A study on the effects of window size on electrocardiogram signal quality classification
The sliding window-based method is one of the most used method for automatic Electrocardiogram (ECG) signal quality classification. Based on this method, ECG signals are generally divided into small segments depending on a window size and these segments are then used in another classification pro...
Saved in:
| Main Author: | |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | English |
| Published: |
2016
|
| Subjects: | |
| Online Access: | http://repo.uum.edu.my/20118/1/KMICe2016%20333%20338.pdf http://repo.uum.edu.my/20118/ http://www.kmice.cms.net.my/kmice2016/files/KMICe2016_eproceeding.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The sliding window-based method is one of the most
used method for automatic Electrocardiogram (ECG) signal quality classification. Based on this method, ECG signals are generally divided into
small segments depending on a window size and
these segments are then used in another classification process, e.g., feature extraction. The segmentation step is necessary and important for signal classification and signal segments with
different window sizes can directly affect the
performance of classification. However, in signal
quality classification, the window size is often
randomly selected and further analysis on the most
appropriate window sizes is thus required. In this
paper, an extensive investigation of the effects of window size on signal quality classification is
presented.A set of statistical-amplitude-based
features widely used in the literature was extracted based on 10 different window sizes, ranging from 1 to 10 seconds.To construct signal quality classification models, four well-known machine learning techniques, i.e., Decision Tree, Multilayer Perceptron, k-Nearest Neighbor, and Naïve Bayes, were employed.The performance of the quality classification models was validated on an ECG dataset collected using wireless sensors from 20 volunteers while performing routine activities, e.g.,sitting, walking, and jogging.The evaluation results obtained from four machine-learning classifiers demonstrated that the performance of signal quality classification using window sizes of 5 and 7 seconds were good compared with other sizes. |
|---|