An Improved QRS Detection Approach based on Two-Lead Bi-Referring

Heartbeat classification using disease-specific feature selection

“…While many methods can be used to locate QRS waves and detect the fiducial points, such as the techniques used in our previous work [20,21] and wavelet-based methods [22,23], the objective of the present study is to develop a heartbeat classifier by feature selection. For convenience and to operate well with the WaveForm DataBase (WFDB), we employ the QRS annotation included in MIT-BIH-AR and use the tool "ecgpuwave" (a QRS detector and waveform limit locator available from the physionet at http://www.physionet.org/physiotools/ecgpuwave/) to detect the fiducial points, which include P-onset, P-peaks, P-offset, QRS-onset, QRS-offset, T-onset, T-peaks and T-offset.…”