Adaptive filtering of physiological noises in fNIRS data

Abstract

The study presents a recursive least-squares estimation method with an exponential forgetting factor for noise removal in functional near-infrared spectroscopy data and extraction of hemodynamic responses (HRs) from the measured data. The HR is modeled as a linear regression form in which the expected HR, the first and second derivatives of the expected HR, a short-separation measurement data, three physiological noises, and the baseline drift are included as components in the regression vector. The proposed method is applied to left-motor-cortex experiments on the right thumb and little finger movements in five healthy male participants. The algorithm is evaluated with respect to its performance improvement in terms of contrast-to-noise ratio in comparison with Kalman filter, low-pass filtering, and independent component method. The experimental results show that the proposed model achieves reductions of 77% and 99% in terms of the number of channels exhibiting higher contrast-to-noise ratios in oxy-hemoglobin and deoxy-hemoglobin, respectively. The approach is robust in obtaining consistent HR data. The proposed method is applied for both offline and online noise removal.

Functional Near-Infrared Spectroscopy and Its Clinical Application in the Field of Neuroscience: Advances and Future Directions

“…Funane et al (2015) showed that hemoglobin signals obtained with short distances channels (∼1.5 cm) were better correlated with laser-doppler flowmetry measures of scalp blood flow than signals obtained from standard emitter/detector distances (∼3 cm) targeted to measurements of blood flow in adult cortex. Additionally, Nguyen et al (2018) showed there was no correlation (r < 0.38) of short separation channels to other physiological noises present in long range separation channels. Thus, the inclusion of short separation channels as a regressor in fNIRS analyses can reduce signal interference from scalp blood flow.…”

“…Nguyen et al (2018) applied this principle to reduce the amount of physiological noise present in fNIRS during a finger tapping task through the linear combination of the expected hemodynamic responses to the prescribed stimuli, short separation channel signals to detect extra-cortical noise, Fourier approximations of physiological noise (heart rate, respiratory, and blood pressure fluctuations), and baseline drift. Unknown model coefficients were estimated using a recursive least-square estimator to produce an adaptive filter that was able to reduce on average 77% of noise in oxyhemoglobin and 99% of noise in deoxyhemoglobin (Nguyen et al, 2018). Some limitations of this type of filtering include parameter tuning, the need to define noise distributions, and biased estimations if the filter is not closed loop (Abdelnour and Huppert, 2009).…”

“…More recently, short separation channels (∼8 mm source-detector distance) have been used to directly measure and remove scalp blood flow from fNIRS ( Gagnon et al, 2014 ; Funane et al, 2015 ; Nguyen et al, 2018 ). The short distance between the light emitters and detectors prevents light penetration to the cortical surface, limiting blood flow measurements to the scalp.…”

Detection of Mild Cognitive Impairment Using Convolutional Neural Network: Temporal-Feature Maps of Functional Near-Infrared Spectroscopy

“…The modified Beer-Lambert law was utilized to convert the optical densities to HbO and HbR (Sassaroli and Fantini, 2004). The converted signals passed 4th-order Butterworth low-and high-pass filters (i.e., cutoff frequencies: 0.001 and 0.1 Hz, respectively) to remove physiological noise, i.e., cardiac noise−1 Hz, respiration−0.25 Hz, and Mayer signal−0.1 Hz (Naseer et al, 2016;Khan and Hong, 2017;Nguyen et al, 2018). In accordance with our previously published evaluation results (Yang et al, 2019) and the relevant literature (Hoshi, 2007), it was observed that HbO is more sensitive and dependable than HbR.…”

Reduction of Onset Delay in Functional Near-Infrared Spectroscopy: Prediction of HbO/HbR Signals

“…where i represents the channel number; y is the measured HbO/ HbR; u is the desired hemodynamic response function (dHRF); w is the physiological noise; ε is the zero-mean Gaussian noise; a n , b m , c p , and c o are unknown coefficients that are recursively estimated; and n o , m o , and p o are the orders of the system, input, and exogenous signals, respectively. For fNIRS, the exogenous signal w consists of specifically three sinusoidal signals representing cardiac, Mayer, and respiration related physiological noises (Abdelnour and Huppert, 2009;Nguyen H.-D. et al, 2018). Also, the exogenous signals can be dropped out in the estimation process (i.e., p o = 0) if the prediction/tracking of the measured signal is focused.…”