Read in an analysis from ImageJ's Analyze Particles and group into different groups using k-means

imagej_kmeans.m

imagejdata=importdata('analyzeParticlesOutput.txt','\t',1)
disp(imagejdata.colheaders)
%% parse the data in matlab
% A tiny piece of code to find which column the given header is in
% note imagej exports do not need the extra quotation marks like R does
findColumn=@(idstrut,colName) find(not(cellfun('isempty',strfind(idstrut.colheaders,[ colName ]))));
% a function to return all of the data in that column
getColumn=@(idstrut,colName) idstrut.data(:,findColumn(idstrut,colName));

%% calculate the nearest neighbor distance and angle
[nndist,nnangle]=findnn(getColumn(imagejdata,'X'),getColumn(imagejdata,'Y'));


%% show the data
figure(1)
subplot(1,2,1);
plot(getColumn(imagejdata,'X'),getColumn(imagejdata,'Y'),'r.')
title('Point Location')
subplot(1,2,2);
hist(nndist)
pause(0.5)

%% make two groups based on area and nearest neighbor distance
groupId=kmeans([nndist' getColumn(imagejdata,'Area')],2,'EmptyAction','singleton');
% show the groups as different colors
xdata=getColumn(imagejdata,'X');
ydata=getColumn(imagejdata,'Y');
figure(2)
plot(xdata(groupId==1),ydata(groupId==1),'r.')
hold on;
plot(xdata(groupId==2),ydata(groupId==2),'bo')
legend({'Group 1','Group 2'});
hold off;