A new approach to estimating the evoked hemodynamic response applied to dual channel functional near infrared spectroscopy
Introduction
Functional Near-Infrared Spectroscopy (fNIRS) is a relatively new noninvasive neuroimaging technique to discover hemodynamic variations within the cortex [1], [2], [3], [4]. It is an increasingly popular technology for brain function assessment that is due to its several advantages over other techniques (like EEG, fMRI, etc.) for being low-cost, portable, safe, having relatively high temporal resolution and a quick setup [5], [6], [7].
fNIRS is based on the fact that activation of a specific localized part of the brain leads to an increase in oxygen consumption in that region, which is accompanied by an increase in the total blood flow, regional blood volume and regional blood oxygenation due to neurovascular coupling [8], [9]. This leads to a change in the concentration of the local oxygenated haemoglobin (oxy-Hb) and deoxygenated haemoglobin (deoxy-Hb) [10]. Since oxy-Hb and deoxy-Hb have specific optical properties in the near-infrared light range (between 700 and 900 nm), the change in the concentration of these chromophores during neurovascular coupling can be detected noninvasively. In fNIRS, the near infrared light photons passing through the tissue are either scattered or absorbed. Photons which are not absorbed travel a banana shaped path and back to the surface of the scalp, the amount of which can be measured using a photodetector. Choosing appropriate wavelengths with regards the absorption coefficients of oxy-Hb and deoxy-Hb, it is possible to calculate variations in the concentrations of these chromophores using a modified Beer–Lambert law (MBLL) [11], [12].
fNIRS has the ability to measure the time variations in both oxy-Hb and deoxy-Hb noninvasively and without imposing a major limitation to the subject. In spite of this and other advantages, the presence of the systemic physiological interferences in the cortex tissue is one of the main problems in using fNIRS to study the brain function. The so called systemic physiological interferences are present not only in the brain tissue but also in all other superficial tissues over the scalp. If the distance between the light source and the photodetector is set to about 3 cm over the scalp, photons penetrate to a depth of about 1.5 cm within the cortex. On the other hand, if a photodetector placed about only 1 cm away from the source on the scalp, only hemodynamic changes of the superficial layers, which are only the physiologic interferences will be recorded [13], [14]. The so called physiological interferences include cardiac pulsation, respiration, blood pressure and Mayer wave fluctuations. These will result in a poor estimation of the evoked hemodynamic responses (EHRs) by fNIRS [15], [16]. Motion artifact and electrical noise are other interferences affecting fNIRS signal.
Various algorithms have been developed for hemodynamic response extraction from fNIRS signals in the past, which are either based on single, dual and/or multichannel fNIRS instruments. Block averaging (BA) and bandpass filtering (BPF) methods have been used to reduce the systemic artifacts from a single channel fNIRS recordings. In the BA method, actual hemodynamic responses are retrieved by time domain averaging of the N EHRs recorded during N uniform stimuli [17], [18]. The higher the number of recorded EHRs the better will be the results in the cost of more processing time.
Based on the assumption that the frequency contents of the physiological interferences are different than those of EHRs’, some fNIRS based BCI studies have used a band-pass filter to remove the physiological interferences [5], [19]. This method is not however very effective since there are partial overlap in the low frequency components of the EHRs and the physiological interferences (i.e. respiration and the Mayer wave) [20].
Both principal component analysis (PCA) and independent component analysis (ICA) are prevalent methods used for EHR extraction from fNIRS signals in multichannel recordings [21], [22], [23]. These methods are based on the assumptions that the spatial distribution of the systemic interferences and the EHRs is different. While unlike the localization of brain functional activity, systemic physiological interferences are generally global and their effects can be seen in all recorded channels [24].
The main problem with both PCA and ICA algorithms is that the noise intercrosses from noisy channels to other channels introducing negative impact on the final results. Several methods for physiological interferences reduction from the fNIRS signal using a dual channel fNIRS system were reported in the past [25], [26], [27], [28], [29], [30], [31], [32], [33]. Adaptive filtering approach is the most common method to reduce the systemic interferences from the fNIRS signals in dual-channel fNIRS systems [34], [35], [36], [37]. In this method, signal of the near channel is considered as the reference input while signal of the far channel is considered as the primary input. Different algorithms are used to find appropriate scaling coefficients for the reference signal in order to subtract the scaled reference signal from the primary input in order to remove interfering trends.
In this article, we present a novel algorithm to extract the brain hemodynamic responses from the fNIRS signals obtained by a dual channel fNIRS instrument (including near and far channels) which is superior to the other previously reported methods in the sense that it has no assumption on the amplitude, shape and duration of the hemodynamic responses, it is computationally low cost and simple, and needs no parameter updating. We employed the wavelet transform to decompose the near and far signals into approximation and detail sub-bands each with distinct frequency content. Next, with an ad-hoc and simple method the physiological interferences are eliminated from the signal of the far channel. In order to evaluate the performance of the proposed method, the results of the proposed method were compared to those produced by the recently reported method of EEMD-RLS [37]. For this comparison the three parameters of relative mean square error (rMSE), percentage estimation error of peak amplitude () and Pearson's correlation coefficient (CC) were used as quantitative criteria. Application of the proposed method to the synthetic data showed a significant improvement in recovery of the EHRs. The results showed that even in the case of a high level of noise and a short data length the proposed method is very effective. In continue a semi-real data was generated by adding simulated EHRs to the real resting state data. Application of the proposed method to this data also showed a remarkable improvement in recovery of the evoked hemodynamic response.
The rest of the paper is organized as follows. Section 2 introduces the proposed algorithm. Section 3 presents the results of applying the proposed algorithm to the synthetic and semi-real data. Finally Section 4 concludes the paper.
Section snippets
Dual channel functional near infrared spectroscopy
A dual channel fNIRS probe arrangement together with five-layered slab human head model schematically illustrated in Fig. 1. In dual channel fNIRS systems there are one near and one far channel. In fNIRS, the measurement depth depends on the distance between the source and the detector. In the near channel where the distance between the source and the detector is about 1 cm, only the hemodynamic variations from the superficial layers are recorded . On the other hand, in the far channel
The results obtained using the synthetic data
Example of the synthetic oxy-Hb changes in near and far channels with a signal to noise ratio (SNR) of 7 dB of synthetic subject 1 are shown in Fig. 5. The signals were passed through a bandpass filter (0.005–1 Hz) which was developed using a 4th order Butterworth filter in order to remove any instrumental and baseline drift noise. The shaded lines represent the onset of stimulation. In the proposed algorithm, first the quantity of L (wavelet decomposition level) and
Discussion
The fNIRS is a noninvasive neuroimaging tool that can be effectively used to study evoked hemodynamic changes within the brain. Using this technique, the variations in the cerebral oxyhemoglobin and deoxyhemoglobin concentrations can be estimated during functional activation of the cerebral cortex. The cerebral hemodynamic fluctuations of the systemic origin (physiological interferences) within the fNIRS signals are an important limitation on the fNIRS modality. These physiological
Conclusion
A novel effective method for EHR estimation from fNIRS signals was presented in this article. Using synthetic and semi-real data, it was demonstrated that the proposed method can reduce the global interference and improve the SNR of the evoked brain activity signals with a higher accuracy and simpler computing compared to the used EEMD-RLS method, making the proposed algorithm appropriate for real-time applications.
Conflict of interest statement
None of the authors have any conflicts of interest to declare.
Acknowledgments
The authors are grateful to the Cognitive Sciences and Technologies Council (CSTC) of Iran for its support of this research (contract number 117).
References (50)
- et al.
Classification of functional near-infrared spectroscopy signals corresponding to the right- and left-wrist motor imagery for development of a brain-computer interface
Neurosci. Lett.
(2013) - et al.
Decoding four different sound-categories in the auditory cortex using functional near-infrared spectroscopy
Hear. Res.
(2016) - et al.
A brief review on the history of human functional near-infrared spectroscopy (fNIRS) development and fields of application
NeuroImage
(2012) - et al.
Detection of dynamic changes in cerebral oxygenation coupled to neuronal function during mental work in man
Neurosci. Lett.
(1993) - et al.
A quantitative comparison of simultaneous {BOLD} fMRI and {NIRS} recordings during functional brain activation
NeuroImage
(2002) - et al.
Different time evolution of oxyhemoglobin and deoxyhemoglobin concentration changes in the visual and motor cortices during functional stimulation: a near-infrared spectroscopy study
NeuroImage
(2002) - et al.
A theoretical study of the signal contribution of regions of the adult head to near-infrared spectroscopy studies of visual evoked responses
NeuroImage
(1998) - et al.
Near-infrared spectroscopy: does it function in functional activation studies of the adult brain?
Int. J. Psychophysiol.
(2000) - et al.
Selective activation of the superior frontal gyrus in task-switching: an event-related fNIRS study
NeuroImage
(2008) - et al.
Improved recovery of the hemodynamic response in diffuse optical imaging using short optode separations and state-space modeling
NeuroImage
(2011)
Adaptive filtering to reduce global interference in non-invasive NIRS measures of brain activation: how well and when does it work?
NeuroImage
A reference-channel based methodology to improve estimation of event-related hemodynamic response from fNIRS measurements
NeuroImage
Evoked hemodynamic response estimation using ensemble empirical mode decomposition based adaptive algorithm applied to dual channel functional near infrared spectroscopy (fNIRS)
J. Neurosci. Methods
Non-synchronous behavior of neuronal activity, oxidative metabolism and blood supply during mental tasks in man
Neurosci. Lett.
Relation between asymmetry of prefrontal cortex activities and the autonomic nervous system during a mental arithmetic task: near infrared spectroscopy study
Neurosci. Lett.
A reference-channel based methodology to improve estimation of event-related hemodynamic response from fNIRS measurements
NeuroImage
A comparison of artifact reduction methods for real-time analysis of fNIRS data, Lecture Notes in Computer Science (including Subseries Lecture Notes in artificial intelligence and Lecture Notes in bioinformatics)
5618 LNCS
Online binary decision decoding using functional near-infrared spectroscopy for the development of brain--computer interface
Exp. Brain Res.
FNIRS-based brain-computer interfaces: a review
Front. Hum. Neurosci.
Decoding of four movement directions using hybrid NIRS-EEG brain-computer interface
Front. Hum. Neurosci.
Estimation of optical pathlength through tissue from direct time of flight measurement
Phys. Med. Biol.
Comment on the modified Beer–Lambert law for scattering media
Phys. Med. Biol.
Comparison of time-resolved and -unresolved measurements of deoxyhemoglobin in brain
Proc. Natl. Acad. Sci. USA
HomER: a review of time-series analysis methods for near-infrared spectroscopy of the brain
Cited by (11)
Functional Near-Infrared Spectroscopy as a Personalized Digital Healthcare Tool for Brain Monitoring
2023, Journal of Clinical Neurology (Korea)Actiview: A MATLAB-based toolbox for realtime cortical activation analysis using functional near-infrared spectroscopy
2019, ACM International Conference Proceeding Series