function STATS = calcShapeFeatures2D(bwImage)
%% compute shape features
% given an image region and returns a structure with the results.
STATS = regionprops (bwImage,'Area', 'Extent', 'Perimeter', ...
'Centroid', 'ConvexArea', 'ConvexImage','ConvexHull',...
'Solidity', 'Eccentricity', 'MajorAxisLength', ...
'EquivDiameter','MinorAxisLength');
%Call custom perimeter and convex perimeter method, because it is more precise
STATS.Perimeter = FacetMidpointPerim(bwImage);
STATS.ConvexPerimeter = ConvexPerim(bwImage);
%Circularity
STATS.Circularity = (4*pi*STATS.Area)/(STATS.ConvexPerimeter^2);
%Roughness
STATS.Roughness = STATS.ConvexPerimeter/STATS.Perimeter;
%Elongation
STATS.Elongation = STATS.MajorAxisLength/STATS.MinorAxisLength;
%Compactness
STATS.Compactness = STATS.Perimeter^2/(4*pi*STATS.Area);
%% Compute Maximum Chord Length
convexHullVertices = STATS.ConvexHull;
maxChordLength = 0;
for j = 1:size(convexHullVertices,1)-1
for k = j+1: size(convexHullVertices,1)
chordLength = sqrt( ( convexHullVertices(j,1) - convexHullVertices(k,1) )^2 + ( convexHullVertices(j,2)- convexHullVertices(k,2) )^2 ) ;
if chordLength > maxChordLength
maxChordLength = chordLength;
end
end
end
STATS.MaxChordLength = maxChordLength;
%% Compute Radial distance Standard Deviation
I3=bwmorph(bwImage,'remove'); % Nodule Outline
DR = NaN; % Vector of Outline Distance for RadialDistanceSD
j = 0;
for r = 1:size(I3,1)
for c = 1:size(I3,2)
if (I3(r,c) == 1)
j = j + 1;
% Distance
DR(j) = sqrt((c-STATS.Centroid(1))^2+(r-STATS.Centroid(2))^2);
end
end
end
% RadialDistanceSD
STATS.RadialDistanceSD = nanstd(DR(:));
%% Compute 2nd Moment
% Vector of Distance to every pixel in the region for 2nd Moment
DSM = NaN;
k = 0;
for r1 = 1:size(bwImage,1)
for c1 = 1:size(bwImage,2)
if (bwImage(r1,c1) == 1)
k = k + 1;
% Distance Square
DSM(k) = (c1-STATS.Centroid(1))^2+(r1-STATS.Centroid(2))^2;
end
end
end
% SecondMoment (Variance)/Area
STATS.SecondMoment = sum(DSM)/(STATS.Area)^2;
end
