3 ms·
I've worked a bit on ECG signals (Physionet, too). You may want to take a look at: - "Baseline Correction with Asymmetric Least Squares Smoothing" by Paul H. C
by Jugurtha 9y ago
I've worked a bit on ECG signals (Physionet, too). You may want to take a look at:
- "Baseline Correction with Asymmetric Least Squares Smoothing" by Paul H. C. Eilers and Hans F.M. Boelens[0]. They used it for chromatography and spectroscopy but I've had great results applying that to detect and remove the baseline from ECG. The article provides 11 lines of MATLAB code to illustrate.
- Savitzky-Golay filter: For smoothing. I liked it for some of the noisy signals.
- Poincaré plot for RR intervals.
[0]: https://zanran_storage.s3.amazonaws.com/www.science.uva.nl/ContentPages/443199618.pdf https://zanran_storage.s3.amazonaws.com/www.science.uva.nl/C...
- JPLeRouzic 9y agoThanks for your detailled answer and those pointers, I appreciate this. I am using phonocardiograms (PCG) not ECG but the processing is quite similar. In my case I record PCG with ultrasound Doppler but a mere microphone could do the job as well. I think I went further than detecting the heart's rate and variability, as I try to recognise each heart sound (S1 ... S4) inside a beat and I developped a technique for having a beat signature in order to apply ML on the sequence of beats with HMM. This beat signature technique could certainly be applied to ECG. It could probably even be used for any series of events. However my problem is not technical, it is about raising awareness, getting help (including resources) and get strategic advising.