Centroid calculation for connected component in 3D volume

I am trying to implement brain tumor segmentation on 3D brain MRI(.mha data type). After preliminary segmentation, I am applying 26-neighbor connected component algorithm(using bwconncomp) to obtain the largest connected component with maximum volume, following which I need to calculate the centroid of this connected component.

I am not sure if my method of calculating the largest connected component and the centroid is correct, because the centroid value obtained and its nearby voxels all have value 0. Also I dont know how 3D index is represented. For eg. if centroid=(x,y,z), does it correspond to x=row,y=column and z=2D slice?

Any help would be appreciated. Below is my code with the relevant part.

 CC=bwconncomp(Ibin,26);  %Input Black & White 3D data of size 240x240x155
 Pixelid=regionprops(CC,'PixelIdxList');
 [prow pcol]=size(Pixelid);
 maxval=Pixelid(1).PixelIdxList;
 index=1;
 for i=1:prow
  number=numel([Pixelid(i).PixelIdxList]); %calculating the component with max number of voxels
  if(number>maxval)
   maxval=number;
   index=i;
  end
 end
 for i=1:prow
    if i~=index
        Ibin(Pixelid(i).PixelIdxList)=0;
    end
 end
 CC1=bwconncomp(Ibin,26); 
 Cent=regionprops(CC1,'Centroid');

Answer 1

[Walter Roberson]

The Centroid returned can be fractional as it involves a mean.

The documentation of regionprops says for Centroid

"Returns a 1-by-Q vector that specifies the center of mass of the region. The first element of Centroid is the horizontal coordinate (or x-coordinate) of the center of mass, and the second element is the vertical coordinate (or y-coordinate). All other elements of Centroid are in order of dimension."

Note though that x coordinate corresponds to column in MATLAB, so if you want to use the centroid to index you should reverse the first two coordinates

   Centstats = regionprops(CC1,'Centroid');
   Cent = Centstats(1).Centroid;
   crow = round(Cent(2));
   ccol = round(Cent(1));
   cplane = round(Cent(3));
   pixel_nearest_centroid = Ibin(crow, ccol, cplane);

Even so keep in mind that for an area that is not strictly convex, that the centroid can be outside of the region. For example, for a region shaped like

*  *
*  *
****

the centroid is going to have to be between columns 2 and 3 (because it is symmetric around the horizontal axis) and it is going to have to be above the baseline (because there is mass above the baseline but no mass below the baseline), so the centroid would have to be at a location where there is no set pixel in the blob.

[Amala Thampi]

Thanks,that answers my question. So, is there a way I can find a voxel that is central to a volume which is necessarily not convex?I am trying to implement region growing algorithm and I wanted to use the centroid as a seed point, but in my case the centroid is lying outside the volume.

[Walter Roberson]

http://www.mathworks.com/help/images/ref/bwdist.html

Use the two-output form and index the second output at the rounded centroid to find the index of the closest pixel in the region that is set.

[Amala Thampi]

Would that give me the distance to the central point?From what I understood, it gives the distance to the nearest non zero pixel. So if the centroid is outside, then wouldn't that give me the distance to a point on the edge of the volume?Kindly clarify.