I have attached my code , in the classification section , I am only getting epidural image and other type of haemorhage is not getting classified. Please can you help

Question

Iqra Rizvi 2018-8-10

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/414328-i-have-attached-my-code-in-the-classification-section-i-am-only-getting-epidural-image-and-other

imgextract.m

if true
  % code
end
%%Image Acquisition %%

[Filename ,Pathname] = uigetfile('Extra Axial Haemorrhage\*.*','Select One Input Image'); I = imread([Pathname Filename]); save Filename Filename

figure('Name','Input Image','NumberTitle','Off'); imshow(I); axis off; title('Input Image','fontsize',12,'fontname','Times New Roman','color','Black');

I = imresize(I,[256 256]); [m,n,ch] = size(I); if ch==3 I = rgb2gray(I); end figure('Name','Resized Image','NumberTitle','Off'); imshow(I); axis off; title('Resized Image','fontsize',12,'fontname','Times New Roman','color','Black');

%% Preprocessing %%

% Power law transformation %

I_temp = double(I); c = 1; %constant gamma = 0.8; I_plt = c*I_temp.^gamma ; figure,imshow(I_plt, []);title('Power Law transformation');

%% Segmentation %%

% Sobel %

% img=imread('a.jpg'); % B=rgb2gray(img); % subplot(2,2,1) % imshow(B) % pause(2)

% Sobel Power Law Transformation % I_sobelplt = double(I_plt);

for i=1:size(I_sobelplt,1)-2 for j=1:size(I_sobelplt,2)-2 %Sobel mask for x-direction: mx=((2*I_sobelplt(i+2,j+1)+I_sobelplt(i+2,j)+I_sobelplt(i+2,j+2))-(2*I_sobelplt(i,j+1)+I_sobelplt(i,j)+I_sobelplt(i,j+2))); %Sobel mask for y-direction:m my=((2*I_sobelplt(i+1,j+2)+I_sobelplt(i,j+2)+I_sobelplt(i+2,j+2))-(2*I_sobelplt(i+1,j)+I_sobelplt(i,j)+I_sobelplt(i+2,j)));

I_plt(i,j)=sqrt(mx.^2+my.^2); end end % subplot(2,2,2) %figure,imshow(I_plt); title('Sobel gradient'); %pause(2) %Define a threshold value %Entropy_Iplt = entropy(I_plt);

Thresh=100; I_plt=max(I_plt,Thresh); I_plt(I_plt==round(Thresh))=0; I_plt=uint8(I_plt);

%subplot(2,2,3) figure,imshow(~I_plt);title('PLT Sobel edge detection');

% Prewitt %

% Prewitt PLT %

I_prplt = edge(I_plt,'prewitt'); figure,imshow(I_prplt);title('PLT Prewitt Edge detected Image');

% Canny %

% Canny PLT %

I_cplt = edge(I_plt,'Canny');

figure('Name','Segmentation Process','NumberTitle','Off'); imshow(I_cplt); axis off; title('PLT Canny Edge Detector','fontsize',12,'fontname','Times New Roman','color','Black');

% Gaussian Filter %

% Gaussian Filter PLT %

H = fspecial('gaussian',[3 3],0.5); Gaussian_Filt = imfilter(I_plt,H,'replicate');

figure('Name','Preprocessing Image','NumberTitle','Off'); imshow(Gaussian_Filt); axis off; title('Gaussian Filtering ','fontsize',12,'fontname','Times New Roman','color','Black');

Igray = Gaussian_Filt;

I_gplt = double(Igray); classType = class(Igray); scalingFactor = double(intmax(classType)); I_gplt = I_gplt/scalingFactor; Gx = [-1 1]; Gy = Gx'; Ix = conv2(I_gplt,Gx,'same'); Iy = conv2(I_gplt,Gy,'same');

figure('Name','Ix','NumberTitle','Off'); imshow(Ix ,[]); axis off; title('Ix_plt','fontsize',12,'fontname','Times New Roman','color','Black');

figure('Name','Iy','NumberTitle','Off'); imshow(Iy ,[]); axis off; title('Iy_plt','fontsize',12,'fontname','Times New Roman','color','Black');

% Fuzzy %

% Fuzzy PLT %

edgeFIS = newfis('edgeDetection'); edgeFIS = addvar(edgeFIS,'input','Ix ',[-1 1]); edgeFIS = addvar(edgeFIS,'input','Iy ',[-1 1]); sx = 0.1; sy = 0.1; edgeFIS = addmf(edgeFIS,'input',1,'zero','gaussmf',[sx 0]); edgeFIS = addmf(edgeFIS,'input',2,'zero','gaussmf',[sy 0]); edgeFIS = addvar(edgeFIS,'output','Iout',[0 1]);

wa = 0.1; wb = 1; wc = 1; ba = 0; bb = 0; bc = 0.7; edgeFIS = addmf(edgeFIS,'output',1,'white','trimf',[wa wb wc]); edgeFIS = addmf(edgeFIS,'output',1,'black','trimf',[ba bb bc]);

r1 = 'If Ix is zero and Iy is zero then Iout is white'; r2 = 'If Ix is not zero or Iy is not zero then Iout is black'; r = char(r1,r2); edgeFIS = parsrule(edgeFIS,r); showrule(edgeFIS)

I_fplt = zeros(size(I_gplt)); for ii = 1:size(I_gplt,1) I_fplt(ii,:) = evalfis([(Ix(ii,:));(Iy(ii,:));]',edgeFIS); end

figure('Name','Segmentation Using Fuzzy Logic','NumberTitle','Off'); imshow(I_fplt,[]); axis off; title('PLT Segmentation Using Fuzzy Logic','fontsize',12,'fontname','Times New Roman','color','Black'); %figure,imshow(~Seg_Im);title('Fuzzy');

I_adf=adaptivethreshold(I_fplt,11,0.1,0);

figure('Name','Segmentation Process','NumberTitle','Off'); imshow(I_adf); axis off; title('Adaptivethreshold','fontsize',12,'fontname','Times New Roman','color','Black');

%% Proposed Segmentation Method %%

% Fuzzy + Sobel % I_sobelf = double(I_adf);

for i=1:size(I_sobelf,1)-2 for j=1:size(I_sobelf,2)-2 %Sobel mask for x-direction: mx=((2*I_sobelf(i+2,j+1)+I_sobelf(i+2,j)+I_sobelf(i+2,j+2))-(2*I_sobelf(i,j+1)+I_sobelf(i,j)+I_sobelf(i,j+2))); %Sobel mask for y-direction:m my=((2*I_sobelf(i+1,j+2)+I_sobelf(i,j+2)+I_sobelf(i+2,j+2))-(2*I_sobelf(i+1,j)+I_sobelf(i,j)+I_sobelf(i+2,j)));

I_adf(i,j)=sqrt(mx.^2+my.^2); end end % subplot(2,2,2) figure,imshow(I_adf); title('fuzzy Sobel gradient'); %pause(2) %Define a threshold value

%Thresh=50; %I_fplt=max(I_fplt,Thresh); %I_fplt(I_fplt==round(Thresh))=0; %I_fplt=uint8(I_fplt);

%subplot(2,2,3) %figure,imshow(~I_fplt);title('fuzzy Sobel edge detection');

% Fuzzy + Prewitt %

I_prf = edge(I_adf,'prewitt'); figure,imshow(I_prf);title('Fuzzy Prewitt Edge detected Image');

% Fuzzy + Canny %

I_cf = edge(I_adf,'Canny');

figure('Name','Segmentation Process','NumberTitle','Off'); imshow(I_cf); axis off; title('Fuzzy Canny Edge Detector','fontsize',12,'fontname','Times New Roman','color','Black');

%% Feature Extraction %

% LoG Lindeberg Algorithm %

LoG_Pt = LoG_Lindeberg(I_adf); draw(~I_adf,LoG_Pt,'LoG Lindeberg');

% Harris Laplace Algorithm % HarrisLaplace_Pt = Harris_Laplace(I_adf); draw(~I_adf,HarrisLaplace_Pt,'Harris Laplace');

% Harris Algorithm % Harris_Pt = Harris(I_adf); draw(~I_adf,Harris_Pt,'Harris');

%% % Features Value %%%

Feat_LoG = [mean(LoG_Pt(1)) mean(LoG_Pt(2)) mean(LoG_Pt(3))]; Feat_HarrisLaplace = [mean(HarrisLaplace_Pt(1)) mean(HarrisLaplace_Pt(2)) mean(HarrisLaplace_Pt(3))]; Feat_Harris = [mean(Harris_Pt(1)) mean(Harris_Pt(2)) mean(Harris_Pt(3))]; Test_Features = [Feat_LoG Feat_HarrisLaplace Feat_Harris]; save Test_Features Test_Features

f=figure('Name','Test Features Values','NumberTitle','off'); t = uitable('Parent',f,'Data',Test_Features');

%% Classification %%

load Train_Features load Test_Features load Label load Filename

f=figure('Name','Train Features Values','NumberTitle','off'); t = uitable('Parent',f,'Data',Train_Features);

% PNN Classifier %

load Label T = ind2vec(Label); net = newpnn(Train_Features',T); Y = net(Test_Features'); Class = vec2ind(Y);

if Class==1 msgbox('Epidural Hematoma'); elseif Class==2 msgbox('Subarachnoid Hematoma'); else msgbox('Subdural Hematoma'); end %% Performance Measure %%

Actual = Label; POS = 3; Predicted = Label; Predicted (POS) =3;

Result = Evaluate(Actual,Predicted);

figure('Name','Performance Measures','NumberTitle','Off','Color','White'); bar(1,Result(1),0.5,'m') ; hold on ; bar(2,Result(2),0.5,'c') ; hold on ; bar(3,Result(3),0.5,'g') ; set(gca, 'XTick',1:7, 'XTickLabel',{'Accuracy' 'Sensitivity' 'Specificity' },'fontsize',12,'fontname','Times New Roman','fontweight','bold');