Abstract—Functional near-infrared spectroscopy (fNIRS) is
an emerging non-invasive optical technique which can measure
cortical brain activities. This study describes the development of
one-channel fNIRS system designed for brain-computer
interface (BCI), six subjects participate in the experiment, the
concentration changes of oxy-hemoglobin (HbO
2
) and
deoxy-hemoglobin (Hb) during mental arithmetic task are
measured. And physiological noise in hemodynamic responses
are reduced based on empirical mode decomposition (EMD).
EMD is a data-driven and adaptive algorithm for analyzing
non-stationary data. This method decomposes the original data
into a set of intrinsic mode functions (IMFs). Noise can be
eliminated by selecting appropriate IMFs. Compared with
conventional butterworth filter, EMD achieves a higher contrast
to noise ratio (CNR) in physiological noise reduction.
I. INTRODUCTION
Neural activity detected by electroencephalography (EEG)
is a typical signal used for representing the brain state, and
EEG is a common modality applied in brain-computer
interface (BCI) [1, 2]. Functional near-infrared spectroscopy
(fNIRS) is a relatively new modality which provides a
non-invasive, low-cost, alternative option for BCI research [3,
4]. This technique utilizes an optical window in the
near-infrared light spectrum, the light within the spectral
range (600 – 900 nm) can easily penetrate brain tissues and be
absorbed by the hemoglobin in blood. Brain functional
activities can be assessed by the concentration changes of
deoxy-hemoglobin (Hb) and oxy-hemoglobin (HbO
2
).
This work is supported by General Reserve Department of PLA under
Grant 9140A26060214ZK63424 and by National Natural Science
Foundation (NNSF) of China under Grant 61203368, 61172145, 51305438.
Xuxian Yin is with State Key Laboratory of Robotics, Shenyang Institute
of Automation (SIA), Chinese Academy of Sciences (CAS), Shenyang
110016, P. R. China and University of Chinese Academy of Sciences,
Beijing 100049, P. R. China (email: yinxuxian@sia.cn).
Gang Shi is with State Key Laboratory of Robotics, Shenyang Institute of
Automation (SIA), Chinese Academy of Sciences (CAS), Shenyang 110016,
P. R. China (email: sg0105@sia.cn).
Zhidong Wang is with State Key Laboratory of Robotics, Shenyang
Institute of Automation (SIA), Chinese Academy of Sciences (CAS),
Shenyang 110016, P. R. China and Dept. of Advanced Robotics, Chiba
Institute of Technology, Chiba 2750016, Japan (e-mail:
zhidong.wang@it-chiba.ac.jp).
Hongyi Li is with State Key Laboratory of Robotics, Shenyang Institute of
Automation (SIA), Chinese Academy of Sciences (CAS), Shenyang 110016,
P. R. China and School of Mechanical Engineering & Automation,
Northeastern University, Shenyang, China (email: hli@sia.cn).
The measured hemodynamic signals are influenced by
cardiac and respiratory activities, also by low-frequency
oscillations (Mayer wave, around 0.1 Hz) [5]. Numerous
methods have been employed to reduce physiological noise,
including low-pass filtering [6], wavelet denoising [7],
independent component analysis (ICA) [8], Bayesian filtering
[9], etc. Low-pass filtering is a common method for reducing
physiological noise, while the disadvantage of low-pass
filtering is that both of hemodynamic responses and noise can
be reduced due to their components overlapping in the range
of frequency spectra.
Empirical mode decomposition (EMD) presented by
Huang et al. [10] is a data-driven algorithm for analyzing
non-stationary signals, this method is adaptive because none
of specified basis functions are used. The basis functions are
obtained from the signal itself.
In this study, one-channel fNIRS system was designed and
six subjects participated in mental arithmetic tasks to validate
the system performance. EMD algorithm was used to reduce
physiological noise which was mixed in the hemodynamic
responses elicited by brain functional activities.
II. MATERIALS AND METHODS
A. fNIRS Instrument
Recently, there are three optical measurement categories:
time domain (TD), frequency domain (FD) and continuous
wave (CW) measurements. TD and FD methods can measure
the path length of photons, and the absolute concentration
values of hemoglobin can be acquired. However, these
instruments are expensive, and technically demanding. In our
instrument, CW method was applied to measure the relative
concentration changes of hemoglobin. In cognitive studies,
the relative values can provide enough information to present
brain activation.
This instrument is made up of four parts: light emission,
light transmission, light conversion and signal processing. The
schematic diagram of this instrument is shown in Figure 1.
Two laser diodes at wavelengths of 695 and 830 nm were
employed as light sources. The laser diodes were modulated
with sinusoidal signals with the frequency of 1 and 2 kHz,
respectively. The light shed on the scalp through the fiber, and
avalanche photodiodes (APDs) were used to detect the light
scattering from the brain tissue. The optical fiber bundle with
a diameter of 2.5 mm was used to transmit the light into the
Development of one-channel fNIRS system and physiological noise
reduction in brain hemodynamic responses
Xuxian Yin, Gang Shi, Zhidong Wang, Hongyi Li
Proceedings of the 2015
IEEE Conference on Robotics and Biomimetics
Zhuhai, China, December 6-9, 2015
978-1-4673-9674-5/15/$31.00 © 2015 IEEE
763