Can you provide an example of how your QRS peaks are written in your file?

I don't know if you can say QRS peak because it is then a complex of tree waves Q, R and S which represent tree deferent steps in heart rythme. I think you just detect the R peaks unless you detect also the onset and offset of Q and S waves.

Dear Maxime,

Thank you for your response.

I mean the detection of R-peak only. 

The file contains two columns of: time and R-peaks.

0.191667    0.214162

1.008333    0.301969

1.819444    0.266861

2.625000    0.401221

3.397222    0.211462

4.205556    0.424435

    .                    .

    .                    .

    .                     .

 Dear Ms. Al-Busaidi,

If you happen to use MATLAB to check the  feasibility of your approach, I may be able to help you in comparing your results with that of the  annotation file from MIT- BIH arrhythmia database. I have already done that once, with the same database . 

Dear Aouache,

I did not work on ROC and AUC methods a lot but from my search I understood that ROC is mainly used to evaluate the method after performing some changes to get the optimal values. This is valid if I already have FP and FN values, which I am trying to calculate.

So, please if you have a clear example or reference on what you suggested, I’ll be grateful if you share it with us.

Dear Jithesh,

Can you share your MATLAB experience, please?

Because I am mainly using MATLAB for evaluating my results and I tried to use bxb.m function provided by the physionet website but the function gets halted every time I run it.

That's why I am trying to use the WFDB C-code library. 

Dear Ms. Asiya, 

First of all let me make sure that what I am thinking is true.

You have developed a method to find all the peaks in a given ECG sample and you need to verify whether the peaks detected by your algorithm is same as that shown in the MIT-BIH database, so that you can evaluate accuracy right?

Jithesh,

Yes, that is exactly what I want to do. I am targeting the R-peaks only. 

Dear Ms. Asiya.  If you have WFDB tools installed you can use the rr2ann command.

I've copied the data you have showed in a previous message in a file called data. After that I write the following command:

rr2ann -r 100 -a new -T -x 360 < data

I have the first record (100) of MIT-BIH database in the same folder. "new" will be the name  of the annotator file I'm going to create. "-T" option indicates that the data corresponds to beat positions, nor to RR distances. "-x360" is necessary because your data are in seconds and  this factor is needed to transform them in samples (360 is the sampling frequency of 100 record).

As a results the file 100.new is created and I could use the bxb command:

bxb -r 100 -a atr new -f 0

The "-f 0" is needed because by default the bxb command begin the comparison 5 minutes after the begin of the record.

Dear Dr. Xose,

Thank you for the detailed answer. It is exactly what I was trying to do.

Currently, I am trying to understand the results and comparing them to the result I did conducted them manually.

Regards

Hi All,

I have the same problem as Asiya. I am using Matlab 2018. I want to compare reference R peak annotation with the R peak detected by my algorithm. As of this discussion, I was trying to add WFDB toolbox to Matlab. As per the instruction, first, I gave directory to install then I pasted the code, but I was getting lots of error. Please solve my problem or give any idea to fulfill my purpose.

Regards

Chhavi

Hello,

Dear Chhavi Kb what are the errors you are getting?

Check your database/mitdb folder.

If the ECG files 100.atr and 100.dat are not available, download them from here:

https://physionet.org/physiobank/database/mitdb/

Thanks for help. Could you tell what is in your data.txt ? In my matlab code, R_location variable contains 1x2272 (double) R peaks location. How can i use rr2ann , in my case?

Welcome Chhavi Kb .

The file contains two columns of: time locations in seconds and R-peak values.

0.191667    0.214162

1.008333    0.301969

1.819444    0.266861

2.625000    0.401221

3.397222    0.211462

4.205556    0.424435

    .                    .

    .                    .

    .                     .

Hi Asiya Al-Busaidi and Xose A. Vila

I am using Matlab 2018b for my algorithm. After running my code, I am getting a 1D array of 1x2272 containing R peak location in samples. Now i want to compare this by reference annotation provided by physionet. Please suggests me how i can do this by the wfdb toolbox?