A multi-modal wearable monitoring system and method for atelectasis assessment

Abstract

The application discloses a kind of multi-modal wearable monitoring systems and methods for atelectasis evaluation, it is related to respiratory system disease monitoring technical field, including bioelectrical impedance measurement module, wearable ultrasound monitoring module, data acquisition and preliminary processing unit, central processing unit and data transmission and remote monitoring module;Bioelectrical impedance measurement module adopts 4 electrode technology to acquire thoracic impedance dynamic signal, wearable ultrasound monitoring module acquires pleural sliding image by multi-point layout, data acquisition and preliminary processing unit realizes double modal signal synchronous acquisition and pre-processing, central processing unit completes atelectasis determination and positioning by multi-modal data fusion algorithm, data transmission and remote monitoring module realizes data transmission and real-time alarm.The application is complementary by two kinds of technologies, overcome the defect that single monitoring means signal is not stable, positioning is not accurate, realize the non-invasive, real-time, accurate monitoring of atelectasis, provide reliable support for early clinical intervention.

Description

Technical Field

[0001] This invention relates to the field of respiratory disease monitoring technology, and in particular to a multimodal wearable monitoring system and method for assessing atelectasis. Background Technology

[0002] Currently, medical monitoring technology is rapidly developing towards non-invasive, real-time, continuous monitoring and early warning. In the field of respiratory diseases, the incidence of chronic obstructive pulmonary disease (COPD), pneumonia, and other lung diseases continues to rise, making timely and accurate detection and assessment of patients' lung function a major challenge in clinical diagnosis and treatment. Traditional lung detection methods mainly rely on radiographic imaging and routine ultrasound examinations, which have shortcomings such as radiation risks, large equipment size, and limited detection frequency, making it difficult to detect early lesions in a timely manner. With the improvement of wearable technology, sensor technology, and the level of intelligent data processing, medical testing equipment is moving towards portability, low power consumption, and multimodal fusion, providing new technical approaches for non-invasive and continuous detection of respiratory abnormalities such as atelectasis.

[0003] Atelectasis is a condition in which part or all of the lung tissue collapses due to insufficient gas supply. Early detection is crucial for clinical treatment and prognosis. Among existing technologies, bioelectrical impedance analysis (BIA) and ultrasound imaging each have their advantages but also limitations: Wearable BIA uses a four-electrode technique to collect dynamic changes in the electrical conduction of chest tissue, reflecting information such as local tissue water content and ventilation status. However, it is affected by skin condition, electrode contact, and motion interference, resulting in insufficient data resolution and a high risk of misjudgment. Wearable ultrasound devices determine the state of lung tissue by observing pleural sliding, enabling distributed real-time monitoring. However, when used alone, it is easily affected by the operator's experience and the distribution of local tissue, resulting in insufficient positioning accuracy and signal consistency.

[0004] Most existing atelectasis monitoring technologies rely on a single detection method, which not only suffers from unstable signals, inaccurate positioning, and difficulty in continuous monitoring, but also often requires complex signal processing and high-precision equipment to improve data acquisition accuracy, increasing manufacturing and maintenance costs, limiting clinical application, and failing to meet the clinical needs for early, accurate, and continuous monitoring of atelectasis. Summary of the Invention

[0005] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present invention is that the existing atelectasis monitoring technology relies on a single detection method, which has the disadvantages of unstable signal, inaccurate positioning, and difficulty in continuous monitoring, and cannot meet the clinical needs for early, accurate and continuous monitoring of atelectasis.

[0006] To achieve the above objectives, this invention provides a multimodal wearable monitoring system for atelectasis assessment, characterized by comprising a bioelectrical impedance measurement module, a wearable ultrasound monitoring module, a data acquisition and preliminary processing unit, a central processing unit, and a data transmission and remote monitoring module; the bioelectrical impedance measurement module and the wearable ultrasound monitoring module are respectively connected to the data acquisition and preliminary processing unit, the data acquisition and preliminary processing unit is connected to the central processing unit, and the central processing unit is connected to the data transmission and remote monitoring module; the bioelectrical impedance measurement module is used to acquire dynamic bioelectrical impedance signals within the pleural cavity, the wearable ultrasound monitoring module is used to acquire pleural sliding image signals, the data acquisition and preliminary processing unit is used for simultaneous acquisition, preprocessing, and preliminary fusion of dual-modal signals, the central processing unit is used for data storage, dynamic signal analysis, and atelectasis determination and localization, and the data transmission and remote monitoring module is used for data transmission and real-time alarm.

[0007] Preferably, the bioelectrical impedance measurement module includes an electrode unit, a signal excitation and detection unit, and a preliminary amplification and filtering module; the electrode unit includes two excitation electrodes and two detection electrodes, which are attached to the left and right sides of the subject's chest; the signal excitation and detection unit is used to apply alternating current and capture impedance change signals; the preliminary amplification and filtering module is used to perform low-noise amplification and multi-stage filtering processing on the acquired analog signals.

[0008] Preferably, the bioelectrical impedance measurement module further includes a self-calibration module, which automatically detects and compensates for signal deviations through periodic hardware calibration and dynamic software calibration; the hardware calibration is achieved through a built-in precision resistor network, and the dynamic software calibration adjusts parameters based on the signal quality index.

[0009] Preferably, the wearable ultrasound monitoring module includes an ultrasound probe unit and an ultrasound signal processing unit; the ultrasound probe unit adopts a multi-point layout design and is attached to the chest, back, or side according to the subject's posture; the ultrasound signal processing unit is used for receiving, amplifying, digitally processing, and extracting pleural sliding features of ultrasound signals.

[0010] Preferably, the imaging mode of the ultrasonic probe unit is M mode, and the ultrasonic signal processing unit has a built-in digital signal processor for real-time envelope detection and noise reduction of the echo signal.

[0011] Preferably, the data acquisition and preliminary processing unit includes a multi-channel data acquisition unit and a preprocessing and data fusion module; the multi-channel data acquisition unit supports simultaneous sampling of multiple analog signals to ensure the time synchronization of dual-modal data; the preprocessing and data fusion module uses an embedded processing chip to filter and reduce noise in the data, and uses time-domain and frequency-domain feature fusion technology to construct a joint data model.

[0012] Preferably, the central processing unit includes a data storage and management module and a dynamic signal analysis module; the data storage and management module is used for caching and storing multimodal data, and the dynamic signal analysis module, based on a multimodal data fusion algorithm, performs real-time analysis of bioelectrical impedance change curves and pleural sliding features in ultrasound images to complete the risk assessment and localization of atelectasis.

[0013] Preferably, the multimodal data fusion algorithm adopts a hybrid model that combines feature-level fusion and decision-level fusion, including data alignment, feature standardization, feature-level weighted fusion, and decision-level classification steps; the decision-level classification uses a support vector machine classifier to output a binary judgment of atelectasis and an abnormality score.

[0014] Preferably, the data transmission and remote monitoring module supports Bluetooth and / or Wi-Fi wireless communication protocols to transmit the processing results to the remote monitoring terminal and trigger different levels of alarms based on the atelectasis determination results.

[0015] Another aspect of the present invention provides a multimodal wearable monitoring method for assessing atelectasis, the method comprising the following steps:

[0016] Step 1: Collect dynamic bioelectrical impedance signals in the pleural cavity using a bioelectrical impedance measurement module, and collect pleural sliding image signals using a wearable ultrasound monitoring module.

[0017] Step 2: The dual-modal signals are synchronously acquired, preprocessed, and initially fused through the data acquisition and preliminary processing unit to obtain preprocessed dual-modal feature data;

[0018] Step 3: The central processing unit performs multimodal fusion analysis on the preprocessed dual-modal feature data to determine whether atelectasis and abnormal locations exist.

[0019] Step 4: Transmit the judgment result to the remote monitoring terminal through the data transmission and remote monitoring module, and trigger the corresponding alarm according to the degree of abnormality.

[0020] Preferably, in step 1, the acquisition of bioelectrical impedance dynamic signals is achieved through a 4-electrode technique, where an alternating current is applied to the excitation electrode and the detection electrode captures impedance change signals; the acquisition of pleural sliding image signals is achieved through a multi-point layout of ultrasound probe units according to the subject's posture.

[0021] Preferably, the preprocessing in step 2 includes: performing low-noise amplification, multi-level filtering, and digital processing on the bioelectrical impedance signal to extract the average impedance change rate and spectral entropy characteristic parameters; and performing denoising, contrast enhancement, and edge detection processing on the ultrasound image signal to extract the average pleural sliding velocity and the coefficient of variation characteristic parameters of the pleural sliding amplitude.

[0022] Preferably, the preliminary fusion in step 2 is achieved through time-domain and frequency-domain feature fusion technology to construct a dual-modal joint data model; the multimodal fusion analysis in step 3 includes data alignment, feature standardization, weighted fusion and classification determination steps.

[0023] Preferably, the data alignment is based on timestamps to achieve spatiotemporal matching of dual-modal data; the feature standardization adopts the Z-score standardization method; the weighted fusion calculates the comprehensive anomaly score through a linear weighted model, and the weights are dynamically adjusted according to the signal quality index; the classification judgment is achieved by a support vector machine classifier to realize the binary judgment and localization of atelectasis.

[0024] Preferably, the alarm in step 4 includes a first-level alarm and a second-level alarm. The first-level alarm is triggered when the comprehensive abnormal score exceeds the first threshold and the classification is determined to be atelectasis; the second-level alarm is triggered when the comprehensive abnormal score exceeds the second threshold and the abnormal location is consistent.

[0025] The technical advantages of this invention are as follows: By fusing bioelectrical impedance analysis with wearable ultrasound multimodal analysis, the limitations of single technologies in terms of anti-interference, data accuracy, and local positioning are overcome, thereby improving the accuracy and real-time performance of atelectasis detection; the system adopts a wearable design, is non-invasive and portable, supports long-term continuous monitoring, and is suitable for various clinical and home application scenarios; through reasonable sensor layout and data fusion algorithms, it maintains high stability and response speed in complex human body environments; the simplified structural design reduces manufacturing and maintenance costs, facilitating clinical application.

[0026] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the overall system structure of the present invention;

[0028] Figure 2This is the core process diagram of the bioelectrical impedance measurement module of the present invention and the data transmission relationship diagram between each unit;

[0029] Figure 3 This is a schematic diagram showing the layout and connection of the wearable ultrasound device of the present invention;

[0030] Figure 4 This is a diagram showing the logical relationships and data flow between the modules of this invention;

[0031] Figure 5 This is a flowchart of the dynamic signal analysis algorithm of the present invention. Detailed Implementation

[0032] The following description, with reference to the accompanying drawings, illustrates several preferred embodiments of the present invention to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0033] In the accompanying drawings, components with the same structure are indicated by the same numerical designation, and components with similar structures or functions are indicated by similar numerical designations. The dimensions and thicknesses of each component shown in the drawings are arbitrary, and the present invention does not limit the dimensions and thicknesses of each component. To make the illustrations clearer, the thickness of some components has been appropriately exaggerated in the drawings.

[0034] Example 1

[0035] This embodiment provides the overall system design process and core technical principles.

[0036] 1. System Overall Design Process

[0037] like Figure 4 As shown, the system employs a multimodal detection approach, combining chest bioelectrical impedance analysis with wearable ultrasound to achieve early warning and localization of atelectasis risk. The overall method mainly includes the following steps:

[0038] (1) Signal acquisition:

[0039] A bioelectrical impedance sensor was configured using 4-electrode technology. The excitation electrodes and detection electrodes on both sides were attached to the left and right sides of the subject's chest, respectively, to collect dynamic impedance signals within the thoracic cavity.

[0040] Wearable ultrasound devices with multiple points distributed on the subject's chest or back according to their actual posture are attached to collect images of pleural sliding and related dynamic information in real time.

[0041] (2) Signal front-end processing:

[0042] The bioelectrical impedance signal is pre-amplified and filtered, and then converted from analog to digital signal using a dedicated analog-to-digital converter (ADC).

[0043] The ultrasound signal undergoes image preprocessing (such as denoising, enhancement, and edge detection), and the embedded signal processing module is used to extract key features of pleural sliding in the image in real time (including displacement, velocity, edge clarity, etc.).

[0044] (3) Multi-channel data acquisition and synchronization:

[0045] A multi-channel acquisition device is used to simultaneously sample bioelectrical impedance and ultrasound signals, and techniques such as timestamp marking and event triggering are used to ensure the precise spatiotemporal alignment of the two modal data.

[0046] (4) Signal feature preprocessing and extraction:

[0047] For digital bioelectrical impedance signals, features such as amplitude, phase, spectrum, and respiratory cycle are extracted using methods such as time-domain filtering, fast Fourier transform (FFT), and wavelet transform.

[0048] For ultrasound data, pleural sliding features are extracted using image processing algorithms, focusing on the continuous motion trajectory of local areas.

[0049] (5) Multimodal data fusion and comprehensive judgment:

[0050] A time-synchronization and feature fusion model was constructed to perform weighted fusion of bioelectrical impedance and ultrasound image features, and to compare overall thoracic function information with local imaging abnormalities.

[0051] A decision support module based on rule bases or machine learning algorithms (such as support vector machines and random forest hybrid models) is introduced to score and locate atelectasis.

[0052] (6) System feedback and alarms:

[0053] When an abnormal signal is detected, the central processing unit transmits the result to the remote monitoring terminal in a timely manner through the communication module (Bluetooth, Wi-Fi, etc.) and triggers alarm information to assist in clinical intervention decisions.

[0054] 2. Core Technology Principles and Key Details

[0055] (1) 4-electrode technology: Medical-grade patches are used to ensure good contact with the skin for a long time. The detection signal is collected after a small sinusoidal alternating current is precisely applied, which effectively suppresses contact resistance interference. Multi-stage amplification and signal filtering design is used to overcome the interference of human motion and power frequency noise to ensure signal quality.

[0056] (2) Wearable ultrasound design: The ultrasound device module is designed to be compact and conform to the curvature of the human body. It adopts an adjustable angle fixing bracket and medical adhesive materials to ensure good acoustic coupling between the pleura and the probe in different postures. The built-in ultrasound signal processing circuit and fast Fourier transform algorithm ensure that the device can quickly capture and analyze the sliding characteristics of the pleura and locate abnormal areas of the lung.

[0057] (3) Multi-channel data acquisition and timing synchronization: The multi-channel data acquisition unit designed by the system supports the parallel acquisition of bioelectrical impedance and ultrasound signals. It uses a unified sampling clock and timestamp to achieve data alignment, providing a sufficient data foundation for multimodal fusion.

[0058] (4) Data fusion and judgment strategy: During feature fusion, Bayesian inference and weighted algorithms are used to comprehensively analyze signal feature parameters to improve detection sensitivity. The feedback adjustment mechanism enables the system to automatically correct electrode drift and changes in ultrasound equipment signals during long-term continuous monitoring to adapt to individual physiological differences.

[0059] Example 2

[0060] like Figures 1-3 As shown, this embodiment provides a multimodal wearable monitoring system for atelectasis assessment, including a bioelectrical impedance measurement module, a wearable ultrasound monitoring module, a data acquisition and preliminary processing unit, a central processing unit, and a data transmission and remote monitoring module;

[0061] The bioelectrical impedance measurement module and the wearable ultrasound monitoring module are respectively connected to the data acquisition and preliminary processing unit, the data acquisition and preliminary processing unit is connected to the central processing unit, and the central processing unit is connected to the data transmission and remote monitoring module.

[0062] The bioelectrical impedance measurement module is used to acquire dynamic bioelectrical impedance signals in the pleural cavity; the wearable ultrasound monitoring module is used to acquire pleural sliding image signals; the data acquisition and preliminary processing unit is used for synchronous acquisition, preprocessing and preliminary fusion of dual-modal signals; the central processing unit is used for data storage, dynamic signal analysis and atelectasis determination and location; and the data transmission and remote monitoring module is used for data transmission and real-time alarm.

[0063] The bioelectrical impedance module and wearable ultrasound module respectively transmit the acquired raw signals to the central processing unit through the multi-channel data acquisition unit. The central processing unit stores and analyzes the data, and transmits the processing results to the remote monitoring terminal through the communication module, realizing comprehensive and real-time monitoring of atelectasis.

[0064] (1) Bioelectrical impedance measurement module

[0065] The bioelectrical impedance measurement module includes an electrode unit, a signal excitation and detection unit, a preliminary amplification and filtering module, and a self-calibration module.

[0066] The electrode unit includes two excitation electrodes and two detection electrodes, using medical-grade conductive patches to ensure good contact between the electrodes and the skin and signal stability during prolonged wear. The position and spacing of the electrodes are ergonomically optimized to balance signal quality and wearing comfort, and are attached to the left and right sides of the subject's chest. The signal excitation and detection unit applies alternating current and captures impedance change signals. It uses a precise current source to apply a small alternating current and uses a highly sensitive detection electrode to capture impedance changes caused by tissue and liquid media. It can capture the dynamic impedance spectrum of the subject's chest in real time, providing quantitative parameters for subsequent atelectasis detection. The preliminary amplification and filtering module performs low-noise amplification and multi-stage filtering on the acquired analog signal to remove environmental interference and human motion noise, ensuring signal authenticity and stability. The self-calibration module automatically detects and compensates for signal deviations through periodic hardware calibration and software dynamic calibration. The hardware calibration is achieved through a built-in precision resistor network, and the software dynamic calibration adjusts parameters based on the signal quality index.

[0067] The specific parameters for this module are as follows:

[0068] Excitation signal: frequency range 50kHz-100kHz (preferably 50kHz), constant current output current 100μA-500μA (preferably 200μA), total harmonic distortion < 1%.

[0069] Signal acquisition: Input impedance > 100MΩ, equivalent input noise voltage <1μV RMS (1Hz-10kHz).

[0070] ADC (Analog-to-Digital Converter): Resolution ≥ 16 bits, programmable sampling rate ranging from 1kSPS to 10kSPS to ensure capture of respiratory frequency (typically 0.1-0.5Hz) and its harmonics.

[0071] Common-mode rejection ratio: >100dB, to suppress power frequency interference.

[0072] (2) Wearable ultrasound monitoring module

[0073] The wearable ultrasound monitoring module includes an ultrasound probe unit and an ultrasound signal processing unit. The ultrasound probe unit adopts a multi-point layout design, is compact and lightweight, and is fixed to the chest, back, or side through a fit design (adjusted according to the subject's posture, such as prioritizing the chest when supine), ensuring good acoustic coupling effect and possessing high-resolution imaging capabilities to capture pleural sliding signals in real time. The ultrasound signal processing unit is used for receiving, amplifying, digitally processing, and extracting pleural sliding features from ultrasound signals. It has a built-in dedicated signal processing circuit and extracts pleural sliding features through algorithms such as Fast Fourier Transform, providing imaging evidence for locating atelectasis.

[0074] The ultrasonic probe unit operates in M-mode (motion mode), and the ultrasonic signal processing unit incorporates a digital signal processor for real-time envelope detection and noise reduction of the echo signal. Specific parameters of this module are as follows:

[0075] Ultrasonic probe: center frequency 2.0MHz-5.0MHz (preferably 2.5MHz), axial resolution < 1mm, lateral resolution < 2mm.

[0076] Imaging mode: M mode (motion mode), scan depth 5cm-15cm (adjustable), frame rate ≥30Hz to ensure continuous capture of pleural sliding.

[0077] Signal processing: Dynamic range > 100dB, built-in digital signal processor performs real-time envelope detection and noise reduction on echo signals.

[0078] (3) Data acquisition and preliminary processing unit

[0079] The data acquisition and preliminary processing unit includes a multi-channel data acquisition unit and a preprocessing and data fusion module. The multi-channel data acquisition unit supports simultaneous sampling of multiple analog signals, ensuring the time synchronization of dual-modal data. The multi-channel acquisition synchronization error is < 1ms, achieved through hardware trigger signals. The preprocessing and data fusion module uses an embedded processing chip (such as the ARM Cortex-M7 series) with a main frequency ≥ 300MHz to support real-time floating-point operations. It performs filtering and noise reduction on the data and uses time-domain and frequency-domain feature fusion technology to construct a joint data model for performing preliminary filtering and feature extraction.

[0080] (4) Central Processing Unit

[0081] The central processing unit includes a data storage and management module and a dynamic signal analysis module. The data storage and management module is used for caching and storing multimodal data to ensure data integrity and continuity. The dynamic signal analysis module is based on a multimodal data fusion algorithm to perform real-time analysis of bioelectrical impedance change curves and pleural sliding features in ultrasound images, completes risk assessment and localization of atelectasis, supports adaptive parameter adjustment, and can achieve fine detection for different body types and postures.

[0082] The central processing unit (CPU) core processor uses a high-performance application processor (such as the ARM Cortex-A series) and is equipped with no less than 1GB of RAM to run complex data fusion and classification algorithms.

[0083] The multimodal data fusion algorithm adopts a hybrid model that combines feature-level fusion and decision-level fusion, including data alignment, feature standardization, feature-level weighted fusion, and decision-level classification steps. The decision-level classification uses a support vector machine classifier to output a binary judgment of atelectasis and an abnormality score.

[0084] (5) Data transmission and remote monitoring module

[0085] The data transmission and remote monitoring module supports Bluetooth and / or Wi-Fi wireless communication protocols with a transmission delay of <100ms, ensuring the real-time nature of alarm information. It is used to transmit processing results to the remote monitoring terminal and trigger different levels of alarms based on the atelectasis determination results.

[0086] Example 3

[0087] like Figure 4 and Figure 5 As shown, this embodiment provides a multimodal wearable monitoring method for atelectasis assessment, using the multimodal wearable monitoring system for atelectasis assessment provided in Embodiment 1. The method includes the following steps:

[0088] Step 1: Bioelectrical impedance data collection and processing

[0089] Dynamic bioelectrical impedance signals within the pleural cavity were acquired using a bioelectrical impedance measurement module, while pleural sliding image signals were acquired using a wearable ultrasound monitoring module. The acquisition of dynamic bioelectrical impedance signals was achieved using a four-electrode technique, with an excitation electrode applying alternating current and a detection electrode capturing impedance changes. The acquisition of pleural sliding image signals was achieved using a multi-point ultrasound probe unit, tailored to the subject's posture. Details are as follows:

[0090] (1) Electrode attachment: Using 4-electrode technology, the outer excitation and inner detection electrodes are attached to the corresponding positions on the left and right chest.

[0091] (2) Signal preamplification: The excitation current passes through the chest tissue and generates a weak impedance change signal. After being processed by a low-noise amplifier and a multi-stage filter, environmental interference is removed to ensure signal stability.

[0092] (3) Signal digitization: High-precision ADC is used to digitize the analog signal to ensure that the sampling frequency matches the human breathing rhythm.

[0093] (4) Digital preprocessing: Time-domain filtering and FFT transformation are performed in the embedded processing unit to extract features such as amplitude, phase and spectrum, and a self-calibration mechanism is used to compensate for signal drift caused by long-term monitoring.

[0094] Specific preprocessing steps for bioelectrical impedance signals:

[0095] Noise reduction: First, a 5th-order Butterworth bandpass filter (passband 0.05Hz-5Hz) is used to filter out baseline drift and high-frequency electromyographic noise.

[0096] Feature extraction: The filtered signal Z(t) is subjected to a moving average to extract its envelope, and the ΔZ curve, which mainly reflects the changes in lung ventilation, is obtained.

[0097] Feature parameter calculation:

[0098] Average impedance change rate (Avg_dZ / dt):

[0099] Avg_dZ / dt=(1 / N)*Σ|dZ (t) / dt|

[0100] Where N is the number of sampling points in one respiratory cycle, and this value may decrease when the lungs are atelectasised.

[0101] Spectral Entropy:

[0102] Perform an FFT on the ΔZ curve to obtain the power spectrum P(f), and calculate the normalized spectral entropy.

[0103] H=-Σ[P_norm (f)*log2 (P_norm (f))],

[0104] A decrease in entropy may indicate that the breathing pattern has become regular or abnormal.

[0105] Self-calibration and parameter adjustment implementation methods:

[0106] Periodic hardware calibration: Every 30 minutes, the system switches the measurement channel to a built-in precision resistor network (e.g., 500Ω, 1kΩ, 2kΩ) via an analog switch to remeasure and update the ADC gain and offset error.

[0107] Software dynamic calibration (Signal Quality Index - SQI):

[0108] Judgment criterion: Real-time calculation of the SQI of the bioelectrical impedance signal, the formula is:

[0109] SQI=Signal_Power / (Noise_Power+Motion_Artifact_Power).

[0110] Among them, Signal_Power comes from the 0.1-0.5Hz frequency band, Noise_Power comes from the 50 / 60Hz power frequency and its harmonics, and Motion_Artifact_Power comes from the 1-10Hz frequency band.

[0111] Parameter adjustment: When the SQI is below a threshold (e.g., 3.0), the system determines that the signal quality is poor. At this time, the algorithm will: automatically increase the order or bandwidth of the digital filter; temporarily increase the sliding window size used for feature calculation to smooth noise; reduce the weight of bioelectrical impedance features in the fusion decision and rely more on the judgment results of ultrasound modality; and prompt medical staff through the terminal with "poor electrode contact or motion interference".

[0112] Step 2: Wearable ultrasound acquisition and processing

[0113] The data acquisition and preliminary processing unit synchronously acquires, preprocesses, and initially fuses the dual-modal signals to obtain preprocessed dual-modal feature data. Preprocessing includes: low-noise amplification, multi-level filtering, and digitization of the bioelectrical impedance signal to extract the average impedance change rate and spectral entropy characteristic parameters; denoising, contrast enhancement, and edge detection processing of the ultrasound image signal to extract the average pleural sliding velocity and the coefficient of variation of the pleural sliding amplitude characteristic parameters; and preliminary fusion is achieved through time-domain and frequency-domain feature fusion techniques to construct a dual-modal joint data model. Details are as follows:

[0114] (1) Select the attachment position according to the human body posture: select the front of the chest when lying supine, and select the back or side in other states to ensure the best collection of pleural sliding information.

[0115] (2) Image preprocessing: The original ultrasound image is subjected to noise reduction, contrast enhancement and edge detection to improve the recognition of pleural motion signals.

[0116] (3) Feature extraction: By comparing consecutive frames of images, dynamic tracking of pleural sliding is achieved, and key parameters such as displacement and velocity are extracted, including the average pleural sliding velocity (calculated by displacement between consecutive frames) and the coefficient of variation of pleural sliding amplitude.

[0117] Step 3: Multimodal data fusion

[0118] The central processing unit performs multimodal fusion analysis on the preprocessed bimodal feature data to determine the presence and location of atelectasis. The multimodal fusion analysis includes data alignment, feature standardization, weighted fusion, and classification. Data alignment is based on timestamps to achieve spatiotemporal matching of the bimodal data. Feature standardization uses the Z-score standardization method. Weighted fusion calculates a comprehensive anomaly score using a linear weighted model, with weights dynamically adjusted based on the signal quality index. Classification uses a support vector machine classifier to achieve binary judgment and localization of atelectasis. Details are as follows:

[0119] (1) Data alignment: Both bioelectrical impedance and ultrasound data are time-stamped and data alignment is achieved using a common sampling clock.

[0120] (2) Feature Standardization: The feature parameters of the two modal signals are normalized to eliminate dimensional differences. The Z-score standardization method is used.

[0121] x_norm=(x-μ) / σ,

[0122] Where μ and σ are the mean and standard deviation of this feature in the patient's resting state, respectively.

[0123] (3) Joint decision-making: The processed feature data are weighted and fused using a weighted fusion algorithm, and a hybrid support vector machine or random forest model is used for classification and judgment to comprehensively assess and locate the risk of atelectasis.

[0124] Specific steps of the fusion and decision-making algorithm:

[0125] Step 1: Feature-level fusion and anomaly probability calculation

[0126] The comprehensive anomaly score S_fusion is calculated using a linear weighted model.

[0127] S_fusion = w1F1_norm + w2F2_norm + w3F3_norm + w4F4_norm

[0128] The weights w1...w4 can be dynamically adjusted according to the Signal Quality Index (SQI) (e.g., when the SQI is low, w3 and w4 increase). F1_norm...F4_norm are the standardized eigenvalues ​​(note: the relationship between features and anomalies, for example, F1 may be negative in atelectasis, so the sign of the coefficients needs to be considered when weighting); F1 is the average impedance change rate, F2 is the spectral entropy, F3 is the average pleural sliding velocity, and F4 is the coefficient of variation of the pleural sliding amplitude.

[0129] Step 2: Decision-level fusion and localization (based on classifier)

[0130] The fused feature vectors [F1_norm, F2_norm, F3_norm, F4_norm] are input into a pre-trained support vector machine classifier.

[0131] Algorithm structure: SVM using radial basis functions as kernel functions.

[0132] Decision function: f(x) = sign(Σ(α_iy_iK(x_i,x)) + b)

[0133] Where x_i is the support vector, y_i is its class label (+1 for normal, -1 for atelectasis), α_i is the Lagrange multiplier, K(x_i,x)=exp(-γ*||x_i-x||²) is the RBF kernel function, and b is the bias term.

[0134] Output: The classifier not only outputs a binary judgment of "normal" or "atelectasis", but the absolute value of its decision function f(x) can also serve as a continuous score of the degree of abnormality. At the same time, the feature (F3, F4) of which ultrasound probe contributes the most indicates the most likely anatomical location of atelectasis (such as the right upper lobe or right lower lobe).

[0135] Step 4: Feedback and Adjustment Mechanism

[0136] The judgment results are transmitted to the remote monitoring terminal via the data transmission and remote monitoring module, and corresponding alarms are triggered according to the degree of abnormality. Alarms include Level 1 and Level 2 alarms. A Level 1 alarm (yellow warning) is triggered when the overall abnormal score exceeds the first threshold and the classification is atelectasis. A Level 2 alarm (red alarm) is triggered when the overall abnormal score exceeds the second threshold and the abnormal location is consistent, and is sent directly to the nurse station terminal via the network. Details are as follows:

[0137] (1) The system automatically adjusts the filtering and parameter settings based on continuous monitoring data to overcome errors caused by changes in electrode contact or attitude adjustment during long-term monitoring.

[0138] (2) The abnormal judgment result is fed back to the central processing unit, triggering alarms and subsequent data storage, forming a complete closed-loop data processing process.

[0139] Example 4

[0140] This example provides an application case for monitoring atelectasis in ICU postoperative patients.

[0141] 1. Application Scenarios

[0142] Patients undergoing general anesthesia will undergo continuous 48-hour lung monitoring to prevent and detect atelectasis at an early stage.

[0143] 2. Specific equipment and layout

[0144] (1) Bioelectrical impedance module:

[0145] Electrodes: Four disposable Ag / AgCl gel electrodes (3M™ Red Dot™) are used.

[0146] Equipment: The excitation and detection unit uses TI's ADS129x series biopotential analog front-end chip, paired with a precision voltage-controlled current source (Howland current pump circuit).

[0147] Attachment locations: The two excitation electrodes are attached to the lateral side of the 2nd intercostal space along the midclavicular line on the right chest and the 8th intercostal space along the midaxillary line on the left chest, respectively; the two detection electrodes are attached 3 cm inside the same vertical line.

[0148] (2) Wearable ultrasound module:

[0149] Probes: Two miniature linear array ultrasound probes (Butterfly Network® iQ+ probes) are used.

[0150] Equipment: The probe's built-in microcontroller is responsible for transmit / receive control and preliminary signal processing.

[0151] Attachment positions: One probe is fixed to the midclavicular line at the 4th intercostal space on the right chest for monitoring the upper lobe of the right lung; the other is fixed to the inferior angle of the scapula on the right back for monitoring the lower lobe of the right lung. The probes are secured with medical waterproof tape and coupling agent.

[0152] (3) Central Processing Unit:

[0153] Equipment: A ruggedized tablet or custom handheld terminal (Ruggon™ series) is used to run the system's analysis software.

[0154] 3. Signal Acquisition Process

[0155] (1) After the system is powered on, the bioelectrical impedance module first performs self-calibration: after the electrodes are attached, the impedance of a known 1kΩ precision calibration resistor is measured, compared with the theoretical value, and the system gain and offset correction coefficients are calculated and stored.

[0156] (2) The central processing unit issues a synchronous acquisition command, and the bioelectrical impedance module starts to acquire impedance data (Z(t)) at a sampling rate of 2kSPS, while the ultrasound module acquires M-mode images at a frame rate of 50Hz.

[0157] (3) All data is tagged with a unified timestamp and transmitted to the central processing unit in real time via Bluetooth 5.0.

[0158] 3. Monitoring Results and Feedback

[0159] (1) The central processing unit runs a multimodal data fusion algorithm in real time to analyze and process the collected data and generate an atelectasis risk assessment report every 5 minutes.

[0160] (2) When the overall abnormal score exceeds the first threshold (T1) and the SVM classifier judges it as "atelectasis", the system triggers a first-level alarm (yellow warning) and the terminal displays the warning information and the suspected abnormal area.

[0161] (3) When the overall abnormal score exceeds the second threshold (T2) and the abnormal location is consistent, the system triggers a level 2 alarm (red alarm), the terminal issues an audible and visual alarm, and sends the alarm information and related data to the nurse station terminal via Wi-Fi, and pushes it to the mobile device of the responsible doctor.

[0162] (4) During the monitoring process, the system automatically records all raw data, processing results and alarm information to form a complete monitoring file, which is convenient for doctors to review and analyze later.

[0163] Through the above specific implementation methods, the present invention utilizes a combination of bioelectrical impedance and wearable ultrasound multimodal data to achieve real-time monitoring and localization of dynamic changes in atelectasis in a non-invasive and portable manner.

[0164] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A multimodal wearable monitoring system for assessing atelectasis, characterized in that, It includes a bioelectrical impedance measurement module, a wearable ultrasound monitoring module, a data acquisition and preliminary processing unit, a central processing unit, and a data transmission and remote monitoring module; The bioelectrical impedance measurement module and the wearable ultrasound monitoring module are respectively connected to the data acquisition and preliminary processing unit, the data acquisition and preliminary processing unit is connected to the central processing unit, and the central processing unit is connected to the data transmission and remote monitoring module. The bioelectrical impedance measurement module is used to acquire dynamic bioelectrical impedance signals in the pleural cavity; the wearable ultrasound monitoring module is used to acquire pleural sliding image signals; the data acquisition and preliminary processing unit is used for synchronous acquisition, preprocessing and preliminary fusion of dual-modal signals; the central processing unit is used for data storage, dynamic signal analysis and atelectasis determination and location; and the data transmission and remote monitoring module is used for data transmission and real-time alarm.

2. The multimodal wearable monitoring system for atelectasis assessment as described in claim 1, characterized in that, The bioelectrical impedance measurement module includes an electrode unit, a signal excitation and detection unit, and a preliminary amplification and filtering module. The electrode unit includes two excitation electrodes and two detection electrodes, which are attached to the left and right sides of the subject's chest. The signal excitation and detection unit is used to apply alternating current and capture impedance change signals. The preliminary amplification and filtering module is used to perform low-noise amplification and multi-stage filtering on the acquired analog signals.

3. The multimodal wearable monitoring system for atelectasis assessment as described in claim 2, characterized in that, The bioelectrical impedance measurement module also includes a self-calibration module, which automatically detects and compensates for signal deviations through periodic hardware calibration and dynamic software calibration. The hardware calibration is achieved through a built-in precision resistor network, and the dynamic software calibration adjusts parameters based on the signal quality index.

4. The multimodal wearable monitoring system for atelectasis assessment as described in claim 1, characterized in that, The wearable ultrasound monitoring module includes an ultrasound probe unit and an ultrasound signal processing unit; the ultrasound probe unit adopts a multi-point layout design and is attached to the chest, back or side according to the subject's posture; the ultrasound signal processing unit is used for receiving, amplifying, digitally processing and extracting pleural sliding features of ultrasound signals.

5. The multimodal wearable monitoring system for atelectasis assessment as described in claim 4, characterized in that, The imaging mode of the ultrasonic probe unit is M mode, and the ultrasonic signal processing unit has a built-in digital signal processor for real-time envelope detection and noise reduction of the echo signal.

6. The multimodal wearable monitoring system for atelectasis assessment as described in claim 1, characterized in that, The data acquisition and preliminary processing unit includes a multi-channel data acquisition unit and a preprocessing and data fusion module. The multi-channel data acquisition unit supports simultaneous sampling of multiple analog signals to ensure the time synchronization of dual-modal data. The preprocessing and data fusion module uses an embedded processing chip to filter and reduce noise in the data, and uses time-domain and frequency-domain feature fusion technology to construct a joint data model.

7. The multimodal wearable monitoring system for atelectasis assessment as described in claim 1, characterized in that, The central processing unit includes a data storage and management module and a dynamic signal analysis module. The data storage and management module is used for caching and storing multimodal data, and the dynamic signal analysis module, based on a multimodal data fusion algorithm, performs real-time analysis of bioelectrical impedance change curves and pleural sliding features in ultrasound images to complete the risk assessment and localization of atelectasis.

8. The multimodal wearable monitoring system for atelectasis assessment as described in claim 7, characterized in that, The multimodal data fusion algorithm adopts a hybrid model that combines feature-level fusion and decision-level fusion, including data alignment, feature standardization, feature-level weighted fusion, and decision-level classification steps; the decision-level classification uses a support vector machine classifier to output a binary judgment of atelectasis and an abnormality score.

9. The multimodal wearable monitoring system for atelectasis assessment as described in claim 1, characterized in that, The data transmission and remote monitoring module supports Bluetooth and / or Wi-Fi wireless communication protocols, which are used to transmit the processing results to the remote monitoring terminal and trigger different levels of alarms based on the atelectasis determination results.

10. A multimodal wearable monitoring method for assessing atelectasis, characterized in that, The method includes the following steps: Step 1: Collect dynamic bioelectrical impedance signals in the pleural cavity using a bioelectrical impedance measurement module, and collect pleural sliding image signals using a wearable ultrasound monitoring module. Step 2: The dual-modal signals are synchronously acquired, preprocessed, and initially fused through the data acquisition and preliminary processing unit to obtain preprocessed dual-modal feature data; Step 3: The central processing unit performs multimodal fusion analysis on the preprocessed dual-modal feature data to determine whether atelectasis and abnormal locations exist. Step 4: Transmit the judgment result to the remote monitoring terminal through the data transmission and remote monitoring module, and trigger the corresponding alarm according to the degree of abnormality.

11. The multimodal wearable monitoring method for atelectasis assessment as described in claim 10, characterized in that, In step 1, the acquisition of bioelectrical impedance dynamic signals is achieved through a 4-electrode technique, with an excitation electrode applying alternating current and a detection electrode capturing impedance change signals; the acquisition of pleural sliding image signals is achieved through a multi-point layout of ultrasound probe units according to the subject's posture.

12. The multimodal wearable monitoring method for atelectasis assessment as described in claim 10, characterized in that, The preprocessing in step 2 includes: performing low-noise amplification, multi-level filtering, and digital processing on the bioelectrical impedance signal to extract the average impedance change rate and spectral entropy characteristic parameters; and performing denoising, contrast enhancement, and edge detection processing on the ultrasound image signal to extract the average pleural sliding velocity and the coefficient of variation characteristic parameters of the pleural sliding amplitude.

13. The multimodal wearable monitoring method for atelectasis assessment as described in claim 10, characterized in that, The initial fusion in step 2 is achieved through time-domain and frequency-domain feature fusion technology to construct a dual-modal joint data model; the multimodal fusion analysis in step 3 includes data alignment, feature standardization, weighted fusion and classification determination steps.

14. The multimodal wearable monitoring method for atelectasis assessment as described in claim 13, characterized in that, The data alignment is based on timestamps to achieve spatiotemporal matching of dual-modal data; the feature standardization adopts the Z-score standardization method; the weighted fusion calculates the comprehensive anomaly score through a linear weighted model, and the weights are dynamically adjusted according to the signal quality index; the classification judgment is achieved by a support vector machine classifier to realize the binary judgment and localization of atelectasis.

15. The multimodal wearable monitoring method for atelectasis assessment as described in claim 10, characterized in that, The alarms in step 4 include a level 1 alarm and a level 2 alarm. A level 1 alarm is triggered when the overall abnormal score exceeds the first threshold and the classification is atelectasis; a level 2 alarm is triggered when the overall abnormal score exceeds the second threshold and the abnormal location is consistent.

Images