Forum: Software

> Pranav Kachhawa has emailed me with this question.
>
> I have completed the Atrial Fibrillation Detection using LabVIEW and Matlab. I would like to know how to implement it using open source software.
>
> Can somebody give him some clues?
hi everyone
i'm new user here , i'm doin my masters in biomed ! i'm doin a project on heart rate variability . if any body has a source code on matlab which will calculate the rr interval from the text data of ecg pls mail me
my email id is
thanks in advance
prasad

> I have completed the Atrial Fibrillation Detection using LabVIEW and Matlab. I would like to know how to implement it using open source software.
>
> Can somebody give him some clues?

It would be great if you can use some open source language for making your algorithm open source.

It depends on what kind of outlook one want to take while making it open. Firstly if you want to distribute your algorithm as an algorithm (PC side/Mobile using WindowsCE) and want people upgrade them its better to stick to Java or C# for that matter. Octave (as Adoran) suggested, is also a good idea. Try putting a testbed/wrapper to read in some data and show results.

But for hardware implementation I would port everything to C (not even C++) taking a little more pain.

> Pranav Kachhawa has emailed me with this question.
>
> I have completed the Atrial Fibrillation Detection using LabVIEW and Matlab. I would like to know how to implement it using open source software.
>
> Can somebody give him some clues?

Can the matlab source code be made available? I think it's ok to assume java for now. If it has to be tightened up because of invalid minimum hardware assumptions, at least there will be a working prototype.

response moved to hardware:
http://www.open-ecg-project.org/tiki-view_forum_thread.php?comments_parentId=35&forumId=4

matlab to C,C++ code conversion and C, C++ to matlab modules is not good because matlab is running on Java, and all function calls take more than 20-30msec. Is not good for low cost hardware, that have no flow control bufferization. matlab is good for prototyping and education, I do not recommended it for fast and open code applications, even this Java is not developed for life support applications. matlab can be used good for algorithm development, but my opinion is that C,C++ is good for this and you can use all open code FFT and signal analysis libraries that matlab use. Just read in goodle about this.

> Pranav Kachhawa has emailed me with this question.
>
> I have completed the Atrial Fibrillation Detection using LabVIEW and Matlab. I would like to know how to implement it using open source software.
>
> Can somebody give him some clues?

Pranav has some possibilities:
1) Matlab has the ability to export its "m-files" to C/C++, you code and then port it to whatever you wants to make.

2) You can export your Matlab application to a DLL and use it on your software.

3) There are some open source alternatives for Matlab and LabView?. An alternative for Matlab is Sysquake, it runs on Windows, Linux, MacOSX, and even on PalmOS (the port for palmos is called LyME). There is also Octave, Scilab, and Sage. For LabView?, a good alternative is "Processing" (www.processing.org), which runs java in a simplified way.

Hope that helps!

Jeferson

Pranav Kachhawa,

Your alternative is to port your program to an open-source mathematics platform. Might I suggest "Octave". It has a very similar layout and feel to MatLab?.

I have begun learning Octave as a means to wean myself off MatLab? which as Gari points out involves expensive license fees that I avoid simply because I am a student. When I graduate I will lose this entitlement.

It is a GNU/GPL certified platform and it can be found http://www.gnu.org/software/octave/

Emlyn Clay

> Pranav Kachhawa has emailed me with this question.
>
> I have completed the Atrial Fibrillation Detection using LabVIEW and Matlab. I would like to know how to implement it using open source software.
>
> Can somebody give him some clues?

Dear Pranav,
This depends on what you mean by 'implement' your software. If you've already written the software, then you need to evaluate it on a known database, such as the AFDB.
Then, in the context of this project, you need to make it work across platforms, and on a variety of data formats. As yet, no data format libraries have been agreed upon, so this is difficult. I would suggest using George Moody's most excellent WFDB libraries that read more data formats than you could ever imagine (including EDF). The problem with Matlab is that your users will need an expensive Matlab license. Matlab and Labview are wonderful tools, but they are going to be difficult to fuse in this project. It's true that you can compile down Matlab code into license-free libraries, but this has to be repeated for many different architectures and operating systems, and represents a heavy burden on the project.
See here and here for more discussion on this topic.
After you have tested your software and reported the results on a standard database, assuming your code out-performs existing methods, I would suggest you post the code and a link to it, so that others can use it as a basis for an AF detector.
Note that the conventional methods using the statistics of P-P intervals may not work on the short periods of data that would be collected for this project application.
All the best,
Gari