Electroencephalogram analysis using fast wavelet transform

• Published in: Computers in biology and medicine Vol. 31; no. 6; pp. 429 - 440
• Main Authors: Zhang, Zhong, Kawabata, Hiroaki, Liu, Zhi-Qiang
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.11.2001; New York, NY Elsevier Science
• Subjects: Algorithms; Biological and medical sciences; Computation accuracy; Computation speed; Computerized, statistical medical data processing and models in biomedicine; Computers; Data Interpretation, Statistical; Diagnosis, Computer-Assisted; EEG analysis; Electrodiagnosis. Electric activity recording; Electroencephalography - statistics & numerical data; Fast algorithm; Frequency domain analysis; General aspects. Methods; Humans; Integral equations; Interpolation; Investigative techniques, diagnostic techniques (general aspects); Medical sciences; Nervous system; Time domain analysis; Wavelet transform; Wavelet transforms; Computation speed; Wavelet transform; Fast algorithm; Computation accuracy; EEG analysis; Human; Evaluation; Time frequency domain method; Electroencephalography; Algorithm; Electrodiagnosis; Biomedical data processing; Accuracy; Sensitivity; Rapid technique; Wavelet transformation; Signal analysis; Biomedical engineering
• ISSN: 0010-4825; 1879-0534
• DOI: 10.1016/S0010-4825(01)00019-1

The continuous wavelet transform is a new approach to the problem of time–frequency analysis of signals such as electroencephalogram (EEG) and is a promising method for EEG analysis. However, it requires a convolution integral in the time domain, so the amount of computation is enormous. In this paper, we propose a fast wavelet transform (FWT) that the corrected basic fast algorithm (CBFA) and the fast wavelet transform for high accuracy (FWTH). As a result, our fast wavelet transform can achieve high computation speed and at the same time to improve the computational accuracy. The CBFA uses the mother wavelets whose frequencies are 2 octaves lower than the Nyquist frequency in the basic fast algorithm. The FWT for high accuracy is realized by using upsampling based on a L-Spline interpolation. The experimental results demonstrate advantages of our approach and show its effectiveness for EEG analysis.