRI scan of the Head neck region. Therefore i made it a logical volume and created a code that detects the first and last one in each row and column. But as you might would expect there are some gaps in this boundary which i would like to close. (in other words i want to connect the found coordinates to each other). Can anybody tell me how to do this?

I will place an example image under my code, this is the code i use to detect the boundary:

if true
  % 
z = size (MRIt);
for j = 1: z(3);
  A = squeeze (MRIt(:,:,j));
  [row column] = size(A);
  Edge = row * zeros(row, column); 
  Edge1 = row * zeros(row, column); 

    for i = 1:row;
        B = A(i,1:end);
        [x,y] = find (B, 1, 'first');
        Edge(i,y)= 1 ;
    end

    for i = 1:row;
        B = A(i,1:end);
        [x,y] = find (B, 1, 'last');
        Edge(i,y)= 1 ;
    end

    for i = 1:column;
        B = A(1:end,i);
        [x,y] = find (B, 1, 'first');
        Edge(x,i)= 1 ;
    end

    for i = 1:column;
        B = A(1:end,i);
        [x,y] = find (B, 1, 'last');
        Edge(x,i)= 1 ;
    end

    MRIe(:,:,j) = Edge;
end
  end

Answer 1

[Walter Roberson]

Consider using "morphological closing", http://www.mathworks.com/help/images/ref/bwmorph.html . This would be "dilating" the existing pixels until they flow together to fill the gaps, then eroding back to give you the continuous outline.

Answer 2

[Image Analyst]

You don't want to do it like that. Get rid of all those loops over rows and columns - they're no good as you've already found out. What you should do, if you want to create a binary "shell" of the skull is to call bwperim and assign that to MRIe:

for j = 1: z(3);
  A = MRIt(:,:,j) > 0;
  MRIe(:,:,j) = bwperim(A);
end