摘要
Nowadays, brain function evaluation using Functional Near Infrared Spectroscopy (fNIRS) is one of the most potential non-invasive monitoring techniques. This paper concerns usefulness of the NIRS signals denoising using the Hemodynamic Evoked Response (HomER) as graphical user interface displays the NIRS data, fast independent component analysis (FASTICA) method to reduce data dimension and the combined Wavelet & PCA method for enhancing NIRS signals. NIRS signals include many types of noise, spread across a broad spectrum of frequencies, such as: low frequency noise from respiratory interference, 0.1 - 0.3 Hz, Mayer wave, about 0.1 Hz, cardiac interference, 0.8 - 2.0 Hz, and other artifacts from head and facial motions. Meanwhile, electronic components generate high frequency noise. Multi-resolution wavelet and PCA was applied successfully to enhance the NIRS signals. It consists of adaptively modifying the wavelet coefficients based on the degree of noise contamination of the processed NIRS signal. This is done subsequently to the signal pre-processing by reducing data dimension using the FASTICA method. We demonstrate, using signal-to-noise ratio and correlation indicators, that the technique used is superior to the wavelet and moving average filter and outperforms the proposed denoising NIRS signal.
Nowadays, brain function evaluation using Functional Near Infrared Spectroscopy (fNIRS) is one of the most potential non-invasive monitoring techniques. This paper concerns usefulness of the NIRS signals denoising using the Hemodynamic Evoked Response (HomER) as graphical user interface displays the NIRS data, fast independent component analysis (FASTICA) method to reduce data dimension and the combined Wavelet & PCA method for enhancing NIRS signals. NIRS signals include many types of noise, spread across a broad spectrum of frequencies, such as: low frequency noise from respiratory interference, 0.1 - 0.3 Hz, Mayer wave, about 0.1 Hz, cardiac interference, 0.8 - 2.0 Hz, and other artifacts from head and facial motions. Meanwhile, electronic components generate high frequency noise. Multi-resolution wavelet and PCA was applied successfully to enhance the NIRS signals. It consists of adaptively modifying the wavelet coefficients based on the degree of noise contamination of the processed NIRS signal. This is done subsequently to the signal pre-processing by reducing data dimension using the FASTICA method. We demonstrate, using signal-to-noise ratio and correlation indicators, that the technique used is superior to the wavelet and moving average filter and outperforms the proposed denoising NIRS signal.
