• home
• current students
• graduate
• events

Presenter: Ping Cheng
Supervisor:

Date: Wed, March 21, 2018
Time: 09:00:00 - 10:00:00
Place: EOW 230

ABSTRACT

ABSTRACT: Abstract— Long-term ECG monitoring systems were previously proposed to solve problems such as the portable problem and the difficulty of capturing intermittent arrhythmias in traditional snapshot ECG devices. However, long-term ECG systems are subject to several practical limitations: battery power restriction, network congestion and heavily-redundant ECG data. To overcome these problems, a novel simple normal and abnormal heartbeat classification algorithm is proposed for long-term ECG monitoring system, to decrease the redundant normal heartbeats in data transmission. The proposed algorithm explores two types of variations, intra-beat variation and inter-beat variation, namely, the waveform change indicator (WCI) and the modified RR interval ratio (modRRIR), to predict a heartbeat change. WCI indicates a waveform change in P/QRS/T segments, and is calculated by applying one-class support vector machines on tens of personal normal heartbeats. modRRIR characterizes the successive heartbeat interval variation, and is obtained from the ratio of consecutive three heartbeats, other than the absolute value of RR intervals. WCI and modRRIR are firstly tuned separately, and then combined to separate normal heartbeats from abnormal ones. The proposed method is evaluated using the publicly available MITDB database and achieves an overall ACC of 78.4%, SE of 76.5%, SP of 93.2%, and PP of 98.9%, which outperforms the results in the literature. Furthermore, the strategy is also validated using the data collected from the ECG platform HeartCarer built in our research group.