Cell array, getting the columns that have non empty cells in the middle of the column

Hi every body. I have a cell array m|x|n , this array is created using a buffer function with an overlap. So i have a 1 dimensional array with some labelings spread in segments and the rest is empty cells. This array i buffer with overlap and an MxN matrix is created. the labelling segments based on the segment length is spread into 2 or three columns of the MxN matrix. But i want to get only the columns where the segment is in the center of the column. Is there any way of achieving it simply? My code is like this:

      %creating matrix cell first and second rows are start and stop of anomalie and the third is the name of anomlie.
      %
      lab={};
      lab(:,1)=x1q(1:4:end);
      lab(:,2)=x1q(3:4:end);
      lab(:,3)=x1q(2:4:end);
      %converting the time into seconds and creating an array of start and stop
      %time
      [~, ~, ~, H, MN, S] = datevec(lab(:,1));
      tanom(:,1)=H*3600+MN*60+S;
      [~, ~, ~, H, MN, S] = datevec(lab(:,2));
      tanom(:,2)=H*3600+MN*60+S+0.5;

      % get the middle time of the anomalie window
      tanom(:,3)=tanom(1:end,1)+(tanom(1:end,2)-tanom(1:end,1))/2;
      %getting the indices of the time nearest to that of the start and stop of
      %the anomalie
      indt(:,1)=nearestpoint(tanom(:,1),t);%start of anomaly
      indt(:,2)=nearestpoint(tanom(:,2),t);%end of anomaly
      indt(:,3)=nearestpoint(tanom(:,3),t);%midle of anomaly
      %create a cell array with the name of anomalie for the length of the event
      %create an empty cell array to hold the anomalies for each row of
      %accelereometer data
      labelat=cell(length(t),1);
      anomalie=zeros(length(t),1);
      for i=1:size(tanom,1);
          for e=indt(i,1):indt(i,2);
              labelat(e)=lab(i,3);
              anomalie(e)=1;
          end;
      end;
      labelat(numel(labelat)+1:length(t))=[];%add the rest of empty cells
      clear('x1q','lab','Y', 'M', 'D', 'H', 'MN', 'S','e','i');
      %Windowing data
      labelw = buffer(labelat, wsize, overlap);%window labels
      anomw = buffer(anomalie, wsize, overlap);%window labels anomalie