A dual-path multi-mode low-power consumption electroencephalogram signal acquisition system

Abstract

This invention discloses a dual-channel, multi-mode, low-power EEG signal acquisition system, relating to the field of brain-computer interface technology. It includes: N electrode channels for acquiring EEG signals; N time-division multiplexing switches, whose inputs are respectively connected to the N electrode channels, and whose outputs are connected in parallel to a shared analog front-end processing link; and N peak detectors, whose inputs are respectively connected to the N electrode channels, for real-time detection of whether the amplitude of the EEG signal in the corresponding channel exceeds a preset threshold. This invention addresses the time-frequency characteristics differences between LFP and AP signals by employing time-division multiplexing and event-driven acquisition mechanisms: for slowly varying continuous LFP signals, the analog front-end and analog-to-digital converter resources are shared; for sparse, bursty AP signals, high-speed sampling is initiated only when a discharge event is detected. Compared to the traditional full-channel high-speed continuous sampling architecture, this significantly reduces the average operating bandwidth and dynamic power consumption of the analog-to-digital converter, improving the system's energy efficiency ratio.

Description

Technical Field

[0001] This invention relates to the field of brain-computer interface technology, specifically to a dual-channel, multi-mode, low-power EEG signal acquisition system. Background Technology

[0002] The brain is one of the most complex biological systems known. High-bandwidth, high-precision brain-computer interface (BCI) chip systems can record and decode neural activity in real time, providing an important technical means for understanding higher brain functions such as consciousness, memory, and decision-making. At the same time, many mental and neurological diseases, such as depression, Parkinson's disease, and epilepsy, originate from abnormal neural circuit function. Chip-level neural signal recording helps to accurately locate pathological neural activity, providing a key basis for the development of new diagnostic and treatment methods.

[0003] However, in the process of recording neural signals at the chip level, the system generally faces the trade-off between bandwidth, size and power consumption. In order to achieve fine analysis of neural activity, the system needs to collect more neuronal information, which usually depends on increasing the number of electrode channels and increasing the sampling bandwidth. However, the resulting increase in data volume and power consumption seriously restricts the scalability and long-term working capability of the system. At the same time, each additional recording channel usually requires the configuration of an independent analog front-end amplifier, filter and analog-to-digital converter, so that the chip area and power consumption increase approximately linearly with the number of channels.

[0004] Multi-channel time-division multiplexing architectures can effectively reduce the number of analog front-end amplifiers, filters, and analog-to-digital converters by sharing high-bandwidth signal processing links, thereby reducing chip area. However, multi-channel time-division multiplexing places higher demands on the instantaneous bandwidth of the analog-to-digital converters, and high-bandwidth analog-to-digital converters are often accompanied by significant power consumption, limiting the applicability of this solution in low-power applications.

[0005] From the perspective of signal characteristics, EEG signals mainly include two types: local field potential (LFP) signals formed by the superposition of the activities of multiple neuronal groups, and action potential (AP) signals generated by the discharge of a single neuron. LFP signals change slowly and have strong continuity, with a frequency range of 1–300 Hz, making them suitable for acquisition using multi-channel time-division multiplexing to reduce the number, area, and power consumption of analog front-ends and analog-to-digital converters. In contrast, AP signals are characterized by high frequency, sparseness, and strong burstiness, with their spectrum mainly concentrated in the range of 300 Hz–10 kHz. If AP signals are also acquired using multi-channel time-division multiplexing, it will significantly increase the bandwidth requirements of the analog-to-digital converter, leading to a substantial increase in power consumption. At the same time, no effective AP events occur for most of the time, and the acquired data contains a large amount of redundant information, further increasing the burden on storage and data transmission. Summary of the Invention

[0006] The purpose of this invention is to provide a dual-channel, multi-mode, low-power EEG signal acquisition system to solve the problems in the prior art where processing LFP signals consumes a lot of hardware resources, while processing AP signals results in a lot of redundant signals and the signal quality needs to be improved.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a dual-channel, multi-mode, low-power EEG signal acquisition system, comprising:

[0008] N electrode channels are used to collect electroencephalogram (EEG) signals;

[0009] N time-division multiplexed switches, whose input terminals are respectively connected to the N electrode channels, and whose output terminals are connected in parallel to a common analog front-end processing link;

[0010] N peak detectors, whose input terminals are respectively connected to the N electrode channels, are used to detect in real time whether the amplitude of the EEG signal of the corresponding channel exceeds a preset threshold, and output the event trigger signal of the corresponding channel when the threshold is exceeded;

[0011] The analog front-end processing link has its input end connected to the output end of the N time-division multiplexing switches, and is used to amplify and convert the selected signal into an analog-to-digital signal;

[0012] The channel selection control module is connected to the output terminal of the peak detector and the control terminal of the time-division multiplexing switch, respectively. It is used to generate the switch control timing sequence according to the polling channel sequence and / or event triggering sequence, so as to control the selection of the N time-division multiplexing switches.

[0013] The channel selection control module includes:

[0014] The polling control module is used to generate the polling channel sequence;

[0015] The event trigger control module is used to generate the event trigger sequence according to the event trigger signals output by each peak detector;

[0016] The selection arbitration module is used to arbitrate the polling channel sequence and the event trigger sequence to generate the switch control timing, ensuring that only one electrode channel is selected at the same time;

[0017] According to different acquisition requirements, the system can be configured as a time-division multiplexing acquisition mode, an event trigger acquisition mode or a hybrid acquisition mode.

[0018] The specific scheme of the time-division multiplexing acquisition mode:

[0019] After the analog signals collected by N electroencephalogram signal acquisition electrodes enter the system, they are divided into two parallel paths. Among them, one path is connected to N peak detectors, and the trigger threshold of the peak detectors is configured to the non-responsive state, so that no peak detection pulse is generated in this mode, and the event trigger sequence always remains invalid; the other path is connected to N time-division multiplexing switches.

[0020] In this mode, the system does not generate a valid event trigger sequence, and the switch control timing is exactly the same as the preset polling switch sequence. Under the control of the switch control timing, the time-division multiplexing switch sequentially conducts each electrode channel according to the established channel order, realizing the cyclic sampling of multi-channel signals. The selected signals are sequentially processed by a low-noise amplifier, a variable gain amplifier, an analog-to-digital converter and a digital logic module and then output for the acquisition of continuous electroencephalogram signals.

[0021] Furthermore, the time-division multiplexing acquisition mode includes a high-speed acquisition sub-mode and a low-speed acquisition sub-mode:

[0022] In the high-speed acquisition sub-mode, the polling channel sequence is M preset channels, where M < N. The switch control timing controls the time-division multiplexing switch to perform high-speed cyclic switching among the M channels, realizing high-time-resolution sampling of the selected channels;

[0023] In the low-speed acquisition sub-mode, the polling channel sequence is all N channels. The switch control timing controls the time-division multiplexing switch to perform cyclic switching among all N channels, and an idle period is inserted between the selections of adjacent channels, effectively reducing the equivalent sampling rate of a single channel to meet the acquisition requirements of low-frequency signals and reduce power consumption and data volume.

[0024] The specific scheme of the event trigger acquisition mode:

[0025] The analog signals acquired by N EEG signal acquisition electrodes are divided into two parallel paths after entering the system. One path is connected to N peak detectors, which are set to an effective trigger threshold. When the signal amplitude of a certain channel exceeds the set threshold, a corresponding peak detection pulse is generated, and an event trigger sequence is generated. The other path is connected to N time-division multiplexing switches.

[0026] In this mode, the polling channel sequence of the system is in an idle state, and the switching control timing is completely driven by the event triggering sequence. When any channel generates an event triggering signal, the switching control timing turns on the channel and performs multiple high-speed samplings on the channel at a fixed sampling interval within a preset event sampling window. The selected signal is processed by a low-noise amplifier, a variable gain amplifier, an analog-to-digital converter, and a digital logic module before being output, thereby realizing event-driven acquisition of transient discharge events of neurons.

[0027] Furthermore, the length of the event sampling window and the sampling interval are preset values; when multiple event trigger signals are generated simultaneously, the selection arbitration module arbitrates each event trigger signal to determine the selection order of each event channel; when the number of channels currently sampling events reaches the system's preset upper limit, newly generated event trigger signals will be ignored.

[0028] Specific scheme for hybrid sampling mode:

[0029] The analog signals acquired by N EEG signal acquisition electrodes are divided into two parallel pathways after entering the system. One pathway is connected to N peak detectors, which are set with effective trigger thresholds to detect neuronal firing events in each channel in real time and generate corresponding event trigger sequences. The other pathway is connected to N time-division multiplexing switches to perform regular channel polling sampling.

[0030] In this mode, the channel selection control module simultaneously enables the polling channel sequence and the event triggering sequence; the selection arbitration module is used to dynamically insert the event triggering sequence into the idle period of the polling channel sequence to generate the switch control timing.

[0031] When no event is triggered, the switch control timing performs normal selection sampling according to the polling channel sequence; when an event is detected, the selection arbitration module allocates the event trigger selection request of the corresponding channel to the nearest idle period to turn on the channel and perform high-speed sampling at a fixed sampling interval within the preset event sampling window, thereby ensuring continuous signal acquisition while achieving accurate capture of key neural events.

[0032] Furthermore, the sampling interval for the event channel within the event sampling window is greater than the selection interval of adjacent channels in the polling channel sequence; in the hybrid acquisition mode, the system alternately selects multiple event channels in a time-sharing manner, and schedules the selection arbitration module between event sampling and polling sampling to ensure that only one channel is selected at any given time.

[0033] Specific solutions for the signal processing link:

[0034] The analog front-end processing link includes a low-noise amplifier, a variable gain amplifier, and an analog-to-digital converter connected in sequence. The selected electrode channel signal is amplified by the low-noise amplifier, the gain of the variable gain amplifier is adjusted, and then the analog-to-digital converter converts it into a digital signal.

[0035] The system also includes a digital logic module connected to the output of the analog-to-digital converter, used to receive and process the digital signals, and to perform time synchronization, data fusion and subsequent digital processing on the converted digital signals.

[0036] Signal types and applications:

[0037] The EEG signals include continuous local field potential (LFP) signals and sparse action potential (AP) signals. The polling channel sequence is used to acquire the continuous LFP signals, and the event triggering sequence is used to acquire the sparse AP signals in an event-driven manner.

[0038] Compared with existing technologies, the dual-channel multi-mode low-power EEG signal acquisition system provided by this invention has the following beneficial effects:

[0039] 1. Significantly reduces overall system power consumption

[0040] This invention addresses the time-frequency characteristics differences between LFP and AP signals by employing time-division multiplexing and event-driven acquisition mechanisms respectively: for slowly changing continuous LFP signals, analog front-end and analog-to-digital converter resources are shared; for sparse and bursty AP signals, high-speed sampling is initiated only when a discharge event is detected. Compared with the traditional full-channel high-speed continuous sampling architecture, this significantly reduces the average operating bandwidth and dynamic power consumption of the analog-to-digital converter, thereby improving the system's energy efficiency ratio.

[0041] 2. Improve bandwidth utilization and reduce redundant data.

[0042] This invention monitors neural discharge events in real time using a peak detector, allocating sampling resources only when valid events occur. This significantly reduces the storage and transmission bandwidth occupied by invalid data, increases the density of effective data information, and alleviates the burden on backend processing and wireless transmission.

[0043] 3. Enhance system scalability and channel expansion capabilities.

[0044] This invention employs a dual-path architecture combined with time-division multiplexing technology, which can expand the number of electrode channels without significantly increasing the number of analog front-ends. Compared with the traditional structure of "independent analog front-end and analog-to-digital converter per channel", the chip area and power consumption no longer increase linearly with the number of channels, which is conducive to high-channel-count integration and large-scale expansion, and is suitable for large-scale neural signal recording scenarios.

[0045] 4. Supports multiple working modes, offering high application flexibility.

[0046] This invention supports three configurable working modes: time-division multiplexing acquisition, event-triggered acquisition, and hybrid sampling. These modes can be flexibly switched according to different application requirements: the time-division multiplexing mode is used for continuous state monitoring scenarios; the event-triggered mode is used for neural discharge research or regulation scenarios; and the hybrid mode is used for scenarios that require simultaneous acquisition of slowly varying background signals and transient discharge information. The configurability of the modes enhances its applicability in scientific research and clinical applications.

[0047] 5. High-quality acquisition of both continuous and transient signals.

[0048] In the hybrid sampling mode, this invention uses an arbitration and timing fusion mechanism to dynamically insert the event trigger sequence into the polling sequence, enabling the coordinated work of continuous signal acquisition and key event capture. While ensuring the integrity of continuous signals such as LFP, it provides higher time resolution sampling capability for transient events such as AP, thus balancing signal integrity and time accuracy.

[0049] 6. Suitable for long-term implantable low-power applications.

[0050] By reducing average power consumption and data transmission burden, this invention helps to extend the battery life of implantable brain-computer interface systems, reduce the risk of overheating and tissue damage, improve the long-term stable operation capability of the system, and provide an efficient solution for the miniaturization and low-power design of implantable neural interface chips. Attached Figure Description

[0051] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0052] Figure 1 This is a system architecture block diagram of the present invention;

[0053] Figure 2 This is a timing diagram of the high-speed acquisition mode under the time-division multiplexing acquisition mode of the present invention;

[0054] Figure 3This is a timing diagram of the low-speed acquisition mode under the time-division multiplexing acquisition mode of the present invention;

[0055] Figure 4 This is a timing diagram of the event-triggered acquisition mode of the present invention;

[0056] Figure 5 This is a timing diagram of the hybrid sampling mode of the present invention. Detailed Implementation

[0057] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0058] like Figure 1 As shown, this embodiment provides a 64-channel EEG signal acquisition system. The system includes a channel selection execution module and a channel selection control module, wherein the channel selection control module includes an event trigger control module, a polling control module, and a selection arbitration module, used to generate channel selection control information based on neural signal characteristics and acquisition strategies.

[0059] After the 64 electrode signals enter the system, they are divided into two processing paths: the first path is connected to 64 time-division multiplexers, whose channel conduction status is controlled by the channel selection control signal, and the outputs of the 64 time-division multiplexers are connected in parallel to a shared analog front-end processing link; the second path is connected to the corresponding event detection module (peak detector), which is used to detect the neural signals of each channel in real time. When the amplitude of a certain channel signal exceeds a preset threshold, the event detection module outputs the event trigger signal of the corresponding channel.

[0060] Channel selection control mechanism: To meet different acquisition requirements, the channel selection control module can selectively enable event-triggered sequences, polling switch sequences, or both simultaneously. The event-triggered sequence and the polling control sequence generated by the digital control module are input to the selection arbitration module. The selection arbitration module arbitrates the channel selection requests output by the enabled switch sequences to generate corresponding switch control sequences, ensuring that only one channel's signal is sampled at a time and avoiding aliasing of signals from different channels.

[0061] The channel selection execution module controls the conduction state of the corresponding time-division multiplexing switch according to the switch control sequence. The selected channel signal is processed sequentially by a low-noise amplifier, a variable gain amplifier, and an analog-to-digital converter, and then sent to the digital logic module for further processing and output. The digital logic module is connected to the output of the analog-to-digital converter to receive the converted digital signal, perform timestamp alignment and data fusion on the LFP and AP signals acquired from different channels to ensure the accuracy of subsequent joint analysis, and perform subsequent digital processing such as data packaging and compression.

[0062] Depending on the different sequence activation methods, the system can implement time-division multiplexing acquisition, event-triggered acquisition, or a hybrid sampling mode that combines both.

[0063] Specific implementation methods of time-division multiplexing acquisition mode

[0064] In time-division multiplexing acquisition mode, the system can configure different channel polling strategies according to the target signal type and acquisition requirements, and can make trade-offs between time resolution, number of channels and data rate. Specifically, time-division multiplexing acquisition mode can be further divided into high-speed cyclic sampling mode with a fixed number of channels and low-speed cyclic sampling mode for all channels.

[0065] High-speed acquisition sub-mode: Figure 2 This is a high-speed sampling timing diagram in time-division multiplexing acquisition mode. In this embodiment, the sampling rate of the analog-to-digital converter is set to 128kHz, and the polling control module is configured with six polling channels: channels 1, 8, 15, 23, 31, and 33. At this time, the threshold of the peak detector is configured to be in a non-responsive state, the event trigger sequence is invalid, and it remains at 0. The selection arbitration module generates the switch control timing only based on the polling channel sequence. The switch control timing is completely consistent with the polling channel sequence. Channels 1, 8, 15, 23, 31, and 33 are sequentially selected for cyclic sampling, with a sampling rate of 128kHz / 6 ≈ 21.33kHz for each channel. This sampling rate can cover the frequency range of local field potential (LFP) signals and action potential (AP) signals.

[0066] Low-speed acquisition sub-mode: Figure 3 This is a timing diagram for low-speed sampling in time-division multiplexing acquisition mode. In this embodiment, the sampling rate of the analog-to-digital converter remains at 128kHz, and the polling channel sequence is set to select one channel every two clock cycles among all 64 channels. Similarly, the event trigger sequence is not enabled and is fixed at 0, and the switch control timing is consistent with the polling channel sequence.

[0067] The 64 channels are selected sequentially at two-cycle intervals, meaning each channel is selected once in a complete polling cycle, with two idle cycles inserted between the selection of adjacent channels. The equivalent sampling rate for each channel is 128kHz / (64×3)≈666.67Hz. This sampling rate is sufficient for acquiring low-frequency signals of local field potentials (LFPs), but it cannot cover action potential (AP) signals.

[0068] Detailed Implementation of Event-Triggered Data Acquisition Mode

[0069] Figure 4This is a timing diagram for the event-triggered acquisition mode. In this mode, the analog-to-digital converter's sampling rate is set to 128kHz, the polling control module is disabled, and the polling channel sequence is idle. The system only enables the event-triggered sequence, meaning sampling is only initiated when a valid neuronal firing event is detected.

[0070] The threshold of the peak detector is set to a predetermined value (e.g., 50 μV based on the background noise level). When the signal amplitude of a certain channel exceeds this threshold, the event trigger control module generates an event trigger sequence for the corresponding channel and submits this signal as a channel selection request to the selection arbitration module. After the channel is selected, the system performs high-speed sampling on that channel at a fixed sampling interval within a preset event sampling window.

[0071] In this embodiment, the event sampling window length is 24 sampling points, and the sampling interval is 5 clock cycles. The system allows simultaneous alternating sampling of up to 6 event channels, with an equivalent sampling rate of 128kHz / 6≈21.33kHz for each channel, which can cover the frequency range of action potential (AP) signals.

[0072] Specific timing examples

[0073] In the 5th clock cycle, the peak detector detects that the signal of the 6th channel exceeds the threshold, and the corresponding event trigger signal is sent to the selection arbitration module. In the 6th clock cycle, the selection arbitration module turns on the time-division multiplexing switch of the 6th channel, and the channel 6 signal then enters the analog front-end processing link for amplification and analog-to-digital conversion, and is sampled every 5 cycles (cycles 11, 16, 21...), with a sampling duration of 24 points (corresponding to 1.125ms).

[0074] In the 14th cycle, the peak detector detects that channels 9 and 32 exceed the set threshold and generates the corresponding event trigger sequence. In the 15th clock cycle, after the arbitration module arbitrates the event trigger sequence, the time-division multiplexing switch of channel 9 is turned on first, and channel 32 is turned on and sampled in the subsequent 16th cycle.

[0075] In the 20th cycle, the peak detector detected that channel 17 exceeded the threshold. However, since the number of channels currently sampling events had reached the system's preset upper limit (6), the arbitration module ignored the sampling request of channel 17 and did not sample it.

[0076] Event-triggered acquisition mode can effectively reduce sampling processing when there is no AP signal, reduce system power consumption and data redundancy, and accurately capture neuronal firing events.

[0077] Detailed implementation of hybrid sampling mode

[0078] Figure 5This is a timing diagram for the hybrid sampling mode. In this mode, the system simultaneously enables the polling control module and the event-triggered control module to achieve continuous acquisition of LFP signals and event-driven acquisition of AP signals.

[0079] LFP signal acquisition: For local field potential (LFP) signals, the system performs low-sampling-rate cyclic sampling on all 64 channels according to a preset polling sequence. Each channel is selected sequentially at two clock cycles intervals, corresponding to an equivalent sampling rate of 128kHz / (64×3)≈666.67Hz, which ensures that the low-frequency LFP signal is completely captured.

[0080] AP signal acquisition: For action potential (AP) signals, the system enables an event-triggered mechanism. When the peak detector detects that the channel signal exceeds a preset threshold, the arbitration module allocates it to the nearest idle period for high-speed sampling. In this embodiment, the event sampling window length is 24 sampling points, the sampling interval is 5 clock cycles, and the system allows up to 4 event channels to be sampled simultaneously in a time-division alternating manner during one event sampling process.

[0081] Specific timing examples

[0082] The threshold of the peak detector is set to an appropriate value (e.g., 50 μV).

[0083] In the 5th clock cycle, channel 6 is detected to have exceeded the set threshold, generating an event trigger sequence for channel 6. The arbitration module detects that the 6th clock cycle is an idle cycle, and therefore activates the time-division multiplexing switch for channel 6 in the 6th cycle. The channel 6 signal then enters the analog front-end processing link, and is sampled every 5 cycles (cycles 11, 16, 21...), with a sampling duration of 24 points (corresponding to 1.125ms).

[0084] In the 15th clock cycle, channel 32 is detected to have exceeded the set threshold, generating an event trigger sequence for channel 32. At this time, the 16th clock cycle is occupied by the polling channel sequence, so the arbitration module is selected to prioritize the polling channel. The 17th clock cycle is an idle cycle, so channel 32 is turned on in the 17th clock cycle, and sampling is performed every 5 cycles (cycles 22, 27, 32...), with a sampling duration of 24 points (corresponding to 1.125ms).

[0085] In mixed sampling mode: for LFP signals, all channels are cyclically sampled at a frequency of approximately 666.67 Hz to ensure that low-frequency LFP signals are captured; for AP signals, event-triggered channels are sampled at a high-speed frequency of approximately 21.33 kHz to cover the frequency range of AP signals.

[0086] The hybrid acquisition mode takes into account the synchronous acquisition requirements of LFP and AP signals. While ensuring continuous signal acquisition, it achieves accurate acquisition of key neural events, effectively increases the number of channels in multi-channel time-division multiplexing, and reduces the sampling rate of invalid data.

[0087] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A dual-channel, multi-mode, low-power EEG signal acquisition system, characterized in that, Comprising: N electrode channels for collecting electroencephalogram (EEG) signals; N time-division multiplexing switches, whose input ends are respectively connected to the N electrode channels, and whose output ends are connected in parallel to a common analog front-end processing link; N peak detectors, whose input ends are respectively connected to the N electrode channels, for detecting in real time whether the amplitude of the EEG signal of the corresponding channel exceeds a preset threshold, and outputting an event trigger signal for the corresponding channel when the threshold is exceeded; An analog front-end processing link, whose input end is connected to the output ends of the N time-division multiplexing switches, for amplifying and performing analog-to-digital conversion on the selected signal; A channel selection control module, respectively connected to the output ends of the peak detectors and the control ends of the time-division multiplexing switches, for generating a switch control timing sequence according to a polling channel sequence and / or an event trigger sequence to control the gating of the N time-division multiplexing switches; Wherein, the channel selection control module includes: A polling control module for generating the polling channel sequence; An event trigger control module for generating the event trigger sequence according to the event trigger signals output by each peak detector; A selection arbitration module for arbitrating the polling channel sequence and the event trigger sequence to generate the switch control timing sequence, ensuring that only one electrode channel is gated at the same time; The system can be configured into a time-division multiplexing acquisition mode, an event trigger acquisition mode or a hybrid acquisition mode according to different acquisition requirements.

2. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 1, characterized in that, When the system is configured into the time-division multiplexing acquisition mode, the threshold of the peak detector is configured to be in a non-responsive state, and the event trigger sequence is invalid; The selection arbitration module only generates the switch control timing sequence according to the polling channel sequence; The time-division multiplexing switches are sequentially cyclically gated in a preset channel order under the control of the switch control timing sequence to realize continuous sampling of multi-channel EEG signals.

3. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 2, characterized in that, The time-division multiplexing acquisition mode includes a high-speed acquisition sub-mode and a low-speed acquisition sub-mode; In the high-speed acquisition sub-mode, the polling channel sequence is M preset channels, where M < N, and the switch control timing sequence controls the time-division multiplexing switches to rapidly cycle and switch between the M channels; In the low-speed acquisition sub-mode, the polling channel sequence is all N channels, and the switch control timing sequence controls the time-division multiplexing switches to cycle and switch between all N channels, and an idle period is inserted between the gating of adjacent channels.

4. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 1, characterized in that, When the system is configured into the event trigger acquisition mode, the polling channel sequence is empty; The selection arbitration module only generates the switch control timing sequence according to the event trigger sequence; When an event trigger signal is generated in any channel, the switch control timing sequence conducts this channel, and performs multiple high-speed samplings on this channel at a fixed sampling interval within a preset event sampling window; 5. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 4, characterized in that, The length and sampling interval of the event sampling window are preset values; When multiple event trigger signals are generated simultaneously, the selection arbitration module arbitrates the event trigger signals to determine the gating order of each event channel; When the number of channels undergoing event sampling reaches the preset upper limit of the system, newly generated event trigger signals will be ignored.

6. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 1, characterized in that, When the system is configured in hybrid acquisition mode, the channel selection control module simultaneously enables the polling channel sequence and the event triggering sequence; The selection arbitration module is used to dynamically insert the event trigger sequence into the idle period of the polling channel sequence to generate the switch control timing. When no event is triggered, the switch control timing is selected and sampled in a conventional manner according to the polling channel sequence; When an event is detected, the selection arbitration module allocates the event trigger gating request of the corresponding channel to the nearest idle period to enable the channel and perform high-speed sampling at a fixed sampling interval within a preset event sampling window.

7. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 6, characterized in that, The sampling interval for the event channel within the event sampling window is greater than the gating interval between adjacent channels in the polling channel sequence; In the hybrid acquisition mode, the system alternately selects multiple event channels in a time-sharing manner, and schedules the selection arbitration module between event sampling and polling sampling to ensure that only one channel is selected at any given time. The upper limit of the event sampling channels in the event-triggered acquisition mode is greater than the upper limit of the event sampling channels in the hybrid acquisition mode.

8. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 1, characterized in that, The analog front-end processing link includes a low-noise amplifier, a variable gain amplifier, and an analog-to-digital converter connected in sequence. The selected electrode channel signal is amplified by the low-noise amplifier, and after the gain is adjusted by the variable gain amplifier, it is converted into a digital signal by the analog-to-digital converter.

9. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 8, characterized in that, It also includes a digital logic module, which is connected to the output of the analog-to-digital converter, for receiving and processing the digital signal, and performing time synchronization, data fusion and subsequent digital processing on the converted digital signal.

10. The dual-channel multi-mode low-power EEG signal acquisition system according to claim 1, characterized in that, The EEG signals include continuous local field potential (LFP) signals and sparse action potential (AP) signals. The polling channel sequence is used to acquire continuous local field potential (LFP) signals, and the event triggering sequence is used to acquire sparse action potential (AP) signals in an event-driven manner.

Images