Wearable urodynamics monitoring method and system

Abstract

The invention provides a wearable urodynamics monitoring method and system, and the method comprises the steps: collecting various urodynamics signals for a patient, extracting the time-frequency domain features of the urodynamics signals from the various urodynamics signals, carrying out the feature fusion of the extracted time-frequency domain features of the urodynamics signals, and obtaining a fused feature vector F0; and inputting the fused feature vector F0 into an improved convolutional neural network for identification to obtain a health index.

Description

Technical Field

[0001] The present invention relates to the field of medical imaging, and in particular to a wearable urodynamic monitoring method and system. Background Art

[0002] In recent years, with the rapid development of medical technology, urodynamic examination has become an indispensable part of the field of urinary system evaluation, and its importance has become increasingly prominent. This examination method not only deeply explores the complex physiological mechanisms of the urinary system, but also greatly improves our diagnostic accuracy and scientific evaluation of treatment effects for various urinary dysfunction diseases. As a comprehensive evaluation method, urodynamic examination provides doctors with detailed and intuitive pathophysiological information by comprehensively analyzing the various physiological parameters involved in the urination process.

[0003] Specifically, urodynamic examination covers multiple dimensions of measurement and analysis, including but not limited to electromyography (EMG) recording, urethral pressure, bladder pressure, abdominal pressure, rectal pressure, and the most intuitive urinary flow rate (Uroflowmetry) and other key indicators. The precise measurement of these parameters requires the operator to have deep professional knowledge and rich clinical experience to accurately interpret complex physiological signals and make scientific judgments based on them.

[0004] However, in the face of growing clinical needs and patients' expectations for higher quality medical services, the traditional urodynamic examination method that relies on doctors' subjective judgment faces many challenges. Therefore, a new technical solution is needed to solve the above problems. Summary of the invention

[0005] The purpose of the present invention is to solve the above-mentioned problems in the prior art and provide a wearable urodynamic monitoring method and system to make the dynamic evaluation results more objective.

[0006] The present invention is achieved through the following technical solutions:

[0007] The present invention provides a wearable urodynamic monitoring method, which comprises the following steps:

[0008] S1: Data collection: Collect various urodynamic signals from patients;

[0009] S2: time-frequency domain feature extraction: extracting the time-frequency domain features of the urodynamic signals from the multiple urodynamic signals collected in step S1 respectively;

[0010] S3: feature fusion: fusing the time-frequency domain features of the urodynamic signal extracted in step S2 to obtain a fused feature vector F0;

[0011] S4: Obtain monitoring results: Input the fused feature vector F0 into the improved convolutional neural network for recognition to obtain a health index.

[0012] Furthermore, the multiple urodynamic signals include: electromyographic signals, urethral pressure signals, bladder pressure signals, abdominal pressure signals, rectal pressure signals, and urine volume curves.

[0013] Furthermore, the time-frequency domain feature extraction includes: acquiring the time-frequency domain features of the urodynamic signal based on the Mallat algorithm.

[0014] Furthermore, the improved convolutional neural network includes a sequentially connected TFF extraction unit, a convolution unit, a TFF extraction unit, a convolution unit, a TFF extraction unit, a fully connected layer, and a softmax layer;

[0015] The TFF extraction unit is specifically:

[0016] The input feature of the TFF extraction unit is denoted as Fin, and the output is denoted as Fout;

[0017] Fout = Conv(Fps);

[0018]

[0019] The CAM() represents a channel attention module, and Conv() represents a convolution operation;

[0020] Where, Fp = FFM (Fc); Fs = TFM (Fc);

[0021] FFM() represents the frequency feature module, and TFM() represents the time feature module;

[0022] The frequency feature module specifically inputs the Fc into the average pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fp;

[0023] The time feature module specifically inputs the Fc into the maximum pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fs;

[0024] Among them, Fps, Fp, Fs, and Fc all represent characteristics.

[0025] Furthermore, the convolution unit includes a convolution layer, a normalization layer, and a ReLU layer connected sequentially.

[0026] Furthermore, the present invention provides a wearable urodynamic monitoring system, the system comprising the following units:

[0027] Data acquisition unit: collects a variety of urodynamic signals from patients;

[0028] A time-frequency domain feature extraction unit is used to extract the time-frequency domain features of the urodynamic signals from the various urodynamic signals collected by the data collection unit;

[0029] Feature fusion unit: fuses the time-frequency domain features of the urodynamic signal extracted by the time-frequency domain feature extraction unit to obtain a fused feature vector F0;

[0030] The monitoring result acquisition unit: inputs the fused feature vector F0 into the improved convolutional neural network for recognition to obtain the health index.

[0031] Furthermore, the multiple urodynamic signals include: electromyographic signals, urethral pressure signals, bladder pressure signals, abdominal pressure signals, rectal pressure signals, and urine volume curves.

[0032] Furthermore, the time-frequency domain feature extraction includes: acquiring the time-frequency domain features of the urodynamic signal based on the Mallat algorithm.

[0033] Furthermore, the improved convolutional neural network includes a sequentially connected TFF extraction unit, a convolution unit, a TFF extraction unit, a convolution unit, a TFF extraction unit, a fully connected layer, and a softmax layer;

[0034] The TFF extraction unit is specifically:

[0035] The input feature of the TFF extraction unit is denoted as Fin, and the output is denoted as Fout;

[0036] Fout = Conv(Fps);

[0037]

[0038] The CAM() represents a channel attention module, and Conv() represents a convolution operation;

[0039] Where, Fp = FFM (Fc); Fs = TFM (Fc);

[0040] FFM() represents the frequency feature module, and TFM() represents the time feature module;

[0041] The frequency feature module specifically inputs the Fc into the average pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fp;

[0042] The time feature module specifically inputs the Fc into the maximum pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fs;

[0043] Among them, Fps, Fp, Fs, and Fc all represent characteristics.

[0044] Furthermore, the convolution unit includes a convolution layer, a normalization layer, and a ReLU layer connected sequentially.

[0045] Compared with the prior art, the advantages of the present invention are as follows:

[0046] The present invention collects multiple urodynamic signals from patients, extracts time-frequency domain features of the urodynamic signals, performs feature fusion on the extracted time-frequency domain features of the urodynamic signals to obtain a fused feature vector F0, and inputs the fused feature vector F0 into an improved convolutional neural network for recognition to obtain a health index as a reference for doctors' diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the specific implementation of the present invention or the technical solution in the prior art, the drawings required for describing the specific implementation or the prior art will be briefly introduced below. Obviously, the drawings described below are only one implementation of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 is a flow chart of a wearable urodynamic monitoring method;

[0049] Figure 2 This is the structural diagram of the improved convolutional neural network. DETAILED DESCRIPTION

[0050] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the technical scheme in the specific implementation mode of the present invention is clearly and completely described below to further illustrate the present invention. Obviously, the specific implementation mode described is only a part of the implementation mode of the present invention, rather than all styles.

[0051] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0052] As attached Figure 1 As shown, the present invention provides a wearable urodynamic monitoring method, the method comprising the following steps:

[0053] S1: Data collection: Collect various urodynamic signals from patients;

[0054] S2: time-frequency domain feature extraction: extracting the time-frequency domain features of the urodynamic signals from the multiple urodynamic signals collected in step S1 respectively;

[0055] S3: feature fusion: fusing the time-frequency domain features of the urodynamic signal extracted in step S2 to obtain a fused feature vector F0;

[0056] S4: Obtain monitoring results: Input the fused feature vector F0 into the improved convolutional neural network for recognition to obtain a health index.

[0057] Furthermore, the multiple urodynamic signals include: electromyographic signals, urethral pressure signals, bladder pressure signals, abdominal pressure signals, rectal pressure signals, and urine volume curves.

[0058] Furthermore, the time-frequency domain feature extraction includes: acquiring the time-frequency domain features of the urodynamic signal based on the Mallat algorithm.

[0059] Further, as attached Figure 1 As shown, the improved convolutional neural network includes a sequentially connected TFF extraction unit, a convolution unit, a TFF extraction unit, a convolution unit, a TFF extraction unit, a fully connected layer, and a softmax layer;

[0060] The TFF extraction unit is specifically:

[0061] The input feature of the TFF extraction unit is denoted as Fin, and the output is denoted as Fout;

[0062] Fout = Conv(Fps);

[0063]

[0064] The CAM() represents a channel attention module, and Conv() represents a convolution operation;

[0065] Channel Attention Mechanism (CAM) uses the relationship between features to achieve channel attention. The idea of ​​channel attention is:

[0066] First, perform global maximum pooling and global average pooling on an input feature map F of size H×W×C in the spatial dimension to obtain two 1×1×C feature maps; then, send the results of global maximum pooling and global average pooling to a shared multi-layer perceptron (MLP) for learning, and obtain two 1×1×C feature maps. The number of neurons in the first layer of MLP is C / r, the activation function is Relu, and the number of neurons in the second layer is C; finally, perform the Add operation on the output of MLP, and then perform the mapping process of the Sigmoid activation function to finally obtain the channel attention weight matrix.

[0067] Where, Fp = FFM (Fc); Fs = TFM (Fc);

[0068] FFM() represents the frequency feature module, and TFM() represents the time feature module;

[0069] The frequency feature module specifically inputs the Fc into the average pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fp;

[0070] The time feature module specifically inputs the Fc into the maximum pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fs;

[0071] Among them, Fps, Fp, Fs, and Fc all represent characteristics.

[0072] Furthermore, the convolution unit includes a convolution layer, a normalization layer, and a ReLU layer connected sequentially.

[0073] Furthermore, the present invention provides a wearable urodynamic monitoring system, the system comprising the following units:

[0074] Data acquisition unit: collects a variety of urodynamic signals from patients;

[0075] A time-frequency domain feature extraction unit is used to extract the time-frequency domain features of the urodynamic signals from the various urodynamic signals collected by the data collection unit;

[0076] Feature fusion unit: fuses the time-frequency domain features of the urodynamic signal extracted by the time-frequency domain feature extraction unit to obtain a fused feature vector F0;

[0077] The monitoring result acquisition unit: inputs the fused feature vector F0 into the improved convolutional neural network for recognition to obtain the health index.

[0078] Furthermore, the multiple urodynamic signals include: electromyographic signals, urethral pressure signals, bladder pressure signals, abdominal pressure signals, rectal pressure signals, and urine volume curves.

[0079] Furthermore, the time-frequency domain feature extraction includes: acquiring the time-frequency domain features of the urodynamic signal based on the Mallat algorithm.

[0080] Furthermore, the improved convolutional neural network includes a sequentially connected TFF extraction unit, a convolution unit, a TFF extraction unit, a convolution unit, a TFF extraction unit, a fully connected layer, and a softmax layer;

[0081] The TFF extraction unit is specifically:

[0082] The input feature of the TFF extraction unit is denoted as Fin, and the output is denoted as Fout;

[0083] Fout = Conv(Fps);

[0084] The specific feature fusion is:

[0085]

[0086] The CAM() represents a channel attention module, and Conv() represents a convolution operation;

[0087] Where, Fp = FFM (Fc); Fs = TFM (Fc);

[0088] FFM() represents the frequency feature module, and TFM() represents the time feature module;

[0089] The frequency feature module specifically inputs the Fc into the average pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fp;

[0090] The time feature module specifically inputs the Fc into the maximum pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fs;

[0091] Among them, Fps, Fp, Fs, and Fc all represent characteristics.

[0092] Furthermore, the convolution unit includes a convolution layer, a normalization layer, and a ReLU layer connected sequentially.

[0093] The above describes the main technical features and basic principles of the present invention and the related advantages. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the concept or basic features of the present invention. Therefore, no matter from which point of view, the above specific embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes that fall within the meaning and scope of the equivalent elements of the claims are included in the present invention.

[0094] In addition, it should be understood that although the present specification is described according to various implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each implementation method may also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A wearable urodynamic monitoring method, characterized in that: The method comprises the following steps: S1: Data collection: Collect various urodynamic signals from patients; S2: time-frequency domain feature extraction: extracting the time-frequency domain features of the urodynamic signals from the multiple urodynamic signals collected in step S1 respectively; S3: feature fusion: fusing the time-frequency domain features of the urodynamic signal extracted in step S2 to obtain a fused feature vector F0; S4: Obtain monitoring results: Input the fused feature vector F0 into the improved convolutional neural network for recognition to obtain a health index.

2. A wearable urodynamic monitoring method according to claim 1, characterized in that: The multiple urodynamic signals include: electromyographic signals, urethral pressure signals, bladder pressure signals, abdominal pressure signals, rectal pressure signals, and urine volume curves.

3. A wearable urodynamic monitoring method according to claim 1, characterized in that: The time-frequency domain feature extraction includes: obtaining the time-frequency domain features of the urodynamic signal based on the Mallat algorithm.

4. A wearable urodynamic monitoring method according to claim 1, characterized in that: The improved convolutional neural network includes a sequentially connected TFF extraction unit, a convolution unit, a TFF extraction unit, a convolution unit, a TFF extraction unit, a fully connected layer, and a softmax layer, and outputs a health index; The TFF extraction unit is specifically: The input feature of the TFF extraction unit is denoted as Fin, and the output is denoted as Fout; Fout = Conv(Fps); The CAM() represents a channel attention module, and Conv() represents a convolution operation; Where, Fp = FFM (Fc); Fs = TFM (Fc); FFM() represents the frequency feature module, and TFM() represents the time feature module; The frequency feature module specifically inputs the Fc into the average pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fp; The time feature module specifically inputs the Fc into the maximum pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fs; Among them, Fps, Fp, Fs, and Fc all represent features.

5. A wearable urodynamic monitoring method according to claim 4, characterized in that: The convolution unit consists of a sequentially connected convolution layer, a normalization layer, and a ReLU layer.

6. A wearable urodynamic monitoring system, characterized in that: The system comprises the following units: Data acquisition unit: collects a variety of urodynamic signals from patients; A time-frequency domain feature extraction unit is used to extract the time-frequency domain features of the urodynamic signals from the various urodynamic signals collected by the data collection unit; Feature fusion unit: fuses the time-frequency domain features of the urodynamic signal extracted by the time-frequency domain feature extraction unit to obtain a fused feature vector F0; The monitoring result acquisition unit: inputs the fused feature vector F0 into the improved convolutional neural network for recognition to obtain the health index.

7. A wearable urodynamic monitoring system according to claim 6, characterized in that: The multiple urodynamic signals include: electromyographic signals, urethral pressure signals, bladder pressure signals, abdominal pressure signals, rectal pressure signals, and urine volume curves.

8. A wearable urodynamic monitoring system according to claim 6, characterized in that: The time-frequency domain feature extraction includes: obtaining the time-frequency domain features of the urodynamic signal based on the Mallat algorithm.

9. A wearable urodynamic monitoring system according to claim 6, characterized in that: The improved convolutional neural network includes a sequentially connected TFF extraction unit, a convolution unit, a TFF extraction unit, a convolution unit, a TFF extraction unit, a fully connected layer, and a softmax layer; The TFF extraction unit is specifically: The input feature of the TFF extraction unit is denoted as Fin, and the output is denoted as Fout; Fout = Conv(Fps); The CAM() represents a channel attention module, and Conv() represents a convolution operation; Where, Fp = FFM (Fc); Fs = TFM (Fc); FFM() represents the frequency feature module, and TFM() represents the time feature module; The frequency feature module specifically inputs the Fc into the average pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fp; The time feature module specifically inputs the Fc into the maximum pooling layer, Conv layer, Conv layer, Conv layer, and sigmoid layer connected in sequence to obtain the feature Fs; Among them, Fps, Fp, Fs, and Fc all represent characteristics.

10. A wearable urodynamic monitoring system according to claim 9, characterized in that: The convolution unit consists of a sequentially connected convolution layer, a normalization layer, and a ReLU layer.

Images