Lu, Nan, Wang, Jihong, Zhang, Chenghui, Vatish, Manu and Randeva, Harpal S. (2007) Uterine electromyography features extraction for classification of term and pre-term signals. In: International Conference on Complex Systems and Applications, Jinan, People's Republic of China, June 08-10, 2007. Published in: Dynamics of Continuous Discrete and Impulsive Systems. Series B. Applications & Algorithims, Vol.14 (No.4). pp. 1675-1679.

Abstract

Surface electromyography (EMG) signal is a complex nonlinear signal which can be used in clinical for early diagnosis of preterm labor. But it is difficult to differentiate uterine contractions which lead to preterm birth by using EMG signals with current tools. In this paper, a new algorithm is proposed to extract the features from surface uterine EMG signal and to distinguish the normal term labor from abnormal preterm labor signals. In this algorithm, firstly, the signal is preprocessed to eliminate the noise and high frequency component using threshold de-noising and wavelet de-noising methods. Secondly, the fractal dimension value along the signal is calculated for the extraction of contraction patterns. Then each contraction pattern of the signal is decomposed and the average wavelet packet energy of the whole signal is calculated using wavelet packet transform. Finally, the signals are classified using artificial neural network method. The experimental results show that the classification accuracy of term labor signal and preterm labor signal can reach 64.1% which is an encouraging result.

Data sourced from Thomson Reuters' Web of Knowledge

Request changes to a record

Actions (login required)

View Item