Medical Image Processing - Digital Signal Processing for Medical Imaging Using Matlab - page 83

4.3.2.1 kltformedimagecomp.m
%Medical image compression using KL-Transformation
%Each column of the image is treated as the vector.
load mri
D=double(D);
%collecting the vectors
vect=[];
for i=1:1:27
vect=[vect D(:,:,1,i)];
end
vect=double(vect);
%Computing the inner-product matrix
vect=double(vect);
C=cov(double(vect'));
[E,V]=eigs(C,50);
%Compression starts here
%The co-efficients to be stored (to store all the images ) is as follows.
%Compression for the first image
a=[25 50];
p=1;
for i=1:5:27
I=double(D(:,:,i));
subplot(6,3,p)
p=p+1
colormap(gray)
imagesc(I)
for cr=1:1:2
COEF=[ ];
E1=E(:,1:1:a(cr));
for j=1:1:128
COEF=[COEF E1'*I(:,j)];
end
%Reconstruction
for k=1:1:128
R(:,k)=E1*COEF(:,k);
end
subplot(6,3 ,p)
colormap(gray)
imagesc(R)
p=p+1
end
end
4.4 Feature Extraction and Classification
To help the automated system to classify the medical images into normal and ab-
normal categories, the features are extracted from the medical image to train the
classifier. The feature vector extracted using the wavelet transformation is described
below.