Near-infrared spectroscopy brain signals

Abstract

Brain signal S of near-infrared spectroscopy (fNIRS) in subject 1 CS The method comprises the steps of placing a near-infrared emitter 2 and respective proximal and distal near-infrared detectors 3 on the skin of the head of a subject 1, and recording a near-infrared signal during a baseline recording phase t1 in which the subject is in a resting state, the recorded signal being a baseline deep signal B DS and baseline shallow signal B SS and a given task frequency f t calculating a scaling factor K between the amplitude of the baseline deep signal and the amplitude of the baseline shallow signal at t; and recording a near-infrared signal while the subject is receiving periodic brain stimulation at the task frequency during a stimulation recording phase t2, wherein the recorded signal is a shallow signal S SS and deep signal S DS applying a scaling factor to the shallow signal; and a task frequency f t and calculating the brain signal at the task frequency as the difference between the deep signal at the task frequency and the scaled shallow signal at the task frequency.

Description

[Technical field]

[0001] The present invention relates to a method for obtaining near-infrared spectroscopy brain signals in a subject using periodic brain stimulation at a predetermined task frequency, the brain signals being clean and free of interfering components from other parts of the body, such that the brain signals primarily correspond to brain activity. [Background technology]

[0002] Based on the neurovascular coupling principle, functional near-infrared spectroscopy (fNIRS) aims to detect hemodynamic changes caused by neural oxygen consumption. NIRS is a non-invasive optical imaging technique that is widely used to measure cerebral (most often cortical) activity through relative concentration changes in oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR) (for reviews, see Obrig and Villringer, 2003; Pinti et al., 2018).

[0003] A major challenge in fNIRS studies is to reliably separate hemodynamic responses due to neurovascular coupling from other confounding components (Tachtsidis and Scholkmann, 2016). fNIRS changes evoked by brain activity are low amplitude in nature and unfortunately overlap with other fluctuations that do not originate from the cerebral cortex. Such fluctuations mainly include (i) systemic hemodynamic activity detectable in both brain and extracerebral regions (Bauernfeind, Wriessnegger, Daly, & Muller-Putz, 2014; Minati, Kress, Visani, Medford, & Critchley, 2011; Tachtsidis et al., 2009), (ii) regional blood flow changes in superficial tissue layers throughout the head (Kirilina et al., 2012), and (iii) instrumentation noise and other artifacts. The first two (which are far from simply spontaneous) can also be caused by cognitive, emotional, or physical tasks. Modulation of these non-cortical task-related components of the signal can be an important source of interference and noise if they mimic the dynamics of the brain activation of interest (Nambu et al., 2017; Nasi et al., 2013; Zimeo Morais et al., 2017). The impact of this modulation of task-related components can be so great that Takahashi et al. (2011) showed that task-related changes in skin blood flow (SBF) could explain more than 90% of the NIRS signal in a verbal fluency experiment, and Minati et al. (2011) demonstrated a strong confounding effect of arterial blood pressure (ABP) fluctuations.

[0004] To better infer the presence of a functional response, experimental protocols attempt to increase statistical power by repeating the stimulus a sufficient number of times with intervening control conditions in which a different response (or no response) is expected. Therefore, fNIRS experiments often used blocked or event-related designs, depending on whether one wishes to analyze sustained or transient responses, respectively (Pinti et al., 2018). Event designs usually use brief stimuli whose order is randomized and separated by a fixed or jittered interstimulus interval. Block designs attempt to maintain intellectual engagement by presenting stimuli for a sufficiently long time interval in one condition, followed by an interstimulus interval for a different condition or rest (Amaro and Barker, 2006). Mixed designs can also be used to investigate interactions between "sustained" and "transient" responses (Petersen and Dubis, 2012).

[0005] Depending on the stimulus presentation strategy, various analytical methods have been developed to make inferences about the functional hemodynamic response and to isolate it from confounding interferences (for a review, see Sungho Tak and Ye, 2014). Classical averaging strategies give robust results, but they are being progressively replaced by more powerful methods, since regular averaging-based statistical tests such as t-tests or ANOVAs do not allow inferences about the shape or time course of the fNIRS signal. These methods include the general linear model (GLM) framework (K. Friston, Ashburner, Kiebel, & Nichols, 2007; Schroeter, Bucheler et al., 2004), data-driven approaches such as principal component analysis (PCA) and independent component analysis (ICA) (Kohno et al., 2007; Yiheng and Zhang, 2009). These include the use of GLM (Zhang, Brooks, Franceschini, & Boas, 2005), as well as dynamic state space modeling (Diamond et al., 2006; Kolehmainen, Prince, Arridge, & Kaipio, 2003). GLM is one of the most widely adopted statistical frameworks to quantify how well the measured fNIRS signal fits a hemodynamic model reflecting the expected neural response. It takes advantage of the good temporal resolution of fNIRS and allows the inclusion of various covariates (e.g., physiological signals) in the regression model. In its most basic form, the model is obtained by convolving a hemodynamic response function (HRF) with a stimulus function that encodes a hypothetical time course of the neural response (Koh et al., 2007; Sungho Tak and Ye, 2014).Thus, GLM is a hypothesis-driven approach that requires the combination of specific HRFs (often taken from fMRI studies) with other disturbing regressors to build a linear model, which may not be clear depending on the task type, brain region, and participant specificity. Furthermore, GLM requires special care when applied to fNIRS signals due to several statistical issues (Huppert, 2016; Huppert, Diamond, Franceschini, and Boas, 2009; Koh et al., 2007). In contrast, PCA and ICA methods rely only on the general statistical assumptions of orthogonality and independence, respectively. Separating the mixture components that make up the fNIS signal is useful, but additional processing is required to clarify which of them are task-related and which are not, which is particularly challenging when extra- and intra-cerebral responses are correlated (Zhou, Sobczak, McKay, & Litovsky, 2020). State-space models, mainly based on Kalman filters, allow for the construction of complex hemodynamic models to describe the time-varying characteristics of the fNIRS signal and estimate the HRF. Dynamic analysis appears to provide better estimates of the HRF and better account for non-stationary signals, but still requires improvements in the model specification and state-space estimators.

[0006] Regardless of the advantages and disadvantages of each experimental method, all would benefit from the inclusion of short-distance recordings to obtain a reference of the superficial layers that contribute to the fNIS signal (for reviews, see Fantini, Frederick, & Sassaroli, 2018; Tachtsidis & Scholkmann, 2016; Sungho Tak & Ye, 2014). Multiple-distance measurements are considered to be particularly effective in isolating real brain responses. However, several open questions remain regarding, for example, the ideal range of distances between the signal source and the detector, the optimal number of short channels, and their arrangement configuration relative to the long channels. Ideally, each long channel should be paired with at least one nearby short channel, as there is increasing evidence showing the heterogeneous nature of superficial hemodynamics (Wyser et al., 2020). Unfortunately, such strict spatial organization of paired measurements is not currently possible with most commonly used NIRS devices today. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] U.S. Pat. No. 1,101,6567 Summary of the Invention [Problem to be solved by the invention]

[0008] It is therefore an object of the present invention to provide a method for acquiring near-infrared spectroscopy brain signals in a subject having access only to long and short channels. [Means for solving the problem]

[0009] The present invention describes a method for acquiring near-infrared spectroscopy brain signals in a subject. Essentially, the method includes the steps of placing a near-infrared emitter and respective proximal and distal near-infrared detectors on the skin of the subject's head, recording by a computer near-infrared signals received from the near-infrared emitter at the proximal and distal near-infrared detectors during a baseline recording phase in which the subject is at rest, the recorded signals including a baseline deep signal received by the distal near-infrared detector and a baseline shallow signal received by the proximal near-infrared detector, calculating by the computer a scaling factor between the amplitude of the baseline deep signal and the amplitude of the baseline shallow signal at a predetermined task frequency, and recording by the computer near-infrared signals received from the near-infrared emitter at the proximal and distal detectors while the subject is receiving periodic brain stimulation at the task frequency during a stimulation recording phase, the recorded signals including the deep signal received by the distal near-infrared detector and the shallow signal received by the proximal near-infrared detector. The method then comprises the steps of obtaining by a computer the brain signal at the task frequency during stimulation by applying a scaling factor to the shallow signal and calculating the brain signal at the task frequency as the difference between the deep signal at the task frequency and the scaled shallow signal. The steps of the method can be performed by a computer or data processing device programmed accordingly. The computer can be a single computer including software conveniently programmed to perform the steps of the method. Also, as known to those skilled in the art, embedded devices or distributed computer systems or other known computer arrangements can be used to perform the method.

[0010] Advantageously, performing periodic brain stimulation at a given task frequency, such as a periodic cognitive task, induces periodic hemodynamic fluctuations measurable in fNIRS recordings and thus generates oscillatory conditions suitable for effective analysis in the frequency domain. Thus, a mental arithmetic task can be employed in a periodic block design at a specific frequency while performing multiple distance recordings over the frontal regions.

[0011] Whereas previously surface fluctuations were merely considered a troublesome confound to be removed, this is not the case in the present invention, which instead sees fluctuations as carriers of valuable information, information that may prove essential not only for gaining a deeper understanding of the fNIRS data but also, perhaps equally important, for a more rigorous assessment of the overall dynamics of the neuro-visceral link, allowing the brain signal to be calculated from deep and superficial signals during periodic brain stimulation at a given task frequency.

[0012] Task-related arousal mechanisms require a tight interplay between cognitive functions and autonomic regulation (Forte, De Pascalis et al., 2019; Forte, Favieri et al., 2019; Nicolini et al., 2014; Thayer and Lane, 2009; Wang et al., 2016). Autonomic regulation is therefore thought to be linked to activity levels in executive brain regions, which allows adaptive responses to environmental demands. Conversely, autonomic failure may be associated with deterioration of certain cognitive functions, specifically executive functions (Forte, De Pascalis et al., 2019; Forte, Favieri et al., 2019). Such tight coordination of extra- and intracerebral responses with task frequencies may have great functional value. Correct coupling between physiological responses may be a sign of proper cognitive and / or cardiovascular function and of its breakdown, i.e., a potential early marker of cognitive decline and / or cardiovascular disease. Thus, brain signals at task frequencies and their correlation with extracerebral responses may be useful in the diagnosis and evaluation of dementias such as Alzheimer's disease, neurodevelopmental disorders such as autism and ADHD, and autonomic failure, among other clinical concepts involving altered brain function.

[0013] According to one embodiment of the present invention, the brain stimulation is an intellectual or cognitive activity, such as a mental arithmetic task, so that the near-infrared spectroscopy brain signal is consistent with a brain response evoked by the intellectual or cognitive activity.

[0014] According to another embodiment of the present invention, the brain stimulation may be visual, auditory, olfactory, gustatory, somatosensory, or motor activity, so that the near-infrared spectroscopy brain signal corresponds to the brain response evoked by the corresponding brain stimulation.

[0015] According to one embodiment of the present invention, the step of acquiring the brain signal at the task frequency during stimulation includes acquiring the phase, i.e., phase angle, and amplitude of the brain signal by a computer. Since the brain signal is expected to have a frequency equal to the task frequency, the components required to acquire the brain signal are its amplitude and phase since the frequency is known. Thus, the difference between the filtered deep signal and the scaled shallow signal at the task frequency provides the phase, i.e., phase angle, and amplitude of the brain signal.

[0016] According to one embodiment of the present invention, the step of acquiring the phase and amplitude of the brain signal during stimulation by the computer includes the steps of determining by the computer a deep signal phasor corresponding to the deep signal during stimulation at the task frequency, where the deep signal phasor has a phase that is the phase difference between the deep signal and the shallow signal at the task frequency and an amplitude that is the amplitude of the shallow signal at the task frequency; determining by the computer a shallow signal phasor corresponding to the shallow signal during stimulation at the task frequency, where the shallow signal phasor has a reference phase (e.g., 0°) and the amplitude of the shallow signal at the task frequency multiplied by a scaling factor; determining by the computer the brain signal phasor by subtracting the shallow signal phasor from the deep signal phasor, i.e., calculating their difference; and calculating by the computer the phase and amplitude of the estimated brain signal at the task frequency. Advantageously, it has been found that the significant components of the frequencies of the shallow and deep signals are at a frequency equal to the task frequency, so that the brain signal can be obtained by appropriately combining the phasors at the task frequency derived from the shallow and deep signals.

[0017] According to one embodiment of the present invention, the phase difference between the deep signal and the shallow signal at the task frequency is obtained by calculating an empirical transfer function in the frequency domain between the deep signal and the shallow signal by a computer, and calculating an argument of the transfer function at the task frequency by a computer, so that the phase of the brain signal corresponding to the functional brain activity can be estimated by a computer by using the empirical transfer function to obtain the temporal coordination between the extracerebral response and the cerebral response.

[0018] According to one embodiment of the present invention, the step of calculating by a computer the scaling factor between the amplitude of the base deep signal and the amplitude of the base shallow signal at the task frequency includes the steps of calculating by a computer the approximation of the complex frequency-dependent basis transfer function between the base deep signal and the base shallow signal, and the step of determining by a computer the scaling factor as the gain of the basis transfer function at the task frequency, so that the empirical transfer function can also be used by a computer to estimate the gain of functional brain activity, which can be applied to the acquisition of brain signals.Therefore, both the gain or amplitude and phase of the brain signal can be acquired by a computer using the transfer function applied to the deep signal and the shallow signal.

[0019] According to one embodiment of the present invention, the task frequency is between 0.015Hz and 0.07Hz, so that the natural vibrations of the body, such as the vibrations caused by the heartbeat, breathing, or Mayer arterial pressure waves, do not affect the brain signals. Preferably, the task frequency is between 0.025Hz and 0.05Hz, so that the task frequency is sufficiently far from the known natural vibrations of the body. More preferably, the task frequency is at a frequency of 0.033Hz, so that the corresponding period is 15 seconds, which can be easily measured to calculate the duration of each period during the periodic brain stimulation.

[0020] According to one embodiment of the invention having multiple groups of near-infrared emitters and respective proximal and distal near-infrared detectors grouped by region of interest, the method further includes a step of averaging by a computer the shallow and deep signals acquired for each group, whereby the signals are averaged to reduce the signal-to-noise ratio for each region of interest.

[0021] According to one embodiment of the present invention, the step of recording during stimulation includes alternating half-cycles of stimulating the subject with a periodic intellectual task and half-cycles of baseline rest, the task frequency being maintained during the recording phase, allowing to promote deep and shallow signal components at the task frequency. It is therefore also expected that the stimulation half-cycles and the baseline resting half-cycles have the same duration.

[0022] Also disclosed is a system for acquiring near-infrared spectroscopy brain signals in a subject, the system comprising a device comprising a near-infrared emitter and respective proximal and distal near-infrared detectors, the device adapted to place the near-infrared emitter and respective proximal and distal near-infrared detectors on the skin of the subject's head, recording near-infrared signals received from the near-infrared emitter at the proximal and distal near-infrared detectors during a baseline recording phase in which the subject is in a resting state, the recorded signals including a baseline deep signal received by the distal detector and a baseline shallow signal received by the proximal detector, and measuring the amplitude and baseline depth of the baseline deep signal at a predetermined task frequency. and a computer comprising means for performing the steps of: calculating a scaling factor between the amplitude of the shallow signal and the amplitude of the deep signal at the task frequency, recording near infrared signals received from the near infrared emitters at the proximal and distal detectors while the subject is receiving periodic brain stimulation at the task frequency during a stimulation recording phase, the recorded signals including the shallow signal received by the proximal detector and the deep signal received by the distal detector, obtaining the brain signal at the task frequency during stimulation by applying the scaling factor to the shallow signal and calculating the brain signal at the task frequency as the difference between the deep signal at the task frequency and the scaled shallow signal. The difference between the deep signal at the task frequency and the scaled shallow signal can be calculated by the computer as the difference between the phasor of the deep signal at the task frequency and the phasor of the scaled shallow signal, such that the phase and amplitude of the brain signal can be calculated by the computer in a non-invasive manner.

[0023] As a supplement to the description provided herein, and for the purpose of facilitating a better understanding of the nature of the present invention, this specification is accompanied by a set of drawings, represented below, by way of illustration and not by way of limitation. [Brief description of the drawings]

[0024] [Figure 1]FIG. 1 illustrates an example of brain stimulation for a given task frequency. [Diagram 2] FIG. 1 shows an example of the placement of emitters and respective proximal and distal detectors on the skin of a subject's head. [Diagram 3] FIG. 3 is a schematic diagram showing shallow and deep channels between the emitter and detector of FIG. 2. [Figure 4] FIG. 2 shows recordings of filtered signals of shallow and deep channels at a given task frequency during baseline and stimulation recording phases of the brain stimulation of FIG. 1. [Figure 5a] FIG. 5 illustrates the gain of the transfer function between the filtered signals of the shallow and deep channels during the baseline recording phase of FIG. 4. [Figure 5b] FIG. 5 shows the phase of the transfer function between the filtered signals of the shallow and deep channels during the stimulation-recording phase of FIG. 4. [Figure 6a] FIG. 5 shows the averaged periods of the shallow and deep signals of FIG. 4 during the stimulation-recording phase. [Figure 6b] FIG. 6b shows the scaled shallow and deep signals of FIG. 6a. [Figure 6c] FIG. 6b shows the brain signal obtained as the difference between the scaled shallow and deep signals of FIG. 6b. [Figure 7] FIG. 1 illustrates obtaining a brain signal as the difference between shallow and deep signals scaled using a phasor. [Figure 8] FIG. 13 shows another example of the placement of emitters and respective proximal and distal detectors on the skin of a subject's head. [Figure 9] FIG. 9 is a schematic diagram showing shallow and deep channels between the emitter and detector of FIG. 8. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0025] FIG. 1 shows a system 100 comprising a device 10 and a computer 20 adapted to carry out the steps of the method for measuring a near-infrared spectroscopy (fNIRS) brain signal S in a subject 1 using the method of the invention. CS A predetermined task frequency f t An example of periodic brain stimulation in

[0026] Since brain stimulation is an intellectual or cognitive activity, the corresponding acquired brain signals S CS is associated with this intellectual or cognitive activity. Figure 1 shows the half-period T t / 2 and the half-period of rest T t / 2, i.e., a periodic pattern of mental effort with regular repetition of activation-rest, based on a blocking protocol designed for a given task frequency f t The periodic brain stimulation in Fig. 1 shows the periodic brain stimulation in a human brain. The idea behind this is to induce periodic hemodynamic changes in the form of cycles of certain responses to the subject 1 followed by a return to basal levels. In this way, such oscillation patterns may be analyzed by conventional spectroscopy. This periodic brain stimulation can also be programmed into a computer.

[0027] As shown in FIG. 1, the periodic brain stimulation is organized into three consecutive non-stop recordings: (i) a 300-second baseline in a resting condition as the baseline recording phase t1, (ii) a 300-second mental arithmetic task as the stimulation recording phase t2, and (iii) a 300-second recovery in a relaxed state. In this case, the stimulation recording phase t2 is a 30-second period T t Each trial consisted of 10 consecutive 30-second trials with a half-period T t / 2, followed by a half-period T of 15 seconds of relaxation pause t / 2 followed by 0.01 / 0.02. To perform the mental arithmetic, participants were asked to repeatedly subtract a small number (between 5 and 9) from a three-digit number (between 100 and 199) as quickly as possible. Both digits, chosen randomly in each trial, were presented on a 21.5-inch display monitor 80 cm from the participants' eyes. A pause was then initiated by the presentation of the question "Result?" for 5 seconds, prompting the participant to verbally inform the participant of the final result of the mental arithmetic (to allow for scoring of performance and to ensure that the participant was paying attention), followed by the presentation of a calming image during which the participant was instructed to relax. To signal the start of the mental arithmetic, a fixation cross was presented in the center of the computer screen 2 seconds before the presentation of the subtraction operand. Furthermore, during the stimulus recording phase t2, the importance of exerting mental effort, rather than the amount or accuracy of the operations performed, was repeatedly emphasized.

[0028] In this case, the 30-second mental arithmetic trial was performed at a task frequency of 0.033 Hz. t This frequency was chosen to avoid overlap with well-known natural fluctuations such as ABP (0.08–0.12 Hz) and very slow endothelial activity (0.01–0.02 Hz). Furthermore, the 15-s duration of mental effort matches the duration of a typical hemodynamic response, and the following 15-s pause represents an optimal inter-event interval to allow return to baseline levels and minimize overlap between successive hemodynamic responses. However, other task frequencies, e.g., frequencies between 0.015 Hz and 0.07 Hz, may be used. t are also envisaged, and preferably frequencies between 0.025 Hz and 0.05 Hz may also be used.

[0029] In this case, the periodic brain stimulation is a mental arithmetic task, i.e., an intellectual or cognitive activity, so the brain signal S to be acquired is CS is the subject 1's response to such intellectual or cognitive activity. However, in other embodiments, brain stimulation may be performed without the brain signal S to be obtained as the subject 1's response. CS Depending on the sensory input, the activity may be visual, auditory, olfactory, gustatory, somatosensory, or motor, or even other activity.

[0030] As shown in Figure 2, the functional near-infrared spectroscopy (fNIRS) brain signal S in subject 1 CS To obtain a predetermined task frequency f, a near-infrared emitter 2 and two detectors 3, a proximal detector 3a and a distal detector 3b, mounted on the device 10 of the system 100, are placed on the skin of the subject's 1 head, typically in the frontal pole region. t Near-infrared spectroscopy (fNIRS) brain signals in subject 1 during periodic brain stimulation in CS It is possible to obtain the following.

[0031] The proximal detector 3a is positioned closer to the emitter 2 than the distal detector 3b. For example, the proximal detector 3a is positioned 14 mm from the emitter 2, and the distal detector 3b is positioned 32 mm from the emitter 2. Of course, other arrangements of the various emitters 2 and detectors 3 are possible in the device 10, and the near-infrared spectroscopy brain signal S in the subject 1 may be measured in a similar manner as described below. CS They can be used to obtain

[0032] The wavelengths of the emitter 2 and detector 3 are adapted to detect differences caused by brain stimulation, for example near infrared wavelengths of 740 and 850 nm will be suitable to detect relative changes in the concentrations of oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR), so that the acquired brain signal is proportional to the absorbance of the emitted and received wavelengths.

[0033] 3 shows a representation of a shallow channel SS established between an emitter 2 and a proximal detector 3a, and a deep channel DS established between an emitter 2 and a distal detector 3b. The emitter 2 may be a light emitting diode emitting the wavelength of interest, in this case 850 nm, and the detector 3 may be a photodiode or phototransistor, also known as an optode, adapted to receive the wavelength of interest and generate a corresponding electrical signal, which is transmitted, for example wirelessly, and processed by the computer 20 of the system 100.

[0034] 4-7 show how the brain signal S is calculated by computer 20 based on changes detected in oxygenated hemoglobin (HbO). CS Regarding the method of detecting HbO, other brain signals based on changes detected in other components at other wavelengths may be used, such as deoxygenated hemoglobin (HbR) or even total hemoglobin (HbT) or the isosbestic point at 804-805 nm where the absorbance caused by HbO and HbR becomes the same. Thus, other wavelengths may be emitted or received as needed, depending on the difference in the components to be detected to obtain the corresponding brain signal.

[0035] FIG. 4 shows an example of recordings of signals from the shallow channel SS and the deep channel DS during a baseline recording phase t1, during which the subject is in a resting state, and a stimulation recording phase t2, recorded by the computer 20. As mentioned above, the stimulation recording phase t2 is a half-period T t / 2 and the half-period T during which subject 1 has a baseline rest period during which he recovers from the intellectual task. t / 2, and the stimulation half-period T t Half-cycle T / 2 and baseline rest t Advantageously, the half-period T t / 2 and baseline resting half-cycle T t / 2, the given task frequency f t Periodic brain stimulation in the brain is induced.

[0036] The signal shown in Figure 4 has a task frequency f t are adapted and filtered by computer 20 to be around the same task frequency f t The components of the signal at the task frequency f are acquired. In this case, the raw optical data of each detector 3 are converted to optical density and then converted to the relative concentration changes of oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR) via the extended Beer-Lambert law. In this case, for illustration purposes, only the relative concentration of oxygenated hemoglobin (HbO) is shown. The data is filtered at the task frequency f using a zero-phase filter, in this case with a filter width of 0.015 Hz. t The computer 20 performs digital band filtering to ensure that the task frequency f t Other filtering and filter widths can also be used to obtain the before and after signals.

[0037] As can be seen in FIG. 4, during the baseline recording phase t1 in which the subject is in a resting state, the task frequency f t The near infrared signal at is recorded by computer 20, and the recorded signal is a baseline deep signal B received by the distal near infrared detector. DS and the baseline shallow signal B received by the proximal near-infrared detector. SS It included:

[0038] During the stimulation recording phase t2, the subject recorded the task frequency f t The task frequency f received from the near-infrared emitter 2 at the proximal and distal near-infrared detectors 3 while receiving periodic brain stimulation at t The near infrared signal at is recorded by the computer 20, and the recorded signal is a shallow signal S received by the proximal detector 3a. SS and the deep signal S received by the distal detector 3b. DS It included:

[0039] The fNIRS signal does not originate from the cerebral cortex due to functional activity, but is dominated by confounding hemodynamics originating from blood flow changes in superficial tissues, as well as changes in systemic physiology present both in superficial layers and in the brain tissue itself, albeit on different time scales. An effective strategy to address this issue is the use of multiple distance measurements. Assuming that recordings with short separation distances are only sensitive to extracerebral changes, and recordings with long separation distances are sensitive to both extracerebral and cerebral activity, the shallow component is removed from the deep signal.

[0040] Therefore, the task frequency f t In this case, the brain signal S CS To obtain the deep signal S DS From shallow signal S SS The removal of the shallow components present in is represented by the linear combination: S CS =S DS -(KS SS )

[0041] Task frequency f t These signals in are considered to be sinusoidal and therefore differ only in their amplitude and phase.

[0042] The scaling factor K is the task frequency f t Baseline deep signal in B DS and baseline shallow signal B SS Alternatively, the baseline deep signal B may be calculated by the computer 20 as the ratio of the amplitude between DS and baseline shallow signal B SS Average peak-to-peak ratio between baseline deep signal B and DS and baseline shallow signal B SS and the ratio between the root mean square measurements of the baseline deep signal B DS Amplitude of baseline shallow signal B SS There are several known methods for computationally calculating this scaling factor K as the ratio of the amplitude of

[0043] In addition, because transfer function models have become a common approach to investigate the dynamics of cerebrovascular autoregulation (Claassen et al., 2015; Van Beek, Claassen, Rikkert, and Jansen, 2008) and have also been used to remove noise from systemic physiology from fNIRS signals (Bauernfeind, Bock, Wriessnegger, and Muller-Putz, 2013; Florian and Pfurtscheller, 1997), a transfer function strategy can also be used by a computer to calculate the scaling factor K. SS is the task frequency f t Assuming that the signal contains quasi-periodic oscillations, with energy in the frequency range of interest before and after the signal, the transfer function H(f) can be approximated from experimental fNIRS data as follows (Zhang et al., 1998):

[0044]

number

[0045] In the case of a time series pair, the shallow signal S SS and deep signal S DS is the task frequency f t are the input and output signals, respectively, used to obtain an approximation of the transfer function in μM between the input and output. From the complex-valued results, a magnitude (gain) was obtained, which corresponds to a scaling factor K that represents the relative change in μM between the input and output, and a phase that conveys their temporal relationship (phase difference or time lag). To report the gain, the computer 20 converted the data into a percentage value resulting in a scaling factor K. The scaling factor K thus represents the change in μM between the input and output during the "baseline", i.e., the baseline recording phase t1. tcan be thought of as the gain value of the basis transfer function bTF at CS When the component of does not contribute, the deep signal S DS Shallow signal S present in SS Represents the ratio of the size of

[0046] Therefore, the task frequency f t Deep base signal B in DS and shallow base signal B SS In the step of calculating the scaling factor K between the amplitude of the deep basal signal B, as shown in FIG. DS and shallow base signal B SS The computer 20 calculates an approximation of the complex frequency-dependent basis transfer function bTF between the task frequency f t The scaling factor K can be determined as the gain of the basis transfer function bTF in

[0047] Task frequency f t Deep signal in S DS and shallow signal S SS The phase difference between the task frequency f t The deep signal S filtered at DS and shallow signal S SS This phase difference can be obtained by determining with the computer 20 the delay between the deep signal S during the stimulation recording phase t2. DS and shallow signal S SS 5b, the empirical transfer function TF in the frequency domain between t The brain signal S can also be obtained by calculating the argument of the transfer function TF in the CS The scaling factor K and the phase difference used to calculate

[0048] Figure 6a shows the deep signal S DS Period and shallow signal S SS6b shows the averaged period of the shallow signal S SS 6a shows the signal in FIG. 6a with a scaling factor K applied to it, thus reducing its amplitude.

[0049] The brain signal S mentioned above CS To obtain the deep signal S DS and shallow signal S SS Given a linear combination between t Brain signals in S CS is the task frequency f t Deep signal in S DS and the scaled shallow signal KS SS 6c. The brain signal S can be calculated by the computer 20 as the phase and amplitude difference between CS is obtained.

[0050] Advantageously, the deep signal S DS and shallow signal S SS is the task frequency f t Since the deep signal S is a sinusoidal signal with a frequency of DS and shallow signal S SS are the deep signal phasors X with their corresponding sinusoidal signal amplitudes and phases. DS and shallow signal phasor X SS The linear combinations shown above can also be expressed as linear combinations of the phasors shown below. S CS =S DS -(KS SS ) A CS cos(2πf t t+Φ CS )=A DS cos(2πf t t+Φ DS )-KA SS cos(2πf t t+Φ SS )

[0051]

number

[0052] Phase of the shallow signal Φ SS and the phase of the deep signal Φ DS One of the phases between is expected to be a reference phase, typically a 0 degree reference phase. In this case, the 0 degree reference phase is the shallow signal Φ SS is assigned to the phase of

[0053] In this way, the deep signal S DS and the scaled shallow signal KS SS The difference in phase and amplitude between the task frequency f t Brain signals in S CS , the brain signal phasor X has an amplitude and phase corresponding to the amplitude and phase of CS To obtain the deep signal phasor X, as shown in Figure 7, DS -Shallow signal phasor X SS can be calculated directly by the computer 20 as the subtraction of the phasors of

[0054] Therefore, during stimulation, the brain signal S CS The step of acquiring the phase and amplitude of the task frequency f t Deep signal S during stimulation DS Deep signal phasor X corresponding to DS determining by the computer 20 the deep signal phasor X DS is the task frequency f t Deep signal in S DS and shallow signal S SS and the task frequency f t Deep signal in S DS and a task frequency f t Shallow signal S during stimulation SS Shallow signal phasor X corresponding to SS determining a shallow signal phasor X SShas a reference phase of 0 degrees and is multiplied by a scaling factor K, t Shallow signal S SS Of course, alternatively, the 0 degree reference phase may be set to the deep signal phasor X DS or the reference phase may be any known phase since only the phase difference is relevant.

[0055] Then, the task frequency f t The estimated brain signal S CS The phase and amplitude of the deep signal phasor X is shown in Figure 7. DS From shallow signal phasor X SS By directly subtracting CS This can be calculated by the computer 20 by determining:

[0056] Although only one emitter 2 and a group of one each of proximal and distal detectors 3a and 3b have been used above, other multi-channel, wireless, continuous wave NIRS devices 10 can also be used, such as the Brainspy28 from Newmanbrain, SL, which employs four emitters 2 and ten detectors 3 forming a rectangular grid of 80×20 mm. In this case, each emitter 2 houses two light-emitting-diodes (LEDs) with wavelengths of 740 nm and 850 nm. As shown in FIG. 8 and in more detail in FIG. 9, through rigorous switching cycles, the device 10 combines pairs of optodes with different separation distances to provide 16 short or shallow channels SS and 12 long or deep channels DS, corresponding to distances between the signal source and the detectors that are 14 mm and 32 mm, respectively. Furthermore, it is expected that the device 10 can measure and correct for the contribution of ambient light and may also incorporate a three-axis accelerometer to account for head movement. The data can be wirelessly transmitted (e.g., via Bluetooth) at a sampling rate of 10 Hz to a computer 20, which processes the data to generate a near-infrared spectroscopy (fNIRS) brain signal S. CS Execute the steps to obtain

[0057] As shown in Figure 8, the NIRS probe can be applied to the frontal area centered at AFpz according to the international 10-5 system, mainly covering the fronto-polar area of ​​the prefrontal cortex (PFC). The optode is expected to contact the skin through an intermediate convex lens that presses against the skin when the probe is held firmly to reduce skin blood flow and therefore its hemodynamic interference.

[0058] Advantageously, by having multiple groups of near infrared emitters and respective proximal and distal near infrared detectors grouped by region of interest, the shallow and deep signals acquired per region can be averaged, improving the signal to noise ratio per region of interest and providing a better brain signal S per region of interest. CS You will be able to obtain the following.

Claims

1. A method for acquiring a brain signal (S ＣＳ ) of functional near-infrared spectroscopy (fNIRS) in a subject (1), comprising: Placing a near-infrared emitter (2) and proximal and distal near-infrared detectors (3), respectively, on the skin of the head of the subject (1); During the baseline recording stage (t) when the subject is in a resting state １ a step of recording, by a computer (20), the near-infrared signals received from the near-infrared emitter (2) by the proximal and distal near-infrared detectors (3), wherein the recorded signals are The baseline deep signal (B) received by the distal detector (3b) ＤＳ ), and The baseline shallow signal (B) received by the proximal detector (3a) ＳＳ ), and A step comprising; The amplitude of the baseline deep signal (B ｔ ),) at a predetermined task frequency (f ＤＳ ) and the amplitude of the baseline shallow signal (B ＳＳ ) and calculating a scaling coefficient (K) by the computer (20); Stimulation recording stage (t ２ ), in a state where the subject is receiving periodic brain stimulation at the task frequency (f ｔ ), the step of recording, by the computer (20), the near-infrared signal received from the near-infrared emitter (2) by the proximal and distal detectors (3), wherein the recorded signal is The shallow signal (S) received by the proximal detector (3a) ＳＳ ), and The deep signal (S) received by the distal detector (3b) ＤＳ ), and A step comprising; Applying a scaling coefficient (K) to the shallow signal (S ＳＳ ), and The task frequency (f ｔ ), the deep signal (S ＤＳ ), and the scaled shallow signal (KS ＳＳ ), calculating the brain signal (S ｔ ) at the task frequency (f ＣＳ ) as the difference between them, and By performing, the brain signal (S ｔ ), at the task frequency (f ＣＳ ) during the stimulation, is acquired by the computer (20); A method characterized by comprising.

2. The method according to claim 1, Wherein the brain stimulation is an intellectual or cognitive activity; A method characterized by that.

3. The method according to claim 1, Wherein the brain stimulation is a visual, auditory, olfactory, gustatory, somatosensory, or motor activity; A method characterized by that.

4. The method according to claim 1, Stimulation recording stage (t ２ ), the step of obtaining the brain signal (S ｔ ) at the task frequency (f ＣＳ ) by the computer (20) includes the step of obtaining the phase and amplitude of the brain signal (S ＣＳ ) by the computer (20). A method characterized by that.

5. The method according to claim 4, The step of obtaining, by the computer (20), the phase and amplitude of the brain signal (S ＣＳ ) during the stimulation is The task frequency (f ｔ ), a step of determining, by the computer (20), a deep signal phaser (X ＤＳ ) corresponding to the deep signal (S ＤＳ ) during the stimulation at the task frequency, wherein the deep signal phaser (X ＤＳ ) is, The task frequency (f ｔ ), the phase difference between the deep signal (S ＤＳ ) and the shallow signal (S ＳＳ ), which is the phase (Φ ＤＳ ), and The amplitude (A ｔ ), which is the amplitude of the deep signal (S ＤＳ ) at the task frequency (f ＤＳ ), and Having a step of; The task frequency (f ｔ ), a step of determining, by the computer (20), a superficial signal phasor (X ＳＳ ) corresponding to the superficial signal (S ＳＳ ) during stimulation at the task frequency, wherein the superficial signal phasor (X ＳＳ ) is, The reference phase (Φ ＳＳ ) and The amplitude (A ｔ ), which is the amplitude of the shallow signal (S ＳＳ ) at the task frequency (f ＳＳ ) multiplied by the scaling factor (K), and Having a step of; Subtracting the shallow signal phasor (X ＤＳ ) from the deep signal phasor (X ＤＳ ) to determine a brain signal phasor (X ＣＳ ) by the computer (20); The estimated phase and amplitude of the brain signal (S ｔ ), at the task frequency (f ＣＳ ), are calculated by the computer (20); Including; A method characterized by that.

6. The method according to any one of claims 1 to 5, The phase difference between the deep signal (S ｔ ), and the shallow signal (S ＤＳ ) at the task frequency (f ＳＳ ), is obtained by calculating, by the computer (20), the empirical transfer function (TF) in the frequency domain between the deep signal (S ＤＳ ) and the shallow signal (S ＳＳ ), and calculating, by the computer (20), the argument of the transfer function (TF) at the task frequency (f ｔ ). A method characterized by that.

7. The method according to any one of claims 1 to 5, The step of calculating, by the computer (20), a scaling coefficient (K) between the amplitude of the deep base signal (B ｔ ), at the task frequency (f ＤＳ ), and the amplitude of the shallow base signal (B ＳＳ ), includes a step of calculating, by the computer (20), an approximation of a complex frequency-dependent base transfer function (bTF) between the deep base signal (B ＤＳ ) and the shallow base signal (B ＳＳ ), and a step of determining, by the computer (20), the scaling coefficient (K) as the gain of the base transfer function (bTF) at the task frequency (f ｔ ). A method characterized by that.

8. The method according to any one of claims 1 to 5, The task frequency (f ｔ ) is a frequency between 0.015 Hz and 0.07 Hz, A method characterized by that.

9. The method according to claim 8, The task frequency (f ｔ ) is a frequency between 0.025 Hz and 0.05 Hz, A method characterized by that.

10. The method according to claim 9, The task frequency (f ｔ ) is a frequency of 0.033 Hz, A method characterized by that.

11. The method according to any one of claims 1 to 5, Having a plurality of groups of near-infrared emitters and proximal and distal near-infrared detectors, respectively; wherein the method further includes a step of averaging, by the computer (20), the shallow signals (S ＳＳ ) and the deep signals (S ＤＳ ) obtained for each group A method characterized by that.

12. The method according to any one of claims 1 to 5, said step of recording by said computer (20) during said stimulation recording stage (t ２ ) is Half cycle (T ｔ / 2) for applying a stimulus to the subject, and Half cycle (T ｔ / 2) for taking the baseline quiet, and Including a step of alternately swapping; A method characterized by that.

13. The method according to claim 12, Wherein the half-cycle for performing the stimulation and the half-cycle for taking the baseline quiet have the same duration; A method characterized by that.

14. A system (100) for acquiring a brain signal (S ＣＳ ) of a near-infrared spectroscopy (fNIRS) in a subject (1), A device comprising a near-infrared emitter (2) and proximal and distal near-infrared detectors (3), respectively, and adapted to place the near-infrared emitter (2) and proximal and distal near-infrared detectors (3), respectively, on the skin of the head of the subject (1); a device (10); During the baseline recording stage (t) when the subject is in a resting state, １ a step of recording the near-infrared signals received from the near-infrared emitter (2) by the proximal and distal near-infrared detectors (3), wherein the recorded signals are The baseline deep signal (B) received by the distal detector (3b) ＤＳ and The baseline shallow signal (B) received by the proximal detector (3a) ＳＳ and A step comprising; The step of calculating a scaling factor (K) between the amplitude of the baseline deep signal (B ｔ ), and the amplitude of the baseline shallow signal (B ＤＳ ) at a predetermined task frequency (f ＳＳ ); Stimulation recording stage (t ２ ) during which, while the subject is receiving periodic brain stimulation at the task frequency (f ｔ ), recording, in the proximal and distal detectors (3), the near-infrared signal received from the near-infrared emitter (2), wherein the recorded signal is The shallow signal (S) received by the proximal detector (3a) ＳＳ ), and The deep signal (S received by the distal detector (3b) ＤＳ ) and A step comprising; Apply the scaling factor (K) to the shallow signal (S ＳＳ ), and The task frequency (f ｔ ), the deep signal (S ＤＳ ), and the scaled shallow signal (KS ＳＳ ), calculating the brain signal (S ｔ ) at the task frequency (f ＣＳ ) as the difference between them, By performing, the brain signal (S ｔ ), at the task frequency (f ＣＳ ) during the stimulation, is acquired; A computer (20) provided with means for executing; A system (100) characterized by comprising.

Images