Multi-mode integrated nerve perception fusion wearable helmet

Abstract

The invention discloses a noninvasive portable wearable helmet device integrating visual, auditory, olfactory and electroencephalogram EEG signal synchronization analysis multi-mode integrated monitoring. The noninvasive portable wearable helmet device is used for early screening, differential diagnosis and scientific research of neurodegenerative diseases. The core structure comprises a lightweight helmet main body, a VR glasses visual projection unit, a bone conduction auditory stimulation module, a piezoelectric atomization olfaction release system, a Bluetooth synchronous control module and a dry electrode EEG array acquisition module. And a Windows system is accessed through an external USB (Universal Serial Bus) driver. Multi-modal data can be fused in real time after being input into a computer, and the cost is far lower than that of traditional multi-device and PET joint inspection. According to the equipment, through a standardized sensory stimulation module, simple state and complex task state stimulation is singly or synchronously released, multi-dimensional physiological signals are collected, a high-density network electroencephalogram collection system is combined, electroencephalogram real-time feedback is obtained, artificial intelligence analysis can be combined in the future, and real-time monitoring can be achieved. The method is expected to realize high-precision classification and mechanism research of cognitive impairment such as Alzheimer's disease and dementia with Lewy bodies and neurodegenerative diseases with cognitive sensory impairment. The technical blank of noninvasive and low-cost early-stage multi-modal integrated screening equipment is filled, and the early-stage multi-modal integrated screening equipment is suitable for health management of large-scale and community hospitals and families.

Description

Technical Field

[0001] The present invention belongs to the field of neurodegenerative disease diagnosis and medical wearable devices, and specifically relates to a non-invasive intelligent helmet system with multimodal perception fusion. Background Art

[0002] Impaired vision, hearing, and smell are associated with hundreds of diseases. Sensory dysfunction may be an early warning signal for various neurological and physical diseases and may be an effective tool for early disease detection and prevention. Studies have shown that olfactory, auditory, and visual dysfunction may occur in the early stages of neurodegenerative diseases such as Alzheimer's disease and Lewy body dementia, and specific sensory dysfunction has been confirmed to be a risk factor for their onset. The mechanisms of sensory impairment vary among different neurodegenerative cognitive disorders. For example, a decrease in acetylcholine has been confirmed in Alzheimer's patients, while the auditory and olfactory systems are normally rich in neurotransmitters such as acetylcholine. The decrease in these neurotransmitters may be an important cause of memory impairment and other cognitive dysfunctions. Lewy body dementia, on the other hand, is more likely to be caused by dopamine transport disorders caused by α-synuclein. Furthermore, damage to the visual system in Alzheimer's disease is primarily caused by the loss of ganglion cells, ultimately leading to decreased contrast sensitivity and damage to the color vision pathway. Patients with Lewy body dementia experience impairments in color perception, shape and object recognition, spatial and motion perception, visual construction tasks, and illusions related to visual cortex and network dysfunction. Furthermore, damage to the three sensory systems, both individually and in their connectivity, is closely associated with cognitive impairment and disease progression.

[0003] With the widespread clinical research on sensory impairment testing, traditional diagnostic techniques for neurodegenerative diseases, especially dementia, which are invasive and expensive, may be replaced by this type of sensory testing. Furthermore, because neurodegenerative cognitive disorders can present with abnormalities in multiple white matter fiber tracts, clinically manifested by abnormalities in the ability to integrate multiple senses, different neurodegenerative diseases have different patterns of cognitive impairment and sensory abnormalities. Therefore, the ability to integrate multiple senses is expected to be able to screen for early screening of neurodegenerative cognitive disorders and differential diagnosis.

[0004] The existing sensory detection device (CN202411679981.7) has a single function and only detects one sense organ, but lacks the ability to detect the patient's sensory interaction.

[0005] In addition, the existing technology has the following core defects:

[0006] 1. Invasiveness: Lumbar puncture for the detection of cerebrospinal fluid biomarkers (such as Aβ42 and pTau) requires an invasive procedure, and patient compliance is poor.

[0007] 2. High cost: PET scans (Aβ imaging, Tau imaging, dopamine transporter PET) cost over $5,000 per scan and require the injection of radioactive tracers, making them difficult to popularize.

[0008] 3. Low sensitivity: Traditional cognitive assessments (such as the MMSE and MoCA) are only 60%-70% sensitive for early-stage Alzheimer's disease, insensitive to very early or mild cognitive impairment, and unable to differentiate between different types of dementia. While the diagnostic markers for Alzheimer's disease, ptau181 and ptau217, are highly sensitive, they are susceptible to interference from factors such as diet, age, and stress. Furthermore, effective, highly sensitive, and specific biomarkers for other cognitive impairments have yet to be discovered.

[0009] 4. Single-modality limitations: Traditional testing focuses solely on a single sense (such as a single EEG or olfactory test) and cannot capture abnormalities in cross-modal interactions. Senses influence each other, and measuring only one sense is significantly influenced by the others and cannot accurately reflect disease progression. The mechanisms of single and combined sensory integration in cognitive impairment differ significantly.

[0010] 5. Detection Limitations: Auditory, visual, and olfactory stimulation requires the coordination of multiple devices, making operation complex and the devices difficult to synchronize. Data analysis is fragmented and significantly affected by environmental stimuli, including heterogeneous signals from the same senses. The lack of multi-source signal fusion and accurate capture of multimodal data for application to model algorithms results in insufficient diagnostic sensitivity (e.g., misdiagnosis rates for Alzheimer's disease and Lewy body dementia exceed 30%).

[0011] Therefore, we designed and invented a wearable helmet for multimodal neural perception fusion, which can comprehensively evaluate the single functional impairment of each sense organ and the multi-sensory coordination ability impairment, aiming to improve clinical and research value, so as to improve the early differential diagnosis of neurodegenerative diseases, especially dementia, by medical personnel. Summary of the Invention

[0012] One of the purposes of the present invention is to provide a wearable neurodegenerative disease dementia differential diagnosis helmet for multimodal neural perception fusion, so as to collect the patient's EEG response under different sensory stimulations, and perform early screening and differential diagnosis of the disease based on this.

[0013] The second purpose of the present invention is to strengthen the research on the mechanism of neurological diseases through real-time feedback of multimodal and task-state EEG combined with multi-data analysis.

[0014] To achieve the first objective above, the present invention adopts a technical solution: providing subjects with a realistic physical scene experience through a semi-enclosed helmet, and then collecting the subjects' EEG responses when experiencing different sensory stimuli. The present invention consists of the following parts:

[0015] 1. Helmet body: The main function is to be used directly by the patient and to conform to the ergonomic position effect. It integrates various modular components for wear. Module components: (1) Materials: 3D printed Polymaker PC-Max lightweight magnesium alloy or carbon fiber outer frame (weight <300g) + medical-grade silicone lining. The outer frame is lightweight, dustproof and waterproof, and the inner lining is medical-grade, anti-allergenic, and removable for cleaning. (2) Front module distribution: LED visual projection screen (viewing angle 60°) and VR glasses equipment. (3) Top module distribution: 256 / 128-lead EEG dry electrode array. (4) Side module distribution: bone conduction speaker. (5) Lower module distribution: replaceable odor capsule array, odor release port (ultrasonic atomizer). (6) Rear module distribution: Embedded AI processor (Windows system, running deep learning model). (7) Power supply module: Split rechargeable battery, BYD 7.6V 6000mAh, with long battery life.

[0016] 2. Visual stimulation module: Its main function is to implement single-image and task-based multi-modal visual sensory stimulation in a closed environment. Module components: (1) Tobii Pro VR visual stimulation module. It captures the user's gaze focus in real time and supports gaze interaction and behavior analysis in virtual environments. (2) Rockchip RK3588 main control chip and display output unit. Achieve independent output of 3840×2160 resolution for both eyes, totaling 4K-level images. (3) BOE MV049F8N-40VR display. HDR10+ enhances light and dark contrast, adapting to virtual reality lighting effects. It is directly connected to the RK3588 main control via the MIPI-DSI interface and supports multi-channel signal distribution. (4) USB 3.0 Type-C interface. It realizes efficient data transmission and power management, and enables high-speed data transmission between PC and VR device (such as firmware burning and debugging logs).

[0017] 3. Auditory stimulation module: The main function is to implement dual-mode auditory sensory stimulation of pure tone and semantic speech. Module components: (1) Shokz OpenComm2 bone conduction unit. Implements non-in-ear audio output and supports low-latency encoding. (2) Knowles TEC-00046 microphone array. Parameters: dual-port 8Ω impedance, 20-20kHz frequency response, ambient sound acquisition, voice interaction, active noise reduction, and 3D sound recording. (3) TITPA6132A2 audio amplifier. Used for microphone array input, single-interface I2S input, SNR ≥ 110dB, noise cancellation, and support for Hi-Res audio.

[0018] 4. Olfactory stimulation module: The main function is to implement rapid single odor atomization inhalation olfactory sensory stimulation. Module components: (1) Aromajoin AroMaker Mini micro-scent capsule system. Contains 40 standard micro-capsules (including phenylethyl alcohol, rose oil, putrescine, etc.), packaged in a rotatable card slot (each volume is 1 mL). (2) SMC VQ110U-5G micro solenoid valve. Parameters: Four-port 12VDC, 10ms response time, drives the airflow control valve Infineon IRLZ44N via MOSFET to control the scent release switch and provide precise airflow guidance. (3) MSB1230 micro-core atomizer. Liquid scent atomization, parameter four-port 5V, 20kHz ultrasonic frequency. Connect the PWM port to the scent valve to precisely adjust the scent concentration.

[0019] 5. Synchronous control module: The main function is to remotely control each sensory module through multi-mode Bluetooth connection. Module components: (1) QCA6391 Bluetooth module. Wireless control, transmission distance ≥10m, and anti-2.4GHz interference. (2) Master GPIO. All stimulation time alignment accuracy is ±1ms. Supports 7 combination modes (such as video only, audio and air synchronization, etc.) (3) Software logic. A computer software code design. The function is named as Polymorphic Scene EEG Collector Software APP.

[0020] 6. Signal acquisition module: The main function is to synchronously capture the EEG data of the multimodal sensory module. Module components: (1) Electroencephalogram (EEG): 256 / 128-lead dry electrode array, dynamic impedance calibration (<10 kΩ) (adaptive pressure adjustment and real-time impedance monitoring), sampling rate 2048 Hz, bandpass filtering 0.5-100 Hz, common mode rejection ratio >110 dB. Supports multi-channel synchronous acquisition, covering the entire brain area. (2) EEG caps are widely used in hospitals. The cap is made of elastic silicone, which is comfortable to wear and suitable for most head shapes. The disc-shaped soft gel electrodes are fixed on the elastic mesh cap, which is convenient for positioning. The mold sleeve is soft, and the fabric chin strap can reduce the wearer's pressure pain and sweating. The conductive paste is applied to capture the signal, and it is easy to wear and clean.

[0021] To achieve the second objective, the present invention employs the following technical solutions: designing and assembling a helmet and connecting it to a multifunctional module. Simultaneously, a synchronization control software was developed and integrated with the EEG device. The clinician then initiates the capture of multimodal EEG data under various sensory conditions.

[0022] Modular equipment assembly:

[0023] 1. EEG connection.

[0024] (1) Install the disc electrode. The new disc-shaped EEG cap requires the installation of disc electrodes for the first time. They do not need to be removed after use, and subsequent use does not require reinstallation. Select the appropriate EEG cap based on the patient's head circumference. Insert one end of the disc electrode harness into the cap's mounting bracket, where the electrode is marked, and connect the other end to the host computer.

[0025] (2) Wear the EEG cap and connect the adapter cable to the disc electrode harness. The disc-shaped EEG cap uses the international 10-5, 10-10, and 10-20 positioning systems for convenient and accurate positioning. Position the cap as soon as you put it on. Place the cap on the corresponding positioning points. Plug the disc-shaped electrode harness into the adapter cable at the corresponding markings and connect it to the amplifier.

[0026] (3) Fill the electrode cup with conductive paste. The new disc-shaped EEG cap is made of silicone material with great elasticity. You can distribute the electrode cup, clean the scalp, and add ten20 conductive paste by lifting it.

[0027] 2. Connect each module to the helmet and then to the host.

[0028] (1) The LED screen of the VR glasses fits the forehead, is fixed with straps, and the brightness is automatically adjusted.

[0029] (2) The bone conduction speaker is embedded in the temporal cavity of the helmet and fixed with a 3D printed alloy bracket. A 0.8 mm thick silicone buffer layer is set between the bracket and the helmet lining.

[0030] (3) The odor capsule slot adopts a magnetic adsorption design (supporting quick replacement). 40 odor capsules are loaded into the rotating slot according to their numbers (each capsule corresponds to a unique RFID tag) and connected to the piezoelectric atomizer drive circuit (voltage 12V, frequency 20kHz).

[0031] 3. Module equipment optimization and trial.

[0032] (1) Optimization of EEG electrode calibration signal acquisition. The EEG electrodes utilize a coil spring structure that adapts to the curvature of the scalp. Enable impedance detection mode and adjust the electrode pressure until the contact impedance is <5kΩ. The software interface displays the electrode contact status in real time (green / red indicator). Connect the cable terminals to the amplifier.

[0033] (2) Equipment trial. After installing all drivers on the computer, record sounds, pictures, videos, scenes, etc. in advance. Connect the Bluetooth part, perform a single stimulus, and collect signal data.

[0034] The beneficial effects of the present invention are:

[0035] (1) This helmet adopts a semi-enclosed structure, combining simple and complex sensory module stimulation to reflect changes in EEG signals, which is expected to capture related abnormalities at an early stage, perform differential diagnosis of the spectrum of degenerative diseases with impaired cognitive sensory function of the nervous system, and is expected to detect changes in sensory function of diseases at an early stage.

[0036] (2) The signal synchronization accuracy of this helmet is higher, the spatial multi-sensory consistency is optimized, the motion artifacts are reduced, high user experience and experimental controllability are achieved, multimodal interaction is enhanced, and the depth of task-state data fusion is increased.

[0037] (3) The helmet takes a short time to stimulate and collect signals, is simple to operate, and is easy to teach, meeting the special requirements of clinical neurology.

[0038] (4) This helmet may have great clinical significance and research value, especially helping medical personnel in the neurology department to have a deeper understanding of neurodegenerative diseases, and assist in their differential diagnosis and scientific research. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 This is a diagram of the device working mode.

[0040] Figure 2 A schematic diagram of the helmet structure.

[0041] Figure 3 Schematic diagram of the visual module structure.

[0042] Figure 4 Schematic diagram of the auditory module structure.

[0043] Figure 5 Schematic diagram of the olfactory module structure.

[0044] Figure 6 It is the operating interface of the computer control software. DETAILED DESCRIPTION

[0045] This embodiment is implemented using the following technical solutions. The present invention is a multimodal integrated neural perception fusion wearable helmet, which includes three core component functions: single or synchronous release of simple and complex task-state sensory stimulation, high-density network EEG multi-dimensional data real-time acquisition, remote software synchronous control and data acquisition; and six core devices: lightweight helmet body, VR glasses visual projection unit, bone conduction auditory stimulation module, piezoelectric atomization olfactory release system, Bluetooth synchronization control module, and dry electrode EEG array acquisition module.

[0046] The embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0047] Various embodiments and examples are disclosed below, and different structures of the present invention are shown in the accompanying drawings. For ease of understanding, the components and settings of specific examples are described in detail. These examples are for reference only and do not limit the scope of the invention. Those skilled in the art should realize that the application of other processes or materials is also feasible. For example: for the helmet structure, a more optimized and more comfortable fiber soft material can also be used. For another example, for the olfactory module, an olfactory stick can also be used for stimulation. It is planned to improve the setting of more stable circuits, more precise software control and calibration devices.

[0048] The specific implementation includes 10 standardized testing steps:

[0049] 1. Opening and using the device. After the device is turned on, connect it to the corresponding WeChat applet via Bluetooth, and perform the device power on / off and related operations through the mobile phone. For reference, see the attached operation interface. Figure 6 .

[0050] 2. The patient puts on the helmet and starts the self-test procedure (takes 10 seconds). (1) Visual module: LED screen displays RGB three primary colors in full screen (CIE 1931 coordinate error <0.005). (2) Auditory module: Play a 1kHz calibration tone (SPL 70dB±2). (3) Olfactory module: Perform airway negative pressure detection (target value -5kPa±0.3). (4) EEG system: perform electrode impedance scanning (<5 kΩ threshold) and noise floor detection (RMS <3 μV@1-100 Hz).

[0051] 3. Phase 1 (3 minutes): Olfactory stimulation was alternating according to a preset sequence, and EEG signals were recorded simultaneously. (1) Stimulation parameters: 40 odors were presented (concentration gradient 0.1-1000 ppm), and a single stimulation lasted 500 ms. (2) Airflow control: Piezoelectric valve opening time accuracy of ±0.1ms, laminar flow nozzles are used to reduce odor mixing (flow rate 0.3m / s). (3) EEG detection: Focus on capturing olfactory event-related potentials (ERPs): peak latency and amplitude of the P2 component (latency 200-400ms) and the P3 component (300-500ms). Semantic conflict of olfactory stimuli may trigger N400. Establish an odor identification index (ORI = amplitude / baseline noise). Time-frequency characteristics: gamma band (30-80Hz) power changes in olfactory cortex activity, which are related to odor identification speed. Phase synchronization: Theta (4-8Hz) phase locking value (PLV) reflects the connection between the prefrontal lobe and the limbic system (such as the hippocampus).

[0052] 4. Phase 2 (3 minutes): Alternate visual stimulation according to a preset sequence and simultaneously record EEG signals. (1) Stimulus parameters: monochrome images, common objects, biological images, International Affective Picture System (IAPS) standardized images, spatial frequency 0.5-30 cpd, presentation time 300 ms. (2) Optical flow control: Dynamically adjust the gamma value (2.2-2.6) to compensate for ambient light changes and prevent VEP signal saturation. (3) EEG detection: Focus on capturing visual evoked potentials (VEP): early components C1 (50-100ms, reflecting primary visual cortex activation) and N1 (150-200ms, attention regulation). Frequency band energy: alpha band (8-12Hz) suppression reflects visual attention, while gamma band (>30Hz) enhancement is associated with feature binding. Steady-state visual evoked potential (SSVEP): frequency-specific response to dual-frequency flicker stimulation (e.g., 15Hz / 20Hz stimulation corresponds to 15Hz / 20Hz energy peak). Visual attention index (VAI) is calculated based on the steady-state response.

[0053] 5. Phase 3 (3 minutes): Alternate auditory stimulation according to a preset sequence and simultaneously record EEG signals. (1) Stimulus parameters: pure tone (250-8000 Hz), common biological calls, alternating between named words and complex sentences, dynamic matching of sound pressure level (55-75 dB). (2) Acoustic flow control: Implement Notch filtering (center frequency is offset from screen refresh rate by ±5Hz) to eliminate more than 95% of mechanical resonance. (3) EEG testing: Focus on capturing auditory evoked potentials (AEP): N1-P2 complex wave (80-200ms), calculate the interhemispheric asymmetry index (HAI = left / right amplitude ratio), which reflects the processing of physical characteristics of sound. MMN (mismatch negativity) is used to detect sound differences. Time-frequency decomposition: Cross-frequency coupling (CFC) of theta (4-8Hz) phase and gamma amplitude, reflecting auditory working memory. Bone conduction specificity: Enhanced beta (12-30Hz) synchronization reflects the connection between the temporal lobe and parietal lobe, which is different from air conduction hearing.

[0054] 6. Stage 4 (5 minutes): Visual images + auditory stimulation, simultaneous recording of EEG signals. (1) Visual and auditory combined stimulation: monochrome pictures and color naming words, pictures of common objects and naming words, pictures of common creatures and their sounds. Matching and mismatching groups are used. (2) EEG testing: Comparison of N1 amplitude differences (bimodal suppression or enhancement) and P300 latency between unimodal and bimodal responses, and measurement of P300 amplitude differences. Constructing the audiovisual integration function (AIF) = (synchronous reaction time - asynchronous reaction time) / baseline reaction time. Cross-modal phase synchronization: Theta-band PLV, reflecting the connection between the occipital lobe (visual) and temporal lobe (auditory), reflects spatiotemporal integration.

[0055] 7. Stage 5 (5 minutes): Visual image + olfactory release identification, simultaneous recording of EEG signals. (1) Visual and olfactory combined stimulation: the matching group (lemon picture + lemon fragrance) and the non-matching group (lemon picture + diesel smell) were pseudo-randomly presented. (2) EEG testing: The patient was required to press the confirmation button within 1500ms (force sensing threshold 50g), and the accuracy and reaction time were recorded. The olfactory recognition time (ORT) was calculated as P300 peak latency - odor release instruction delay (15ms). Time-frequency decomposition (Morlet wavelet) was used to calculate the phase synchronization index (PLV) of the visual theta band (4-8Hz) and the olfactory gamma band (30-80Hz). Semantic conflict response: The N400 amplitude was increased in conflict trials (such as lemon image + diesel smell). Cross-band coupling: The modulation relationship between the visual alpha (8-12Hz) phase and the olfactory gamma amplitude. The correlation between reaction time and the orbitofrontal cortex theta oscillation power was calculated.

[0056] 8. Stage 6 (5 minutes): Olfactory release identification + auditory stimulation, simultaneous recording of EEG signals. (1) Auditory-olfactory combined stimulation: Matched pairs (lemon odor + lemon-named words) and non-matched pairs (lemon picture + pencil-named words) were pseudo-randomly presented. 20% of the trials used semantic conflict (e.g., flower scent + alarm sound), and 80% of the trials used congruent pairing (e.g., flower scent + bird song). (2) EEG detection: Based on the difference in the amplitude of the N400 component, the late positive component of the P600 (500-800ms) triggered by semantic conflict (e.g., flower fragrance + alarm sound), and the phase synchronization analysis of different frequency bands.

[0057] 9. Stage 7 (5 minutes): Visual VR scene + semantic speech task + olfactory release identification, simultaneous recording of EEG signals. Example demonstration: (1) Scenario Construction: Use the Unity engine to render a dynamic supermarket scene (with >5 million polygons) and integrate a physics engine to simulate object interactions. Alternatively, semantic interference items can be set in the VR supermarket scene (e.g., a wrench image and gasoline smell appearing in the fruit section). (2) Task design: Patients need to complete complex operations in the virtual shelf. (3) Semantic task: phonetic naming of objects. (4) Olfactory recognition: When a specific product is picked up, the corresponding odor is released (e.g., milk scent → soap scent). (5) Anomaly detection: 5% abnormal objects are randomly inserted into the scene (e.g., a wrench appears in the fruit area), and the time of discovery is recorded. (6) EEG detection: High-order ERP: The spatiotemporal distribution pattern of the late component (e.g., LPC, 500-1000ms) induced by multimodal stimulation. Entropy analysis: Multiscale entropy (MSE) quantifies the complexity of neural signals and reflects the level of cognitive load.

[0058] 10. Data Transmission and Processing. (1) Data transmission. After each test is completed, the data is automatically saved to the software, recording quantitative EEG data such as olfactory recognition time, auditory evoked potentials, and visual evoked potentials, while also recording EEG waveform signals. The data is automatically stored in the host PC processor. Patient information is recorded in different folders for subsequent processing. (2) Data calibration and processing. Single stimulation does not require time window alignment, but multi-dimensional stimulation does require time window alignment: the onset of auditory stimulation is defined as time zero. Abnormal EEG patterns are identified by analyzing single sensory recognition, multi-sensory association, task-based cognition, and sensory integrated processing capabilities through EEG signal analysis.

[0059] It is expected that a database will be established and promoted in the future: the present invention can initially select a large tertiary hospital as a pilot center to provide comprehensive training to professional and technical personnel, including technical operations, troubleshooting, etc. At the same time, necessary equipment and technical support will be provided to ensure that the medical center can effectively implement and operate the system. The effectiveness and accuracy of the system will be verified after a single-center trial. Then, suitable sub-centers will be selected domestically for promotion, and corresponding technologies and equipment will be deployed to ensure consistency of operations and data quality. After collecting data, a large national database will be established. Finally, cooperative relationships can be established with international medical institutions and research institutions to further promote technology, establish an international data sharing platform, promote medical research and cooperation on a global scale, and continuously upgrade and optimize technology based on global usage and feedback to meet the needs of different regions and countries.

[0060] The present invention effectively partially integrates existing technologies with new technologies or new methods, and provides original method designs and uses to ensure the best results. Any parts not elaborated in detail belong to the existing technology.

[0061] The scope of protection of the present invention is clear and flexible to accommodate possible technical improvements and implementation adjustments. Although the above content describes specific embodiments of the present invention in conjunction with the accompanying drawings, the feasibility and operability need to be re-evaluated by those skilled in the art. Various modifications may be made to the specific embodiments without departing from the principles and essence of the present invention. The scope of the present invention is limited solely by the appended claims.

Claims

1. The present invention is a wearable helmet device with multi-modal integrated neural perception fusion, characterized by It integrates visual, auditory, and olfactory stimulation, high-density EEG signal acquisition, and synchronized Bluetooth software control functions. The core structure includes: a lightweight helmet body, a VR glasses visual projection unit, a bone conduction auditory stimulation module, a piezoelectric atomization olfactory release system, a Bluetooth synchronization control module, and a dry electrode EEG array acquisition module. It is connected to the Windows system through an external USB driver. The present invention has three main functions: First, it can collect EEG signals during single sensory stimulation and mixed stimulation of multiple senses. Second, by analyzing the EEG connection under stimulation, it can indirectly infer whether the fiber network between sensory signals is abnormal. Third, different fiber network abnormalities can be used to establish an EEG cognitive omics database, thereby promoting clinical differential diagnosis and scientific research on neurodegenerative diseases, especially cognitive disorders. This product is of moderate size and weighs less than 300g. The helmet is ergonomically designed, clean and comfortable to wear, impact-resistant, and high-temperature resistant. It has a compact structure, supports modular assembly, has a moderate data acquisition time, and is easy to obtain various parameters.

2. The visual stimulation described in claim 1 utilizes a visual EEG acquisition module, combining EEG equipment with VR wearable glasses. This module seamlessly integrates with the HTC Vive cable and headset module, allowing for reuse of eye scenes and stimuli, creating scenes using the Unity VR engine. It boasts a robust sampling rate and scene projection capabilities. On the front of the headset, VR glasses and a display screen display various pure-color lights, patterns, and scenes, achieving high refresh rates and clear output for binocular VR images. The other end connects to a host computer via an HTC Vive cable for data transmission. The greatest advantage of this module is that it can integrate existing VR glasses with EEG systems to display different task states and VR scenes, providing multimodal data in various scenarios with guaranteed data accuracy and time precision. The relevant data can be directly transmitted to the host computer and is suitable for analysis using Matlab, Python, C, or .NET. The device supports firmware flashing and real-time debugging. Furthermore, in semi-enclosed environments, it effectively reduces interference from ambient stimuli with the same sensory organs but with different signals.

3. The auditory stimulation described in claim 1 utilizes an auditory EEG acquisition module, combining EEG equipment with a bone conduction speaker, supporting both pure tone and semantic speech stimulation. The auditory acquisition module utilizes sound waves of varying frequencies and amplitudes. A bone conduction player is secured to the mastoid process of the temporal bone and connected to a computer via Bluetooth. Suspended straps secure the connection and place it inside the helmet on either side. Ensure that both sides maintain the same position, orientation, angle, and distance from the nose tip. A bone conduction unit provides non-in-ear audio output, supports low-latency encoding, reduces power consumption by 80% in silent mode, and features adjustable bone conduction vibration amplitude. A microphone array is used for ambient sound acquisition, voice interaction, active noise reduction, and 3D audio recording. An audio amplifier is used for microphone array input to amplify loudness. Low-latency connectivity to a host computer is achieved through BLE 5.0, TTL serial communication, and Qualcomm devices. The module's greatest advantage is its pre-recorded audio recording of various pure tone and semantic scenarios, enabling customized patient perception. Furthermore, the semi-enclosed design effectively reduces noise and prevents interference from other sound waves.

4. Olfactory stimulation according to claim 1, we apply olfactory EEG acquisition module, use EEG equipment and piezoelectric atomization olfactory release system in combination, which includes 40 standardized odor capsules and piezoelectric atomizers, and the odors are selected according to the WHO olfactory recognition test specifications. Through the nose, there is no gas residue, the odor stock solution is stored by micro-odor capsules, the piezoelectric atomizer liquid is micronized, the response time is short, the single odor release accuracy is high, and the switching delay is small. In addition, the SMC electromagnetic airflow control valve can accurately guide the airflow to achieve directional odor release (such as wind out of the left nose side), and a small dose can ensure rapid olfactory recognition, and the inhaled dose can be controlled by giving the amount through the mask instead of the air, so as to avoid different inhaled liquid contents caused by physical factors such as liquid density, airflow influence, and volatility.

5. According to the visual operation interface described in claim 1, we use C++ and other programming technologies to design a visual operation interface. The interface can control the visual module, auditory module and olfactory module individually or simultaneously. The interface can display the connection status, power level, and single operation time of different modules, can select scene modes, can play pre-recorded video and audio, and can release special gases at the same time to give patients multi-sensory stimulation. The operation interface can adjust the brightness and volume, and can display the real-time monitoring results of EEG brain waves. At the same time, in order to ensure the uniformity of the timestamps of multi-sensory stimulation, the operation interface also includes a quick calibration button. In order to avoid any discomfort and emergencies caused by the patient while wearing the helmet, the operation interface includes an emergency brake button to stop the operation of the equipment.

6. A dementia screening and classification method based on the device of claim 1, characterized by EEG parameters under multimodal synchronous control. Including multimodal stimulation timing synchronization, EEG signal synchronous acquisition, and multi-task deep machine learning diagnostic process. Our features of interest are intended to be selected from EEG signals under single sensory stimulation and mixed multi-sensory stimulation, as well as task-state sources. In the future, they can be used to develop new models to output disease risk indexes based on big data models. Combined with a variety of machine learning methods, such as support vector machines, gradient boosting machines, random forests, Boosting, stacked generalization, etc., multivariate classification analysis is performed. In the future, multimodal data combined with functional magnetic resonance imaging under task state is expected to accurately predict subtle structural and functional changes in various diseases.

Images