Multi-channel near-infrared high-sampling-rate modulation method and system compatible with brain magnetism

Abstract

The invention discloses a brain magnetism compatible multi-channel near-infrared high-sampling-rate modulation method and system, and belongs to the field of multi-mode brain function imaging research, and the method comprises the steps: modulating a driving signal of an fNIRS light source to a high-frequency carrier; transmitting the modulated driving signal to each fNIRS light source by adopting a multi-channel frequency division multiplexing strategy, so that each fNIRS light source emits infrared light for detecting a target; and demodulating the detected infrared light to obtain the amplitude of the detected infrared light. According to the invention, non-contact perception of the target can be realized. According to the invention, electromagnetic interference in the MEG and fNIRS synchronous imaging process can be reduced, synchronous acquisition of brain magnetic signals and multi-channel high-sampling-rate near-infrared signals is realized by using a specific frequency modulation technology, and the near-infrared signals are accurately and rapidly demodulated.

Description

Technical Field

[0001] The present invention relates to the research field of multimodal brain functional imaging, and in particular to a brain magneto-compatible multi-channel near-infrared high sampling rate modulation method and system. Background Art

[0002] Functional brain imaging is an effective tool for revealing the mechanisms of brain activity. However, single-modality techniques, due to limitations in temporal and spatial resolution and detection depth, cannot meet the demands of complex brain science research and clinical precision diagnosis. Multimodal functional brain imaging, by integrating complementary technologies, enables the coordinated analysis of multidimensional data and has become a key development direction in neuroscience. Multimodal imaging offers the advantage of complementary temporal and spatial resolution. Functional magnetic resonance imaging (fMRI) has high spatial resolution (millimeter-level) but low temporal resolution (second-level); electroencephalography (EEG) has millisecond-level temporal resolution but insufficient spatial resolution; functional near-infrared spectroscopy (fNIRS) strikes a balance between the two but has limited penetration depth. Multimodal fusion can overcome the bottlenecks of single technologies. Secondly, multimodal imaging can achieve multidimensional coverage of physiological signals. EEG and magnetoencephalography can capture information on neuronal electrical activity, while near-infrared and ultrasound imaging can obtain information on blood oxygen metabolism. This allows for multi-parameter joint analysis, enabling a comprehensive understanding of brain functional networks.

[0003] Magnetoencephalography (MEG) based on optically pumped magnetometer (OPM) uses the spin polarization effect of alkali metal atoms (such as rubidium and cesium) in a magnetic field to achieve extremely weak magnetic field (10 -14 Compared with the traditional magnetoencephalography based on superconducting quantum interference device (SQUID), it can get rid of the limitation of liquid helium cooling and operate at low temperature. The sensor is placed close to the scalp, which significantly improves the signal sensitivity of deep brain regions. At the same time, the difficulty of wearable design is greatly reduced. The use of miniaturized OPM probe (<5cm 3 / unit), supports natural head movement and improves compatibility with other modalities. Active compensation technology can operate in partially shielded or even unshielded environments, reducing equipment deployment costs. However, compared to SQUID, OPM-MEG measurements require a near-zero magnetic field environment, and the magnetic field in most measurement environments must be controlled below 10nT. The requirements for magnetic shielding and ambient magnetic field interference are more stringent than those for SQUID MEG.

[0004] fNIRS uses the absorption characteristics of near-infrared light (650-950nm) in brain tissue to reflect the local cerebral hemodynamic response by measuring the concentration changes of oxyhemoglobin and deoxyhemoglobin. Near-infrared does not require magnetic shielding and can be monitored in real time in natural environments (such as exercise and social interaction). The time resolution is about 0.1-1Hz and the spatial resolution is 2-3cm, which is suitable for cortical function positioning. Single-modality near-infrared imaging equipment uses optical fiber or LED light sources. The equipment is lightweight and suitable for portable detection and multimodal fusion. Traditional continuous-wave near-infrared brain functional imaging equipment is limited by the modulation frequency of the light source. The sampling rate is usually 10Hz or below, which makes it difficult to capture high-frequency neural oscillation signals.

[0005] Fiber-free fNIRS significantly improves its ability to resist motion interference by replacing optical signals transmitted by optical fibers with electrical signals. The integrated design of the probe and electronic module can effectively reduce motion artifacts caused by limb movement or fiber drag, making it particularly suitable for children, patients with ADHD, or dynamic behavioral research (such as real-time monitoring of cerebral blood flow in sports rehabilitation). However, when used in conjunction with magnetoencephalography (MEG), it faces magnetic compatibility challenges: the electronic components integrated in fiber-free devices (such as light source driver modules and photoelectric conversion modules) may generate broadband electromagnetic noise, which can intrude into the extremely weak brain magnetic signals detected by MEG, especially interfering with the extraction of key signals such as neural oscillations in the 1-100 Hz frequency band. Traditional fiber-optic fNIRS, without active electronic components, is more likely to meet the stringent weak magnetic environment requirements of MEG.

[0006] Although MEG and fNIRS have complementary potential in terms of spatiotemporal characteristics and system configuration, there are still technical conflicts when combining the two for fiber-free near-infrared devices. The core goal of this invention is to break through the multi-channel near-infrared light source driving signal and MEG ultra-weak magnetic field measurement (10 -14To address spectral conflicts between signals of the order of magnitude (T), an electromagnetically compatible synchronous acquisition architecture is constructed. Traditional continuous-wave near-infrared imaging technology employs a time-division multiplexing (TD) driving light source strategy, essentially using square wave modulation. The light source undergoes high-frequency on-off switching, and transient current changes in the driving circuit are coupled to the OPM probe via electromagnetic induction. This switching noise, generated by the switching noise, falls within the OPM-MEG magnetic field measurement band (0.1-100 Hz). Neural signals are submerged in this noise, resulting in inability to measure EEG signals properly. As the number of fNIRS channels increases, the duty cycle limitations of the modulating square wave force the modulation frequency to exceed 50 kHz. This high-frequency switching noise further exacerbates low-frequency interference through Fourier transform harmonic components (e.g., the third harmonic of a 10 kHz square wave is 30 kHz, which is down-converted to the OPM measurement bandwidth after mixing in a nonlinear circuit). Continuously modulating the driving signal requires distinguishing between the numerous near-infrared channels based on the modulation frequency. In addition to high-frequency switching noise, in the MEG-NIR experiment, the MEG detector is close to the near-infrared light source, so the signal driving the light source is easily coupled into the measurement of the MEG signal, mixing with its internal signal, and generating interference frequencies that fall into the MEG measurement bandwidth. Therefore, the challenge lies in taking into account the performance of the signal output device and the impact of the frequency difference between different channels on the interference frequency in the MEG bandwidth. In addition, how to quickly and accurately demodulate and process the near-infrared signal under frequency modulation is also a difficult problem. Existing technologies use physical isolation methods to reduce electromagnetic interference, but sacrifice the synchronization of multimodal data. Low-frequency modulation technology reduces the modulation frequency of the fNIRS light source to avoid the MEG measurement bandwidth, but it results in a decrease in sampling rate and cannot meet the needs of high-dynamic brain activity detection.

[0007] Therefore, existing technologies fail to effectively balance the requirements of multi-channel, high sampling rate, low electromagnetic interference, and multimodal synchronization. This patent proposes a near-infrared high-sampling rate frequency modulation method compatible with brain magnetometry. By optimizing the light source driving strategy and signal demodulation algorithm, it increases the fNIRS sampling rate to over 100Hz. At the same time, it is compatible with brain magnetometry and achieves precise spatiotemporal alignment of cross-modal data. This technological breakthrough will promote the application of multimodal brain imaging in cutting-edge fields such as neural decoding and brain-computer interfaces. Summary of the Invention

[0008] In view of the above, the purpose of the present invention is to provide a magnetoencephalography-compatible multi-channel near-infrared high-sampling rate modulation method and system, which can reduce electromagnetic interference during the synchronous imaging process of MEG (magnetoencephalography) and fNIRS (functional near-infrared spectroscopy), utilize specific frequency modulation technology to achieve synchronous acquisition of magnetoencephalography signals and multi-channel high-sampling rate near-infrared signals, and accurately and quickly demodulate the near-infrared signals.

[0009] To achieve the above objectives, the technical solution of the present invention includes the following contents.

[0010] A multi-channel near-infrared high sampling rate modulation method compatible with electroencephalography, the method comprising:

[0011] Modulate the driving signal of the fNIRS light source to a high-frequency carrier;

[0012] A multi-channel frequency division multiplexing strategy is used to transmit the modulated driving signal to each fNIRS light source, so that each fNIRS light source emits infrared light to detect the target;

[0013] The detected infrared light is demodulated to obtain the amplitude of the detected infrared light.

[0014] Furthermore, the algorithm for modulating the driving signal of the fNIRS light source to a high-frequency carrier includes: a prime number frequency modulation algorithm or a frequency point modulation algorithm with a difference greater than a set value.

[0015] Furthermore, when the multi-channel frequency division multiplexing strategy is adopted, the frequency intervals of the modulated drive signals are calculated by the following steps:

[0016] Get the sampling rate of the modulated output device, the total number of channels in the multi-channel frequency division multiplexing strategy, and the cutoff frequency of the filter used for demodulation;

[0017] The frequency interval of each modulated driving signal is obtained according to the sampling rate of the modulation output device, the total number of channels of the multi-channel frequency division multiplexing strategy and the cutoff frequency of the filter used for demodulation.

[0018] Furthermore, demodulating the detected infrared light to obtain the amplitude of the detected infrared light includes:

[0019] The infrared light after detection Where t represents time, ω c is the target frequency, H c (t) is the amplitude of the infrared light after detection that changes with time, is the phase of the target signal, H(t) is the amplitude of the other frequency signal that varies with time, ω is the frequency of the other frequency signal, is the phase of other frequency signals, said other frequency signals including: low-frequency environmental noise and other received modulated drive signals;

[0020] The infrared light Y c Respectively with cos(ω c t) and sin(ω c t) After multiplication, use a low-pass filter to filter and get the filtering result And the filtering results

[0021] based on Get the amplitude H of the infrared light after detection c (t).

[0022] Furthermore, the model of the low-pass filter is selected by the following steps, including:

[0023] Calculate the minimum value of the difference between the frequencies used for modulation Δω=|ω-ω c |;

[0024] The model of the low-pass filter is selected based on the minimum value Δω.

[0025] Furthermore, a GPU is used to complete the demodulation of the detected infrared light.

[0026] Furthermore, the method further comprises:

[0027] The brain magnetic signals are collected based on optically pumped magnetoencephalography or SQUID magnetoencephalography.

[0028] A magnetoencephalogram-compatible multi-channel near-infrared high-sampling rate modulation system, comprising:

[0029] A high-frequency modulation module is used to modulate the driving signal of the fNIRS light source into a high-frequency carrier using prime frequency modulation;

[0030] A multi-channel carrier generation module is used to transmit the modulated driving signal to each fNIRS light source using a multi-channel frequency division multiplexing strategy, so that each fNIRS light source emits infrared light for detecting the target;

[0031] The signal demodulation module is used to demodulate the detected infrared light to obtain the amplitude of the detected infrared light.

[0032] An electronic device comprises: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, it implements any of the above-mentioned magnetoencephalogram-compatible multi-channel near-infrared high sampling rate modulation methods.

[0033] A computer-readable storage medium having computer program instructions stored thereon, wherein the computer program instructions, when executed by a processor, implement any of the above-mentioned magnetoencephalogram-compatible multi-channel near-infrared high sampling rate modulation methods.

[0034] A computer program product, when running on a computer device, enables the computer device to execute any one of the above-mentioned magnetoencephalogram-compatible multi-channel near-infrared high sampling rate modulation methods.

[0035] Compared with the prior art, the present invention has at least the following beneficial effects.

[0036] 1. High sampling rate and high channel count compatibility: Through high-frequency modulation technology and frequency division multiplexing strategies with prime frequency points, the synchronous driving of multi-channel near-infrared light sources is achieved, the sampling rate is increased to over 100Hz, and high-density detection of 128 channels or less is supported, meeting the needs of high-dynamic brain activity detection. By controlling the frequency spacing, the performance requirements of the hardware output device are reduced.

[0037] 2. Low electromagnetic interference design: High-frequency prime frequency modulation technology is used to greatly weaken the noise of the light source driving signal within the MEG measurement bandwidth (0.1-100Hz), significantly reducing electromagnetic interference and ensuring the measurement accuracy of brain magnetic signals.

[0038] 3. Synchronous acquisition of multimodal data: High-frequency clock synchronous acquisition can achieve high-precision time alignment of near-infrared and brain magnetic signals, providing a reliable basis for the joint analysis of multimodal data.

[0039] 4. Flexibility and scalability: Modulation and demodulation schemes can be flexibly adjusted according to experimental requirements. For example, by combining GPU acceleration optimization with phase-preserving coherent demodulation technology, the system is suitable for brain science research and clinical application scenarios of different scales, and provides the possibility of adding other imaging modalities and experimental devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Flowchart of the multi-channel near-infrared high sampling rate modulation method for brain magnetometry compatibility.

[0041] Figure 2 This is the flow chart of near-infrared signal demodulation.

[0042] Figure 3 A scheme for modulating the frequency of 32 dual-wavelength light sources. DETAILED DESCRIPTION

[0043] In the following description, the modulation method of the present invention is further described through specific embodiments to facilitate a more thorough understanding of the features and advantages of the present invention by those skilled in the art. It should be noted that the description below is merely a representative typical application. Obviously, the present invention is not limited to any specific structure, function, device, and method described herein, and may have other embodiments or combinations of other embodiments. The software / hardware modules described in the present invention or shown in the accompanying drawings may also be flexibly adjusted as needed.

[0044] The present invention realizes the compatibility of fNIRS and optically pumped magnetoencephalography through prime number frequency modulation, multi-channel frequency division multiplexing strategy and a specific demodulation algorithm, while achieving a multi-channel functional near-infrared spectroscopy light intensity modulation and fast demodulation method with high sampling rate and high signal-to-noise ratio.

[0045] Specifically, the brain magnetocompatible multi-channel near-infrared high sampling rate modulation method of the present invention is as follows: Figure 1 As shown, the following steps 1 to 3 are included.

[0046] Step 1: Modulate the driving signal of the fNIRS light source to a high-frequency carrier.

[0047] The present invention modulates the driving signal of the fNIRS light source to a high-frequency carrier (eg, 2-100 kHz), so that the circuit noise energy is concentrated outside the MEG measurement bandwidth (0.1-100 Hz).

[0048] In one embodiment, the present invention can utilize prime frequencies or a frequency modulation algorithm with large interpolation values ​​for modulation. The proximity of magnetoencephalography and near-infrared signals results in signal coupling and frequency mixing, and frequency differences can fall within the MEG measurement bandwidth, causing interference. Using prime frequencies allows the differences between modulation frequencies to be controlled to different values, distributing all modulation frequency differences to different values ​​within the MEG measurement bandwidth. This weakens any interfering frequency points within the measurement bandwidth, reducing the noise caused by frequency differences within the measurement bandwidth.

[0049] Step 2: Using a multi-channel frequency division multiplexing strategy, the modulated driving signal is transmitted to each fNIRS light source, so that each fNIRS light source emits infrared light to detect the target.

[0050] The present invention adopts a multi-channel frequency division multiplexing strategy, assigning an independent modulation frequency to each near-infrared light source. Different detection channels are distinguished by frequency coding, avoiding the sampling rate loss and electromagnetic interference caused by traditional time division multiplexing. This technology can increase the fNIRS sampling rate to over 100 Hz while reducing crosstalk between adjacent channels.

[0051] In one embodiment, the present invention reduces the temporal resolution requirements of the output device by controlling the frequency spacing, alleviating the pressure on filter performance during demodulation while supporting the driving of more channels of light sources. For example, when using a device with a 100kHz sampling rate for modulated signal output and a 4th-order Butterworth filter with a cutoff frequency of 200Hz for demodulation, the minimum modulation frequency spacing used by the two light source channels is greater than or equal to 450Hz, which can meet the high sampling rate requirement of 100Hz-200Hz for near-infrared data from 32 light sources.

[0052] Step 3: Demodulate the detected infrared light to obtain the amplitude of the detected infrared light.

[0053] The present invention uses a coherent demodulation method for demodulation. However, unlike traditional coherent demodulation, the present invention combines phase-preserving filtering technology to extract the light intensity information of the target frequency band during the signal demodulation stage, perform rapid demodulation, and effectively suppress the aliasing effect of environmental noise (such as power frequency interference and motion artifacts). The steps for demodulating the near-infrared signal are as follows: Figure 2 As shown. The collected near-infrared signal can be expressed as

[0054]

[0055] Where t represents time, ω c is the target frequency, H c (t) is the amplitude obtained by demodulation, representing the light intensity changing with time, is the phase of the target signal, H(t),ω, Represent the amplitude, frequency and phase of other frequency signals respectively, Contains low-frequency ambient noise and other modulated frequency signals that may be received.

[0056] First, the present invention makes Equation 1 and cos(ω c t) and combined with the product-sum-difference formula of trigonometric functions, we get

[0057]

[0058] Let formula 1 and sin(ω c t) and combined with the product-sum-difference formula of trigonometric functions, we get

[0059]

[0060] Next, perform low-pass filtering on Equations (2) and (3). The parameters of the low-pass filter are determined by the signal component with the lowest frequency value in Equations (2) and (3). The cutoff frequency should be less than this frequency value. The smaller this frequency value is, the larger the order should be to ensure the filtering effect. When using high-frequency modulation, the lowest frequency in Equations (2) and (3) should be |ω-ω c |, which is the minimum value of the difference between the frequency points used for high-frequency modulation. If the lowest frequency is 500Hz, a Butterworth filter with a cutoff frequency of 200Hz and an order of 4 can ensure the filtering effect. After filtering, equations (4) and (5) are obtained, and A and B are used to represent the obtained results, respectively, to illustrate the next step of the calculation. This step can filter out the ω contained in equation (1), that is, the noise and other modulation frequencies, effectively suppressing environmental noise while extracting the target frequency.

[0061]

[0062] It should be noted that traditional coherent demodulation does not consider the phase of the demodulated signal. and In this step, equation (5) becomes That is, the amplitude of the target frequency signal is considered to be obtained. However, in actual signal acquisition, due to the performance of the signal output device and the circuit components passed through, the received signals of different channels often have a certain phase difference, and the obtained B will not be the accurate target signal amplitude. Therefore, it is necessary to consider and The impact on demodulation is calculated in the next step.

[0063] Finally, by calculating The formula (5) contains The part of is eliminated, and then the accurate amplitude H corresponding to the target frequency in the signal is obtained c (t), downsampling and smoothing filtering are performed according to the target sampling rate, and the result is used as the light intensity for near-infrared signal preprocessing;

[0064] In an optimization solution of the present invention, GPU acceleration can be applied to the above steps to optimize demodulation, thereby achieving real-time demodulation and improving experimental efficiency.

[0065] The modulation and demodulation methods described above enable both frequency-divided driving of near-infrared light sources and rapid analysis of multimodal data. In practice, the implementation of each module can be flexibly adjusted based on actual conditions and needs.

[0066] In addition, the electroencephalographic signal processing part of the present invention can use OPM-MEG (Optically Pumped Magnetometer Electroencephalography) and is also applicable to traditional SQUID electroencephalography.

[0067] The present invention is described below by taking the selection of the modulation frequency of a near-infrared light source actually used in a brain magneto-near-infrared imaging experiment.

[0068] The number of modulation frequencies is determined according to the number of light sources. The number of modulation frequencies is the product of the number of light sources and the number of wavelengths used by a single light source. For example, if 32 dual-wavelength light sources are used, 64 modulation frequencies are required. Prime modulation frequencies are allocated according to the relative positions of the light source and the detector. The detector distinguishes light sources in different positions based on the modulation frequency, that is, all light sources that a detector can receive cannot use the same modulation frequency. The prime frequency of 2687Hz is selected as the starting point. In order to achieve a high sampling rate of 200Hz for the demodulated signal, an interval of 450Hz or more is selected as the prime frequency selection standard. Based on the maximum sampling rate of 100kHz of the hardware used to collect the modulation signal, the modulation frequency is controlled to less than one-quarter of the maximum sampling rate to prevent frequency aliasing. It can eventually be arranged to 24967Hz. The specific frequency is as follows: Figure 3As shown, there are 50 prime frequencies, and the total number of selected modulation frequencies is less than 64. To distinguish all light sources that may be received by the same detector, seven light sources are further selected for frequency multiplexing based on the relative position and distance between the light sources and detectors. The distance between each of the detectors that can receive the seven light sources is controlled within 9 cm, thus completing the arrangement of the modulation frequencies.

[0069] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Any skilled artisan may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be as set forth in the appended claims.

Claims

1. A multi-channel near-infrared high sampling rate modulation method compatible with brain magnetoelasticity, characterized in that: The method comprises: Modulate the driving signal of the fNIRS light source to a high-frequency carrier; A multi-channel frequency division multiplexing strategy is used to transmit the modulated driving signal to each fNIRS light source, so that each fNIRS light source emits infrared light to detect the target; The detected infrared light is demodulated to obtain the amplitude of the detected infrared light.

2. The method according to claim 1, characterized in that Algorithms for modulating the driving signal of the fNIRS light source to a high-frequency carrier include: a prime number frequency modulation algorithm or a frequency point modulation algorithm with a difference greater than a set value.

3. The method according to claim 1, characterized in that When the multi-channel frequency division multiplexing strategy is adopted, the frequency interval of each modulated driving signal is calculated by the following steps: Get the sampling rate of the modulated output device, the total number of channels in the multi-channel frequency division multiplexing strategy, and the cutoff frequency of the filter used for demodulation; The frequency interval of each modulated driving signal is obtained according to the sampling rate of the modulation output device, the total number of channels of the multi-channel frequency division multiplexing strategy and the cutoff frequency of the filter used for demodulation.

4. The method according to claim 1, wherein Demodulating the detected infrared light to obtain the amplitude of the detected infrared light includes: The infrared light after detection Where t represents time, ω c is the target frequency, H c (t) is the amplitude of the infrared light after detection that changes with time, is the phase of the target signal, H(t) is the amplitude of the other frequency signal that varies with time, ω is the frequency of the other frequency signal, is the phase of other frequency signals, said other frequency signals including: low-frequency environmental noise and other received modulated drive signals; The infrared light Y c Respectively with cos(ω c t) and sin(ω c t) After multiplication, use a low-pass filter to filter and get the filtering result And the filtering results based on Get the amplitude H of the infrared light after detection c (t).

5. The method according to claim 4, characterized in that The model of the low-pass filter is selected by the following steps, including: Calculate the minimum value of the difference between the frequencies used for modulation Δω=|ω-ω c |; The model of the low-pass filter is selected based on the minimum value Δω.

6. The method according to claim 1, characterized in that A GPU is used to complete the demodulation of the detected infrared light.

7. The method according to claim 1, characterized in that The method further comprises: The brain magnetic signals are collected based on optically pumped magnetoencephalography or SQUID magnetoencephalography.

8. A multi-channel near-infrared high sampling rate modulation system compatible with brain magnetoelasticity, characterized in that: The system comprises: A high-frequency modulation module is used to modulate the driving signal of the fNIRS light source into a high-frequency carrier; A multi-channel carrier generation module is used to transmit the modulated driving signal to each fNIRS light source using a multi-channel frequency division multiplexing strategy, so that each fNIRS light source emits infrared light for detecting the target; The signal demodulation module is used to demodulate the detected infrared light to obtain the amplitude of the detected infrared light.

9. An electronic device, characterized in that: The electronic device includes: a processor and a memory storing computer program instructions; when the processor executes the computer program instructions, it implements the brain magnetocompatible multi-channel near-infrared high sampling rate modulation method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the magnetoencephalometric-compatible multi-channel near-infrared high-sampling rate modulation method according to any one of claims 1 to 7.

Images