Hi guys, I’m currently working with intracranial EEG data with one ECG channel. I need to precisely identify all rpeak in my ECG signal.
By doing the automatic detection Event > Detect Heartbeat I can mark some of them, like 10% of them but not all. Usually I can but for this particular patient the data for ECG is really noisy. I can see some Rpeak but the signal is noisy. PSD shows a big contamination by 50hz, 244hz & harmonics. I try several method to clean ECG, including notch filter, bandpass, DC offset. I can obtain a way more clear ECG signal with good QRS pattern. But despite that visual clear amelioration on data, the automatic detection give me the same number of rpeak, nothing change. Sometimes it can even give me less rpeak detected.
Is there a way to enhance this detection or am I doing something wrong here ?
I did the preprocessing on both review file and imported in database files giving me same result.
Since your data is noisy, you may want to reduce the amplitude threshold from 4*std
In the shared plots, the signal is called ECG+, does it imply there is ECG-, if so, have you run the R-peak detector on their difference?
Another alternative is to process the ECG signal in a different software to detect the R peaks (and export as simple text file), then import the times in Brainstorm.
The file can be one row per R-peak time (in seconds). Import in Brainstorm as Array of times, Example of file:
