added clist gneration counters and progenitor fields. Added segment d…

…eletion and creation features to Edit Segments.

Internal/intFixPos.m

@@ -0,0 +1,17 @@
+function x = intFixPos( x, ss )
+
+
+if x(1) < 1
+    x(1) = 1;
+elseif x(1) > ss(2)
+    x(1) = ss(2);
+end
+
+if x(2) < 1
+    x(2) = 1;
+elseif x(2) > ss(1)
+    x(2) = ss(1);
+end
+
+
+end

cell/trackOptiClist.m

@@ -89,6 +89,9 @@
     index_dlminOld = strcmp(tmpFields,'dlmin');
     index_ehist = find(strcmp(tmpFields,'error_frame'));
 
+
+    prog  = [];
+    prog0 = [];
     % loop through all the images (*err.mat files)
     for i = 1:num_im
         data_c = loaderInternal([dirname,contents(i).name]);
@@ -182,6 +185,57 @@
 
             daughter1_id = drill(data_c.regs.daughterID,'(1)');
             daughter2_id = drill(data_c.regs.daughterID,'(2)');
+            mother_id    = data_c.regs.motherID;
+
+            % get lineage stuff done here
+
+            if i == 1
+                prog =    ID;
+                gen0 =  0*ID;
+                gen  =  0*ID;
+
+            else
+                % copy info from previous IDs
+                prog = nan( size( ID ) );
+                gen0 = nan( size( ID ) );
+                gen  = nan( size( ID ) );
+
+                prog(~birthID_flag) = prog_(ID(~birthID_flag));
+                gen0(~birthID_flag) = gen0_(ID(~birthID_flag));
+                gen(~birthID_flag)  = gen_(ID(~birthID_flag));
+
+                % now take care of new cells
+                mother_list = mother_id(birthID_flag);
+
+                prog_tmp = nan( size( mother_list ));
+                gen0_tmp = nan( size( mother_list ));
+                gen_tmp = nan( size( mother_list ));
+
+                prog_tmp(mother_list>0) = prog_(mother_list(mother_list>0));
+                gen0_tmp(mother_list>0) = gen0_(mother_list(mother_list>0))+1;
+                gen_tmp(mother_list>0)  = gen_(mother_list(mother_list>0))+1;
+
+                ID_ml = ID(birthID_flag);
+                prog_tmp(mother_list==0) = ID_ml(mother_list==0);
+                gen0_tmp(mother_list==0) = nan;
+                gen_tmp(mother_list==0)  = 0;  
+
+                prog(birthID_flag) = prog_tmp;
+                gen0(birthID_flag) = gen0_tmp;
+                gen(birthID_flag)  = gen_tmp;
+            end
+
+            IDmax = max(ID);
+
+            gen0_     = nan( [1,IDmax] );
+            gen0_(ID) = gen0;
+
+            gen_      = nan( [1,IDmax] );
+            gen_(ID)  = gen;
+
+            prog_     = nan( [1,IDmax] );
+            prog_(ID) = prog;
+            % done lineage stuff.    
 
             length1 = drill(data_c.CellA,'.length(1)');
             length2 = drill(data_c.CellA,'.length(2)');
@@ -421,6 +475,9 @@
     {'Neck width'},{'neckWidth'},0,0;
     {'Maximum width'},{'maxWidth'},0,0;
     {'Cell dist to edge'},{'cell_dist'},1,0;
+    {'Generation'},{'gen'},0,0;
+    {'Generation0'},{'gen0'},0,0;
+    {'Progenitor'},{'prog'},0,0;
     ];
 
 

segmentation/defineGoodSegs.m

@@ -1,5 +1,4 @@
-function [data] = defineGoodSegs(data, ws, phaseNorm, C2phaseThresh, ...
-    mask_bg, A, CONST, calcScores)
+function [data] = defineGoodSegs(data, ws, CONST, calcScores)
 % defineGoodSegs sets the segments to good, bad and 3n set by the watershed algorithm
 % "Good" segments (segs_good) are the ones that lie along a real cellular
 % boundary, "bad" segments, lie along spurious boundaries
@@ -11,9 +10,6 @@
 % INPUT :
 %   data : data segmentation frame file
 %   ws : watersehd image
-%   phaseNorm : normalized phase image
-%   C2phaseThresh : c2 (curvature) calculated image
-%   mask_bg : background mask
 %   A : scoring coefficients
 %   CONST : segmentation parameters
 %   caclScores : boolean to calculate scores
@@ -39,6 +35,13 @@
 % along with SuperSegger.  If not, see <http://www.gnu.org/licenses/>.
 
 
+%   phaseNorm : normalized phase image
+%   C2phaseThresh : c2 (curvature) calculated image
+%   mask_bg : background mask
+mask_bg       = data.mask_bg;
+phaseNorm     = data.phaseNorm;
+C2phaseThresh = data.C2phaseThresh;
+A             = CONST.superSeggerOpti.A;
 
 sim = size( phaseNorm );
 

segmentation/intRemovePillars.m

@@ -0,0 +1,63 @@
+function mask = intRemovePillars(phase, CONST)
+
+
+
+if exist( 'CONST' ) && ~isempty( CONST ) && ...
+        isfield( CONST.superSeggerOpti, 'remove_pillars' )
+    radius   = CONST.superSeggerOpti.remove_pillars.radius;
+    cut      = CONST.superSeggerOpti.remove_pillars.cut;
+    Area_Cut = CONST.superSeggerOpti.remove_pillars.Area_Cut;
+    debug    = CONST.superSeggerOpti.remove_pillars.debug; 
+
+else
+    radius = 2;
+    cut = 0.05;
+    Area_Cut = 700;
+    debug = false;
+end
+
+
+[~,~,~,D] = curveFilter(  double(phase), radius );
+
+D = D/max(D(:));
+
+
+D(D<cut) = cut;
+
+ws = logical(watershed(D));
+
+lab = bwlabel(ws);
+
+props = regionprops( lab, {'Area'} );
+
+A = [props.Area];
+
+
+
+mask = ismember( lab, find(A<Area_Cut) );
+
+se = strel('disk',1);
+mask = imdilate(mask,se);
+
+
+if debug
+    figure( 'name', ['intRemovePillars: Area cut: A = ',num2str(Area_Cut)] );
+    clf;
+    comp( {phase}, {lab,'label',A} );
+    drawnow;
+
+
+
+    figure( 'name', ['intRemovePillars: Masked region' ] );
+    clf;
+    comp( {phase}, {mask,'r',.5} );
+    drawnow;
+end
+
+
+
+
+
+
+end
+

segmentation/superSeggerOpti.m

@@ -84,13 +84,13 @@
 
 % Load the constants from the package settings file
 
+%% Handles inputs here
 MIN_BG_AREA     = CONST.superSeggerOpti.MIN_BG_AREA;
 MAGIC_RADIUS    = CONST.superSeggerOpti.MAGIC_RADIUS;
 MAGIC_THRESHOLD = CONST.superSeggerOpti.MAGIC_THRESHOLD;
 CUT_INT         = CONST.superSeggerOpti.CUT_INT;
 SMOOTH_WIDTH    = CONST.superSeggerOpti.SMOOTH_WIDTH;
 MAX_WIDTH       = CONST.superSeggerOpti.MAX_WIDTH;
-A               = CONST.superSeggerOpti.A;
 verbose         = CONST.parallel.verbose;
 
 
@@ -134,6 +134,8 @@
     data.phase = phaseOrig;
 end
 
+%% Start segementation process here
+
 % Initial image smoothing to reduce the camera and read noise in the raw
 % phase image. Without it, the watershed algorithm will over-segment the
 % image.
@@ -150,6 +152,17 @@
 phaseNorm(phaseNorm > (mult_max*mean_phase)) = mult_max*mean_phase;
 phaseNorm(phaseNorm < (mult_min*mean_phase)) = mult_min*mean_phase;
 
+% Filter to remove internal structure from phase image ofcells here 
+% (and little pieces of debris from the background
+if isfield( CONST.superSeggerOpti, 'remove_int_struct' ) && ...    
+        CONST.superSeggerOpti.remove_int_struct.flag
+        tmp_im = imclose( phaseNorm, strel('disk',...
+           CONST.superSeggerOpti.remove_int_struct.radius ));
+       W = CONST.superSeggerOpti.remove_int_struct.weight;
+       phaseNorm = (1-W)*phaseNorm + W*tmp_im;
+end
+
+
 % if the size of the matrix is even, we get a half pixel shift in the
 % position of the mask which turns out to be a problem later.
 
@@ -212,6 +225,14 @@
     adapt_flag=1;
 end
 
+
+% Remove CellASIC pillars here
+if isfield( CONST.superSeggerOpti, 'remove_pillars' ) &&  ...
+    CONST.superSeggerOpti.remove_pillars.flag 
+    mask_to_remove = ~intRemovePillars(phaseOrig,CONST);
+    mask_bg(mask_to_remove) = false;
+end
+
 %% Split up the micro colonies into watershed regions to assemble cells
 % if data exists, reconstruct ws
 if dataFlag
@@ -287,7 +308,12 @@
 %% Remake the data structure
 
 % Determine the "good" and "bad" segments
-data = defineGoodSegs(data,ws,phaseNormUnfilt,C2phaseThresh,mask_bg, A,CONST, ~dataFlag );
+data.mask_bg       = mask_bg;
+data.phaseNorm     = phaseNormUnfilt;
+data.C2phaseThresh = C2phaseThresh;
+
+data = defineGoodSegs(data, ws, CONST, ~dataFlag );
+
 data.mask_colonies = mask_colonies;
 
 % copy the existing score into the data structure
@@ -315,7 +341,7 @@
 
 
 if disp_flag
-    figure(1)
+    figure( 'name', 'SuperSegger frame segmentation' )
     clf;
     showSegDataPhase(data);
     drawnow;