Automatic detection of onset and offset of QRS complexes independent of isoelectric segments

作者：Hu, Xiao^*; Liu, Jingjing; Wang, Jiaqing; Xiao, Zhong; Yao, Jing

来源：Measurement, 2014, 51: 53-62.

DOI：10.1016/j.measurement.2014.01.011

• Electrocardiogram
• QRS complexes
• Onset and offset
• Fitting line
• Relative distribution

Accurate location to onset and offset of QRS complex from ECG signal is help measuring other features and it is a important and challenge task to recognize these onsets and offsets from ECG signal without isoelectric segment. The algorithm, which is introduced in this paper, does not depend on isoelectric segment and can accurately detect these onsets and offsets. In the algorithm, for every sample within about a cardio cycle of locally normalized ECG signal, two fitting straight lines are obtained respectively, from two sets of samples before and after the sample. And then two gradients and a smaller included angle corresponding to the two fitted lines are calculated. Third, those samples whose two gradients satisfied a pre-set value are taken as candidates for the R peak, and the sample with the minimal included angle is regarded as the R peak in this cardiac cycle. Finally, respectively before and after the detected R peak, the algorithm researches the onset and the offset of the US complex according to the same decision strategy. All 32 sets of 12-lead ECG records from http://physionet.org/pn3/twadb/ are employed to evaluate the method. The experimental results show that compared with three traditional methods, the method can more accurately detect all onsets and offsets and has better robust on interference.

• 出版日期2014-5
• 单位广州大学