Feature extraction for pap-smear images

I am doing my project on image classification on pap-smear images. for segmention i got some codes that can segment nucleus and cytoplasm. but for the image like i have attached here, i cant segment the nucleus and cytoplasm separately. pls help me with the code that can segment the nucleus from the cytoplasm. Also i would like to know how can i extract the features for these pap-smear images. i require code for feature extraction of the images? please help me with it

thank you in advance.

[Rithika]

hey! Im doing a similar project. May I know how you segmented the nucleus?

Answer 1

[Raunak Gupta]

Hi,

From the similar question here I can see you are using Image Segmenter App right now for creating ROI labels. This is a good tool for creating a mask or filter for similar images. If you want to segment different images and not all images are correctly segmented by single filter in Image Segmenter App you may try using Deep Learning for Image Segmentation, here lot of examples are given to get started.

Since for using deep learning you required to have ground truths for training the model network for which some manual labelling is required. You may use Image Labeler App for the same. Since it’s a medical image segmentation task, segnetLayers and unetLayers would be best suited.

[vidhya v]

thank you sir for your response.

[Raunak Gupta]

Always happy to help!

[vidhya v]

sir,

I have to classify the input image of my dataset. Based on the below example code (Brain MRI detection), I am doing my project. for classification i have to use fitcsvm(). As i am new to matlab, I dont know how to implement it, because i have to pass features into ClassificationSVM. svmtrain() and svmclassify() are not supporting. please suggest on how can i replace the functions to get my result. please help me!!!

example code:

g = graycomatrix(G);

stats = graycoprops(g,'Contrast Correlation Energy Homogeneity');

Contrast = stats.Contrast;

Correlation = stats.Correlation;

Energy = stats.Energy;

Homogeneity = stats.Homogeneity;

Mean = mean2(G);

Standard_Deviation = std2(G);

Entropy = entropy(G);

RMS = mean2(rms(G));

%Skewness = skewness(img)

Variance = mean2(var(double(G)));

a = sum(double(G(:)));

Smoothness = 1-(1/(1+a));

Kurtosis = kurtosis(double(G(:)));

Skewness = skewness(double(G(:)));

% Inverse Difference Movement

m = size(G,1);

n = size(G,2);

in_diff = 0;

for i = 1:m

for j = 1:n

temp = G(i,j)./(1+(i-j).^2);

in_diff = in_diff+temp;

end

end

IDM = double(in_diff);

feat = [Contrast,Correlation,Energy,Homogeneity, Mean, Standard_Deviation, Entropy, RMS, Variance, Smoothness, Kurtosis, Skewness, IDM];

load Trainset.mat

xdata = meas;

group = label;

svmStruct1 = svmtrain(xdata,group,'KernelFunction', 'linear');

species = svmclassify(svmStruct1,feat,'showplot',false);

if strcmpi(species,'MALIGNANT')

helpdlg(' Malignant Tumor ');

disp(' Malignant Tumor ');

else

helpdlg(' Benign Tumor ');

disp(' Benign Tumor ');

end

[Raunak Gupta]

Hi,

There are few changes from svmtrain to fitcsvm and svmclassify to predict. Although the syntax remains the same you may change the two lines in the code using svmtrain and svmclassify as follows

% Old Code
svmStruct1 = svmtrain(xdata,group,'KernelFunction', 'linear');
species = svmclassify(svmStruct1,feat,'showplot',false);
% New Code
svmStruct1 = fitcsvm(xdata,group,'KernelFunction', 'linear');
species = predict(svmStruct1,feat);

This should work fine.

[vidhya v]

thank you sir, but its showing error:

Error in classreg.learning.classif.CompactClassificationSVM/score (line 576)

f = score(this.Impl,X,true,varargin{:});

Error in classreg.learning.classif.ClassificationModel/predict (line 411)

scores = score(this,X,varargin{:});

Error in classreg.learning.classif.CompactClassificationSVM/predict (line 417)

[e-mail](this,X,varargin{:});

Error in img_process>classify_Callback (line 280)

species = predict(svmStruct1,feat);

Error in gui_mainfcn (line 95)

feval(varargin{:});

Error in img_process (line 42)

gui_mainfcn(gui_State, varargin{:});

Error in

matlab.graphics.internal.figfile.FigFile/read>@(hObject,eventdata)img_process('classify_Callback',hObject,eventdata,guidata(hObject))

Error while evaluating UIControl Callback.

[Raunak Gupta]

Hi,

Please check feat matrix in the workspace so that the type of data in each row is similar to that in xdata which is used in training. From the error above I think the feat matrix is not able to form or data in it is missing. You can try with passing only one row of feat matrix if it is formed so as to check if any missing data is causing the issue.

[vidhya v]

ok sir. I will see to it. Thank you so much sir.