True Negative can not detect when do clasiffication with libsvm

Hello , I had the probelm with libsvm. I've tried to solve this problem more than 2 weeks but I can not solve it . I hope someone can help my problem.

My problem is.

I run libsvm in matlab with 40 FP and 13 TN. I want to classify become cancer and no cancer. When I try with this data using svm classification, I got 90.57% . with the confusion matrix: 13 5 0 35 and when I tried with libsvm, it gives an strange output. it gives 79.49% but it can not detect True negative. I used 5-cross validation . this is the confusion matrix that i got.

 cmTest 1  ==== cmTest2 ====cmTest 3====cmTest4 ====cmTest5 =

0 3            0 3          0 3         0 2           0 2
0 10           0 10         0 10        0 10          0 10

I dont know why the true negatif is 0. when I tried with data testing=data training, it gives 100%. and also when I tried with dataset from libsvm, it's working well. I can not find the problem until now. this is I put my main data train function.

      [labels,data]=training('D:\classification\fp\',...
         'D:\th\tp\');

opts = '-s 0 -t 2 -c 0.0313 -g  3.0518e-05';    %# libsvm training options
nfold=5;
indices = crossvalind('Kfold', labels, nfold);
acc = zeros(nfold,1);
    for i=1:nfold
        testIdx = (indices == i); trainIdx = ~testIdx;
          trainLabel = labels(trainIdx);
          trainData = data(trainIdx,:);
          testData=data(testIdx,:);
          testLabel=labels(testIdx);  

            model = svmtrain(trainLabel, trainData, opts);
  [predTestLabel, acc, decVals] = svmpredict(testLabel, testData, model);
  acc = mean(predTestLabel == testLabel)
  cmTest = confusionmat(testLabel,predTestLabel)
      akurasi(i) = sum(predTestLabel == testLabel) ./ numel(testLabel)
      end
  result_accuracy=mean(akurasi)

and this is my train function:

function [C,F ] = training(fpDir,tpDir)
db=dir(fullfile(fpDir,'*.dcm')); 
F=[];
%C=zeros(numel(db),1);
C=[]
for ii= 1:numel(db)
    name=fullfile(fpDir, db(ii).name); 
    im1=dicomread(name);
    F = [F; func_feature1(im1)];  
    C = [C;1];

db=dir(fullfile(tpDir,'*.dcm'));
for ii = 1:numel(db)
    im1=dicomread(fullfile(tpDir, db(ii).name));
    F = [F; func_feature1(im1)];
    C = [C;0];
end
end

is there something wrong with my code ? can someone help me to correct this code. Thank you ..