JIAN-JIUN DINGHuang C.-W.YI-LWUN HOCHI-SHENG HUNGYEN-HUNG LINYING-HSIEN CHEN2020-12-282020-12-2820141205-6626https://www.scopus.com/inward/record.uri?eid=2-s2.0-84906962405&partnerID=40&md5=d5303cb83d28ad2e084e717d68c44036https://scholars.lib.ntu.edu.tw/handle/123456789/534416A new R-wave peak detection algorithm for electrocardiogram (ECG) signal with very high accuracy and efficiency is developed in this paper. With the proposed techniques of R-wave peak candidate sifting and a Haar-like matched filter, we can accurately determine which point is likely to be an R-wave peak. The point that is impossible to be an R-wave peak will not be further processed. After this process, only about 3-5 points per second are needed to be further checked, which is very helpful for reducing the computation time. Moreover, instead of a conventional filter, the proposed technique of the variation ratio test is used to exclude the peak caused by noise. Furthermore, several posterior processing techniques, such as adaptive thresholds and a regularity test, are proposed to further improve the accuracy. To make the proposed algorithm even more efficient, any transform, including the Fourier and the wavelet transforms, is avoided in our algorithm. The simulations for the MIT/BIH arrhythmia database show that our proposed algorithm achieves an error rate of 0.189%, which is much lower than that of existing methods.Electrocardiogram; Fast algorithm; R-wave peak candidate selection; R-wave peak detection; Variation ratio testjournal article2-s2.0-84906962405