segment3D.m

function separatedIm=segment3D(imCell, vars)
%% Adam Tyson | 2018-03-26 | adam.tyson@icr.ac.uk
%takes a cell array of individual objects, segments and returns a cell
%array of separated cells

conn_3d=26; % 3D connectivity
imScale=2^16; 

for object=1:length(imCell)

     image=imCell{object};

    % depth correction
     for z=1:size(image,3)
         tmpIm=imScale*double(image(:,:,z));
         im.scaled(:,:,z)=tmpIm/mean(tmpIm(:));
     end

    im.smoothed = imgaussfilt3(im.scaled,vars.smoothSigma);
    
    im.med = medfilt3(im.smoothed, [5 5 5]);
    
    levelOtsu = vars.threshScale*multithresh(im.smoothed);
    im.thresh=im.smoothed;
    im.thresh(im.thresh<levelOtsu)=0;
    im.thresh(im.thresh>0)=1;
    
    for z=1:size(image,3)
        im.closed(:,:,z)=~(bwareaopen(~im.thresh(:,:,z), vars.holeSize));
    end
    
    im.open=bwareaopen(im.closed,vars.noiseRem); % remove small objs
    im.ws=ws3d(im.open, vars, conn_3d); %watershed
    im.open2=bwareaopen(im.ws,vars.noiseRem); % again, after ws
    
    if strcmp(vars.edgeRem, 'Yes')
        im.edgeRem = imclearborder(im.open2, conn_3d); % clear obj at edges
    else
        im.edgeRem=im.open2;
    end
    
    separatedIm{object} = bwlabeln(im.edgeRem); %opening removes labels
    
    clear im
end
end

function separatedImage=ws3d(im3d, vars, conn_3d)
sze=size(im3d);% size of array for initalising
%% external markers
distTran=bwdist(im3d); % distance transform of positive data
wsExt=double(watershed(distTran)); % watershed to generate ext markers
extMark=zeros(sze); % initialise external markers
extMark(wsExt==0)=1; % markers between cells

%% internal markers
negDistTran=bwdist(~im3d);
intMark = imextendedmax(negDistTran, vars.localMaxThresh,conn_3d);

%% impose these minima onto the distance transform and run watershed
watershedBasins = imimposemin(distTran, intMark | extMark);
% watershed again based on basins
finalwatershed=double(watershed(watershedBasins));
separatedImage=im3d.*finalwatershed; % apply ws to thresholded image
end