Prostatitis ultrasonic drug therapy planning system, method and terminal

Abstract

The invention provides a prostatitis ultrasonic drug therapy planning system, a prostatitis ultrasonic drug therapy planning method and a prostatitis ultrasonic drug therapy planning terminal, and the prostatitis ultrasonic drug therapy planning method comprises the steps: extracting a multivariate prostatitis discrimination feature tensor from a collected photoacoustic RF signal, and then discriminating whether prostatitis has inflammations or not and the degree of the inflammations according to the multivariate prostatitis discrimination feature tensor. Once the inflammation is judged to exist, the system can immediately generate an inflammation degree judgment result, and a personalized inflammation treatment scheme is customized according to the result and the feature tensor. According to the method, the feature tensor in the opto-acoustic RF signal is extracted through an algorithm, accurate and intelligent evaluation of the prostatitis degree of the patient can be realized on the basis of the constructed discrimination network and decision network, and an ultrasonic-drug collaborative treatment planning scheme is output in a personalized manner, so that the drug curative effect is maximized, and the treatment period is effectively shortened.

Description

Technical Field

[0001] The present invention relates to the technical field of prostatitis treatment, and in particular to a prostatitis ultrasonic drug treatment planning system, method and terminal. Background Art

[0002] Chronic prostatitis is the most common urogenital disease in men. It is prone to cause infertility in young and middle-aged men and prostate cancer in elderly men. In addition, the recurrent symptoms of chronic prostatitis, such as pelvic pain, dysuria and sexual dysfunction, seriously affect the patient's quality of life and even cause mental problems.

[0003] Antibiotics are the main treatment for chronic prostatitis. However, the blood-prostate barrier structure in the prostate tissue prevents microorganisms in the prostate ducts from entering the blood circulation, while also preventing antibiotics from circulating through the blood to the prostate ducts to exert their therapeutic effects. Studies have reported that the concentration of antibiotics in the prostate ducts is only 12%-29% of that in the blood. Therefore, even with high-dose, long-course antibiotic treatment, only 47% to 56% of patients can improve their symptoms, and about 50% of patients relapse within a short period of time after stopping the medication due to the inability to completely eliminate the microorganisms in the ducts, which imposes a great burden on patients and the medical system.

[0004] Existing research results show that low-dose ultrasound irradiation can improve the permeability of biological barriers and is expected to increase the drug concentration in the prostate gland ducts. The continuous wave and pulse wave ultrasound treatment devices currently used achieve the purpose of treatment by irradiating the prostate area as a whole to increase the concentration of antibiotics in the prostate gland ducts. However, clinical studies have found that the distribution of inflammatory tissues in chronic prostatitis is uneven, and the severity of inflammation in inflammatory tissues is also different, resulting in the coexistence of normal prostate and different degrees of inflammatory tissues, which brings great obstacles to the control of ultrasound irradiation dose, that is, excessive doses will damage or even destroy the normal blood-prostate barrier, causing it to lose its beneficial barrier function; if the irradiation dose is insufficient, this biological barrier cannot be opened to increase the concentration of antibiotics in the gland ducts and affect the efficacy, especially for moderate to severe inflammation, because the firmness of the blood-prostate barrier is proportional to the degree of inflammation. Therefore, it is very important to judge the degree of prostate inflammation in the treatment of prostatitis. The existing technology lacks a high-precision intelligent way to judge the severity of prostatitis and realize intelligent output of treatment planning schemes. Summary of the invention

[0005] In view of the shortcomings of the prior art mentioned above, the purpose of the present invention is to provide a prostatitis ultrasonic drug treatment planning system, method and terminal, which are used to solve the technical problem that the prior art lacks a high-precision intelligent way to determine the severity of prostatitis and realize intelligent output of treatment planning solutions.

[0006] To achieve the above-mentioned purpose and other related purposes, the present invention provides a prostatitis ultrasonic drug treatment planning system, the system comprising: a photoacoustic RF signal acquisition module, used to acquire the collected photoacoustic RF signal; a feature extraction module, connected to the photoacoustic RF signal acquisition module, used to extract a multivariate prostate inflammation discrimination feature tensor from the photoacoustic RF signal; a discrimination network module, connected to the feature extraction module, used to discriminate whether the prostate is inflammatory and the degree of inflammation based on the multivariate prostate inflammation discrimination feature tensor, and obtain an inflammatory degree discrimination result if it is judged to be inflammatory; a decision network module, connected to the feature extraction module and the discrimination network module, used to obtain an inflammation treatment plan based on the inflammatory degree discrimination result and the prostate inflammation discrimination feature tensor.

[0007] In one embodiment of the present invention, the feature extraction module is used to extract five types of prostate inflammation discrimination feature tensors from the photoacoustic RF signal based on the feature tensor extraction algorithm; wherein the five types of prostate inflammation discrimination feature tensors include: a first feature tensor PAABTF, a second feature tensor PAAIB, a third feature tensor PAFIAB, a fourth feature tensor PAFSAB and a fifth feature tensor PASCS.

[0008] In one embodiment of the present invention, the algorithm for extracting characteristic tensors includes: a first characteristic tensor PAABTF extraction algorithm, including: using the periodogram method to perform classical power spectral density estimation on the photoacoustic RF signal, and then extracting the first characteristic tensor PAABTF; a second characteristic tensor PAAIB extraction algorithm, including: calculating the integral of the first characteristic tensor PAABTF within the effective frequency band, and extracting the second characteristic tensor PAAIB; a third characteristic tensor PAAIB extraction algorithm, including: extrapolating the best fitting straight line of the second characteristic tensor PAAIB within the set frequency range to the value at zero frequency, and extracting the third characteristic tensor PAFIAB; a fourth characteristic tensor PAFSAB extraction algorithm, including: extracting the fourth characteristic tensor PAFSAB from the value of the slope of the best fitting straight line of the second characteristic tensor PAAIB within the set frequency range; a fifth characteristic tensor PASCS extraction algorithm, including: obtaining the fifth characteristic tensor PASCS based on the photoacoustic RF signal.

[0009] In one embodiment of the present invention, the discriminant network module adopts a three-classification deep neural network based on a multi-stream fusion architecture, which is used to input five types of prostate inflammation discriminant feature tensors and output three-classification results corresponding to mild inflammation, moderate inflammation, and severe inflammation.

[0010] In one embodiment of the present invention, the three-classification deep neural network based on the multi-stream fusion architecture includes: a parameter tensor normalization module, which is used to standardize the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS respectively by using the standard deviation method; a multi-stream pyramid parameter tensor distillation and fusion module, connected to the parameter tensor normalization module, which is used to standardize the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS. The feature tensor PASCS shrinks the dimensions of the pyramid structure on each data flow line to perform distillation and selection of the data tensor, and while performing distillation and selection, multiple data streams are fused on different tower layers to obtain fused feature tensors corresponding to the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS respectively; a multi-layer perceptron module is connected to the multi-stream pyramid parameter tensor distillation and fusion module to merge the fused feature tensors and classify them to obtain the prediction probabilities of three classification results of mild inflammation, moderate inflammation and severe inflammation.

[0011] In one embodiment of the present invention, the decision network module is an iterative linear regression deep neural network based on multimodal data fusion, which is used to output a corresponding prostate inflammation treatment plan based on the input data of three modalities; wherein the input data of three modalities include: five types of prostate inflammation discrimination feature tensors, classification results and original treatment plans.

[0012] In one embodiment of the present invention, the iterative linear regression deep neural network based on multimodal data fusion includes: a parameter tensor standardization module, which is used to standardize the data of the three modalities using the maximum and minimum method; a heterogeneous multimodal tensor feature alignment and extraction module, which is connected to the parameter tensor standardization module and is used to perform feature tensor dimension alignment and data feature extraction on the standardized data of the three modalities using their respective corresponding heterogeneous networks to obtain feature data corresponding to the data of the three modalities; a multimodal fusion module, which is connected to the heterogeneous multimodal tensor feature alignment and extraction module, and is used to perform preliminary fusion of the feature data of the three modalities using a tensor splicing method, and then use a linear layer cascade softmax activation function to perform weight fusion to obtain a multimodal fusion tensor; a multilayer perceptron module, which is used to perform linear regression on the fused multimodal fusion tensor to iteratively update the treatment plan tensor; a tensor inverse standardization module, which is used to obtain an inflammation treatment plan by performing an inverse standardization operation on the iteratively updated treatment plan tensor.

[0013] In one embodiment of the present invention, the inflammation treatment plan includes: ultrasonic acoustic parameters, duration of stimulation mode, administration method and drug dosage.

[0014] To achieve the above-mentioned purpose and other related purposes, the present invention provides a method for planning ultrasonic drug treatment for prostatitis, the method comprising: acquiring collected photoacoustic RF signals; extracting a multivariate prostate inflammation discrimination feature tensor from the photoacoustic RF signals; judging whether the prostate is inflammatory and the degree of inflammation based on the multivariate prostate inflammation discrimination feature tensor, and obtaining an inflammation degree discrimination result if it is judged to be inflammatory; and obtaining an inflammation treatment plan based on the inflammation degree discrimination result and the prostate inflammation discrimination feature tensor.

[0015] To achieve the above-mentioned purpose and other related purposes, the present invention provides an electronic terminal, comprising: one or more memories and one or more processors; the one or more memories are used to store computer programs; the one or more processors are connected to the memories and are used to run the computer program to execute the prostatitis ultrasonic drug treatment planning system.

[0016] As described above, the present invention is a system, method and terminal for planning ultrasonic drug treatment for prostatitis, which has the following beneficial effects: the present invention extracts a multivariate prostate inflammation discrimination feature tensor from the collected photoacoustic RF signal, and then discriminates whether the prostate is inflammatory and the degree of inflammation based on the multivariate prostate inflammation discrimination feature tensor. Once it is determined that inflammation exists, the system will immediately generate a discrimination result of the degree of inflammation, and customize a personalized inflammation treatment plan based on this result and the feature tensor. The present invention extracts the feature tensor in the photoacoustic RF signal through an algorithm, and then based on the constructed discrimination network and decision network, it can realize accurate and intelligent assessment of the degree of prostate inflammation in patients, and outputs a personalized ultrasound-drug synergistic treatment planning plan, aiming to maximize the efficacy of the drug, thereby effectively shortening the treatment cycle. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 Shown is a structural schematic diagram of a prostatitis ultrasound drug treatment planning system in one embodiment of the present invention.

[0018] Figure 2 Shown is a schematic diagram of the structure of a three-classification deep neural network based on a multi-stream fusion architecture in one embodiment of the present invention.

[0019] Figure 3 Shown is a schematic diagram of the structure of an iterative linear regression deep neural network based on multimodal data fusion in one embodiment of the present invention.

[0020] Figure 4 Shown is a flow chart of a method for planning ultrasonic drug treatment for prostatitis in one embodiment of the present invention.

[0021] Figure 5 Shown is a schematic structural diagram of an electronic terminal in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0023] It should be noted that in the following description, reference is made to the accompanying drawings, which describe several embodiments of the present invention. It should be understood that other embodiments may also be used, and that mechanical composition, structural, electrical and operational changes may be made without departing from the spirit and scope of the present invention. The following detailed description should not be considered restrictive, and the scope of the embodiments of the present invention is limited only by the claims of the published patents. The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. Spatially related terms, such as "upper", "lower", "left", "right", "below", "below", "lower", "above", "upper", etc., may be used in the text to facilitate the description of the relationship between an element or feature shown in the figure and another element or feature.

[0024] Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case of "direct connection" but also the case of "indirect connection" by placing other elements therebetween. In addition, when a part is said to "include" a certain constituent element, unless otherwise stated, it does not exclude other constituent elements, but means that other constituent elements may be included.

[0025] The terms first, second and third mentioned herein are used to describe various parts, components, regions, layers and / or segments, but are not limited thereto. These terms are only used to distinguish a certain part, component, region, layer or segment from other parts, components, regions, layers or segments. Therefore, the first part, component, region, layer or segment described below may refer to the second part, component, region, layer or segment within the scope of the present invention.

[0026] Furthermore, as used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless there is an indication to the contrary in the context. It should be further understood that the terms "comprise", "include" indicate the presence of the described features, operations, elements, components, items, kinds, and / or groups, but do not exclude the presence, occurrence or addition of one or more other features, operations, elements, components, items, kinds, and / or groups. The terms "or" and "and / or" used herein are interpreted as inclusive, or mean any one or any combination. Therefore, "A, B or C" or "A, B and / or C" means "any of the following: A; B; C; A and B; A and C; B and C; A, B and C". Exceptions to this definition will only occur when the combination of elements, functions or operations is inherently mutually exclusive in some way.

[0027] The present invention provides a prostatitis ultrasound drug treatment planning system, which extracts a multivariate prostate inflammation discrimination feature tensor from the collected photoacoustic RF signal, and then discriminates whether the prostate is inflammatory and the degree of inflammation based on the multivariate prostate inflammation discrimination feature tensor. Once it is determined that inflammation exists, the system will immediately generate a discrimination result of the degree of inflammation, and customize a personalized inflammation treatment plan based on this result and the feature tensor. The present invention extracts the feature tensor in the photoacoustic RF signal through an algorithm, and then based on the constructed discrimination network and decision network, it can achieve accurate and intelligent evaluation of the patient's prostate inflammation degree, and outputs a personalized ultrasound-drug collaborative treatment planning plan, aiming to maximize the efficacy of the drug, thereby effectively shortening the treatment cycle.

[0028] The following is a detailed description of the embodiments of the present invention with reference to the accompanying drawings so that those skilled in the art can easily implement the present invention. The present invention can be embodied in many different forms and is not limited to the embodiments described herein.

[0029] like Figure 1 A structural schematic diagram of a prostatitis ultrasonic drug treatment planning system in an embodiment of the present invention is shown.

[0030] The system comprises:

[0031] The photoacoustic RF signal acquisition module 1 is used to collect the photoacoustic RF signal; wherein, the photoacoustic RF signal can guide the operator to collect through the acquired real-time image of the prostate.

[0032] The feature extraction module 2 is connected to the photoacoustic RF signal acquisition module 1 and is used to extract multi-dimensional prostate inflammation discrimination feature tensors from the photoacoustic RF signal; specifically, the module receives data from the photoacoustic RF signal acquisition module 1 and uses a machine learning algorithm to extract multi-dimensional prostate inflammation discrimination feature tensors from complex photoacoustic RF signals. These feature tensors not only contain the structural information inside the prostate tissue, but also reflect the physiological and pathological state of the tissue, especially the specific changes related to inflammation.

[0033] The discriminant network module 3 is connected to the feature extraction module 2, and is used to discriminate whether the prostate is inflammatory and the degree of inflammation according to the multivariate prostate inflammation discriminant feature tensor, and obtain the discrimination result of the degree of inflammation when it is judged to be inflammatory; specifically, this module constructs a discriminant network, which receives the multivariate prostate inflammation discriminant feature tensor output by the feature extraction module as input, and discriminates whether the prostate is inflammatory and the degree of inflammation by learning a large amount of labeled sample data. When it is judged to be inflammatory, the discriminant network can output the predicted probabilities of three classification results of mild inflammation, moderate inflammation, and severe inflammation, indicating the severity of the patient's prostate inflammation.

[0034] The decision network module 4 is connected to the discriminant network module 3 and the feature extraction module 2, and is used to obtain an inflammation treatment plan based on the inflammatory degree discrimination result and the prostate inflammation discrimination feature tensor. Specifically, this module constructs a decision network, which generates a personalized inflammation treatment plan for the patient based on the inflammatory degree discrimination result output by the discriminant network module and the multivariate prostate inflammation discrimination feature tensor. The plan includes ultrasonic acoustic parameters (such as frequency, intensity), excitation mode and duration (such as heat therapy, ultrasound-guided drug injection, etc.), administration method (oral, injection, local administration, etc.) and drug dosage.

[0035] In one embodiment, the feature extraction module is used to extract five types of prostate inflammation discriminant feature tensors from the photoacoustic RF signal based on a feature tensor extraction algorithm; wherein the five types of prostate inflammation discriminant feature tensors include:

[0036] The first feature tensor PAABTF (Prostate Acoustic Biomarker Tensor Feature) mainly focuses on the changes in acoustic properties within prostate tissue, such as sound velocity and sound attenuation. These changes are often related to tissue microstructure changes caused by prostate inflammation.

[0037] The second feature tensor PAAIB (Prostate Acoustic Intensity-based Feature) is based on the intensity information of the photoacoustic RF signal. It extracts features related to the degree of inflammation by analyzing the distribution and change trend of signal intensity. The intensity information can reflect conditions such as tissue congestion and edema caused by inflammation.

[0038] The third feature tensor, PAFIAB (Prostate Acoustic Frequency-Intensity Analysis-based Feature), combines the frequency and intensity information of the photoacoustic RF signal, and extracts features that are indicative of prostate inflammation through comprehensive consideration of frequency domain analysis and intensity distribution. This combined analysis can more comprehensively reflect the pathological changes inside the tissue.

[0039] The fourth feature tensor PAFSAB (Prostate Acoustic Frequency-Spatial Analysis-based Feature) uses advanced biometrics technology to extract advanced biometric features from photoacoustic RF signals, such as texture features, morphological features, etc. These features play an important role in identifying specific lesions of prostate inflammation.

[0040] The fifth feature tensor PASCS (Prostate Acoustic Spectral-Correlation-basedSpatial Feature) focuses on analyzing the temporal and spatial change sequence of the photoacoustic RF signal. By capturing the dynamic characteristics in the signal, it reveals the development process of prostate inflammation and its relationship with changes in tissue function.

[0041] In a specific embodiment, the algorithm for extracting feature tensors includes:

[0042] The first characteristic tensor PAABTF extraction algorithm includes: using the periodogram method to perform classical power spectrum density estimation on the photoacoustic RF signal, and then extracting the first characteristic tensor PAABTF;

[0043] Specifically, for the N-point discrete photoacoustic RF signal time series x[n], (0≤n≤N-1) obtained, the periodogram method is used to perform classical power spectral density estimation, that is,

[0044]

[0045] Where X(ω) is the discrete Fourier transform of the signal x[n], N is the length of the signal, and ω is the angular frequency.

[0046] Then the first characteristic tensor PAABTF is extracted to characterize the prostate characteristics. PAABTF is obtained by calculating the logarithmic difference between the power spectrum of the region of interest (SOI) and the power spectrum of the reference signal, and the calculation formula is:

[0047] ABTF(f)=10log 10 P SOI (f)-10log 10 P r (f); (2)

[0048] Among them, the power spectrum of the region of interest and the reference signal is P SOI and P r , which is calculated based on Fast Fourier Transform (FFT); the reference signal is obtained by reflecting the excitation signal off a steel plate in pure water.

[0049] The second characteristic tensor PAAIB extraction algorithm includes: calculating the integral of the first characteristic tensor PAABTF within the effective frequency band, and extracting the second characteristic tensor PAAIB;

[0050] Specifically, the first eigentensor PAABTF is calculated in the effective frequency band [f min , f max ] to extract the PAAIB to characterize the prostate characteristics, and the calculation formula is:

[0051]

[0052] Among them, f max and f min are the maximum and minimum values ​​of the analysis frequency band, respectively.

[0053] The third characteristic tensor PAAIB extraction algorithm includes: extracting the third characteristic tensor PAFIAB by extrapolating the best fitting straight line of the second characteristic tensor PAAIB within the set frequency range to the value at zero frequency;

[0054] Specifically, within the selected frequency range [f1, f2], the second eigentensor PAAIB is linearly fitted to obtain the best fitting straight line, the frequency is extrapolated to zero (ie, f=0), and the value corresponding to the straight line at this time is calculated. This value is the third eigentensor PAFIAB.

[0055] The fourth feature tensor PAFSAB extraction algorithm comprises: extracting the fourth feature tensor PAFSAB (dB) from the value of the slope of the best fitting straight line of the second feature tensor PAAIB within a set frequency range;

[0056] Specifically, within the selected frequency range [f1, f2], a linear fit is performed on the second eigentensor PAAIB to obtain a best fitting straight line, and the slope of the straight line is the fourth eigentensor PAFSAB (dB / MHz).

[0057] The fifth characteristic tensor PASCS extraction algorithm includes: obtaining the fifth characteristic tensor PASCS based on the integration of the photoacoustic RF signal in a set frequency range;

[0058] Specifically, for the fifth feature tensor PASCS, it can be described as:

[0059] First, the minimum value fmin and maximum value fmax of the ultrasonic transducer bandwidth are determined based on the reference signal; for the signal of the region of interest (SOI), its power spectrum density PSOI(f) in the frequency domain is calculated. Two integrals are calculated separately, one is the integral of the product of frequency ff and power spectrum density PSOI(f) in the range of fmin to fmax, and the other is the integral of power spectrum density PSOI(f) in the same frequency range. The calculation method is:

[0060]

[0061] Where f0 is the center frequency of the transducer.

[0062] In one embodiment, the discriminant network module 3 adopts a three-classification deep neural network based on a multi-stream fusion architecture. The network input is the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS extracted based on the photoacoustic RF signal, and the output is the three classification results of mild inflammation, moderate inflammation and severe inflammation.

[0063] Through this network design, the five types of tensors, namely the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS, can realize the functions of numerical standardization, multi-dimensional parameter distillation and multi-dimensional fusion. Finally, the fused tensors are classified into three categories using a multi-layer perceptron. Through model training with a large number of samples, the discriminant network can learn the mapping between the five types of data tensors and the three diagnostic classification results, thereby realizing the function of using the prostate inflammation judgment feature tensor as input, and constructing a three-classification deep neural network model with a multi-stream fusion architecture to judge whether the prostate is inflammatory and the degree of inflammation, and outputting the judgment results of mild inflammation, moderate inflammation and severe inflammation.

[0064] In a specific embodiment, if Figure 2 As shown, the three-classification deep neural network based on the multi-stream fusion architecture includes:

[0065] The parameter tensor standardization module is used to standardize the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS to a mean of 0 and a standard deviation of 1 respectively by using the standard deviation method. The main purpose of this parameter standardization is to eliminate network problems caused by different scales or dimensions so that different variables have equal weights in calculations. At the same time, it can also improve the convergence speed of the network model, reduce errors, and improve prediction accuracy.

[0066] The multi-stream pyramid parameter tensor distillation and fusion module adopts a multi-stream fusion architecture design and is connected to the parameter tensor standardization module to shrink the dimensions of the standardized first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS into a pyramid structure on each data flow line for distillation and selection of data tensors, and fuse multiple data streams on different tower layers while distilling and selecting to obtain fused feature tensors corresponding to the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS, respectively. This enables the five types of tensors to achieve multi-level data fusion through the network, and can fully explore the relationship between data for classification, while also effectively reducing the dimension of the data tensor and reducing the number of network model parameters.

[0067] The multi-layer perceptron module is connected to the multi-stream pyramid parameter tensor distillation and fusion module, which is used to merge the fused feature tensors and classify them to obtain the predicted probabilities of three classification results: mild inflammation, moderate inflammation, and severe inflammation. This module can deeply explore and integrate the complex relationship between input features through its carefully designed multi-layer hidden layer structure. At the output end, the softmax activation function is introduced to ensure that all output values ​​are normalized to the range of 0 to 1, and the sum is strictly equal to 1, which is directly mapped to the predicted probabilities of the three classification results of mild inflammation, moderate inflammation, and severe inflammation. This process not only achieves accurate prediction of the degree of inflammation, but also outputs the judgment results in an intuitive probability form, which is convenient for subsequent analysis and display. In addition, these predicted probabilities are seamlessly passed to the subsequent decision network module as new features, providing strong support for higher-level decision making.

[0068] In one embodiment, the decision network module is an iterative linear regression deep neural network based on multimodal data fusion, which is used to output the corresponding prostate inflammation treatment plan according to the input three modal data; wherein the input three modal data include: five types of prostate inflammation discrimination feature tensors, classification results and original treatment plans. The decision network module realizes the solidity of the close connection between the mild, moderate and severe discrimination results and the degree of inflammation, and realizes the intelligent output of the inflammation treatment plan through the constructed iterative linear regression deep neural network based on multimodal data fusion.

[0069] Through this network design, the three modal data can be standardized, multimodal aligned, multimodal feature extracted, and multimodal feature fused. The fused features are linearly regressed through a multi-layer perceptron, and then the iteratively updated treatment plan is obtained through tensor inverse normalization. Through model training with a large number of samples, the model can learn the mapping between the photoacoustic RF feature tensor, the classification result tensor, the original treatment plan tensor and the new treatment plan, so as to realize the function of outputting the iteratively updated prostate inflammation treatment plan by constructing an iterative linear regression deep neural network model with multimodal data fusion with five types of prostate inflammation discriminant feature tensors, classification results and original treatment plans as inputs. In particular, the initial treatment plan is set to a full 0 tensor (i.e., indicating an untreated state). The model will generate a new treatment plan based entirely on the patient's inflammatory characteristics and classification results, and the prostate inflammation treatment plan will be input into the network as the original treatment plan for the next training.

[0070] In a specific embodiment, if Figure 3 , the iterative linear regression deep neural network based on multimodal data fusion includes:

[0071] The parameter tensor standardization module is used to standardize the data of the three modes using the maximum and minimum methods; specifically, the parameter tensor standardization module uses the maximum and minimum methods to standardize each parameter in the multivariate prostate inflammation discriminant feature tensor, classification result tensor, and original treatment plan tensor to the range of 0 to 1. The main purpose of this parameter standardization is to eliminate the influence caused by different scales or dimensions, so that different variables have equal weights in calculations. At the same time, it can also improve the convergence speed of the network model, reduce errors, and improve prediction accuracy.

[0072] A heterogeneous multimodal tensor feature alignment and extraction module is connected to the parameter tensor standardization module and is used to perform feature tensor dimension alignment and data feature extraction on the standardized three modal data using their respective corresponding heterogeneous networks, so that the feature tensor dimensions of the three modal data can be aligned and the features of each modal data can be extracted to obtain feature data corresponding to the three modal data;

[0073] Specifically, heterogeneous networks for the characteristics of each modal data are selected, and these networks can deeply understand and adapt to the unique structure and properties of their respective modal data. By applying these heterogeneous networks, the module first realizes the dimension alignment of the feature tensor, ensuring that the representation of different modal data in the feature space has consistent dimensions and structures, laying the foundation for subsequent data fusion and analysis. On the basis of dimensional alignment, each network further performs the data feature extraction task. This process aims to extract the most representative feature information from the data of each modality, and these features can comprehensively and accurately reflect the essential attributes and internal laws of the data. The result of feature extraction is a series of refined feature data, which correspond to the data of the three modalities, respectively, and provide strong data support for subsequent classification, regression or decision-making tasks. Preferably, the heterogeneous network adopts RESNET50, multi-layer perceptron, and encoding and decoding module, respectively inputting the photoacoustic RF feature tensor, the classification result tensor, and the original treatment plan tensor, and outputting tensors of uniform dimensions, such as 128 dimensions.

[0074] The multimodal fusion module is connected to the heterogeneous multimodal tensor feature alignment and extraction module, and is used to use the tensor splicing method to initially fuse the feature data of the three modalities into an overall tensor, and then use the linear layer cascade softmax activation function to perform weight fusion to obtain the multimodal fusion tensor.

[0075] The multi-layer perceptron module is connected to the multi-modal fusion module and is used to perform linear regression on the fused multi-modal fusion tensor to iteratively update the treatment plan tensor. The multi-layer perceptron module performs linear regression on the fused tensor. By adding a sigmoid activation function to the output end of the multi-layer perceptron and mapping all outputs to the interval between 0 and 1, the iterative update of the treatment plan tensor can be achieved.

[0076] The tensor inverse normalization module is connected to the multi-layer perceptron module and is used to obtain an inflammation treatment plan, i.e., a personalized treatment plan, by inverse normalization of the iteratively updated treatment plan tensor. The inflammation treatment plan includes: ultrasonic acoustic parameters, duration of the excitation mode, administration method, and drug dosage.

[0077] For example, output ultrasonic acoustic parameters: 450mW / cm2, focal depth range 20cm~40mm, low-intensity pulse mode, 20% duty cycle, excitation mode: 20min. Output administration method: Levofloxacin injection, oral: 250 / 500mg once a day, blood drug concentration reaches peak 1-2h after administration (this should be the time point when we need to open the barrier) intravenous drip: 250 / 500mg intravenous drip (1 hour), blood drug concentration reaches peak 1h after administration (this should be the time point when we need to open the barrier). Output administration dosage: 250mg for mild dosage, 500mg for moderate / severe dosage. Reason: It has the characteristics of broad antibacterial spectrum and strong antibacterial effect; fat-soluble, acid-base → the best antibiotic permeability in the prostate, but the drug concentration in the prostatic fluid is still only 1 / 3 of that in the blood.

[0078] Similar to the principle of the above embodiment, the present invention provides a method for planning ultrasonic drug treatment of prostatitis.

[0079] The following provides specific embodiments in conjunction with the accompanying drawings:

[0080] like Figure 4 A flow chart showing a method for planning ultrasonic drug treatment for prostatitis according to an embodiment of the present invention.

[0081] The method comprises:

[0082] Step S1: Acquire the collected photoacoustic RF signal.

[0083] Step S2: extracting a multivariate prostate inflammation discrimination feature tensor from the photoacoustic RF signal.

[0084] In one embodiment, step S2 includes: extracting five types of prostate inflammation discrimination feature tensors from the photoacoustic RF signal based on a feature tensor extraction algorithm; wherein the five types of prostate inflammation discrimination feature tensors include: a first feature tensor PAABTF, a second feature tensor PAAIB, a third feature tensor PAFIAB, a fourth feature tensor PAFSAB and a fifth feature tensor PASCS.

[0085] In one embodiment, the algorithm for extracting feature tensors includes:

[0086] The first characteristic tensor PAABTF extraction algorithm includes: using the periodogram method to perform classical power spectrum density estimation on the photoacoustic RF signal, and then extracting the first characteristic tensor PAABTF;

[0087] The second characteristic tensor PAAIB extraction algorithm includes: calculating the integral of the first characteristic tensor PAABTF within the effective frequency band, and extracting the second characteristic tensor PAAIB;

[0088] The third characteristic tensor PAAIB extraction algorithm includes: extracting the third characteristic tensor PAFIAB by extrapolating the best fitting straight line of the second characteristic tensor PAAIB within the set frequency range to the value at zero frequency;

[0089] The fourth feature tensor PAFSAB extraction algorithm comprises: extracting the fourth feature tensor PAFSAB from the value of the slope of the best fitting straight line of the second feature tensor PAAIB within a set frequency range;

[0090] The fifth characteristic tensor PASCS extraction algorithm includes: obtaining the fifth characteristic tensor PASCS by calculating the ratio of the frequency-weighted integral to the unweighted integral of the signal intensity within the bandwidth of the ultrasonic transducer minus the center frequency.

[0091] Step S3: Determine whether the prostate is inflammatory and the degree of inflammation based on the multivariate prostate inflammation discrimination feature tensor, and obtain the inflammation degree discrimination result if it is judged to be inflammatory.

[0092] In one embodiment, step S3 includes: a three-classification deep neural network based on a multi-stream fusion architecture, inputting five types of prostate inflammation discrimination feature tensors, and outputting three-classification results corresponding to mild inflammation, moderate inflammation, and severe inflammation.

[0093] In one embodiment, the three-classification deep neural network based on the multi-stream fusion architecture includes:

[0094] A parameter tensor standardization module, used to standardize the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS respectively by using a standard deviation method;

[0095] A multi-stream pyramid parameter tensor distillation and fusion module is connected to the parameter tensor standardization module, and is used to shrink the dimensions of the standardized first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB, and the fifth feature tensor PASCS into a pyramid structure on each data flow line to perform distillation and selection of data tensors, and to fuse multiple data streams on different tower layers during the distillation and selection to obtain fused feature tensors corresponding to the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB, and the fifth feature tensor PASCS respectively;

[0096] The multi-layer perceptron module is connected to the multi-stream pyramid parameter tensor distillation and fusion module, and is used to merge the fused feature tensors and classify them to obtain the prediction probabilities of three classification results of mild inflammation, moderate inflammation, and severe inflammation.

[0097] Step S4: Obtaining an inflammation treatment plan based on the inflammation degree discrimination result and the prostate inflammation discrimination feature tensor.

[0098] In one embodiment, step S4 includes: an iterative linear regression deep neural network based on multimodal data fusion, outputting a corresponding prostate inflammation treatment plan according to the input data of three modalities; wherein the input data of three modalities include: five types of prostate inflammation discrimination feature tensors, classification results and the original treatment plan.

[0099] In one embodiment, the iterative linear regression deep neural network based on multimodal data fusion includes:

[0100] The parameter tensor normalization module is used to normalize the data of the three modes using the maximum and minimum value methods;

[0101] A heterogeneous multimodal tensor feature alignment and extraction module, connected to the parameter tensor standardization module, is used to perform feature tensor dimension alignment and data feature extraction on the standardized three modal data using their respective corresponding heterogeneous networks to obtain feature data corresponding to the three modal data;

[0102] A multimodal fusion module, connected to the heterogeneous multimodal tensor feature alignment and extraction module, is used to preliminarily fuse the feature data of the three modalities by using a tensor splicing method, and then use a linear layer cascade softmax activation function to perform weight fusion to obtain a multimodal fusion tensor;

[0103] The multi-layer perceptron module is used to perform linear regression on the fused multi-modal fusion tensor to iteratively update the treatment plan tensor;

[0104] The tensor inverse normalization module is used to obtain the inflammation treatment plan through the inverse normalization operation of the iteratively updated treatment plan tensor.

[0105] In one embodiment, the inflammation treatment program includes: ultrasonic acoustic parameters, duration of stimulation mode, administration method and drug dosage.

[0106] Since the implementation principle of the prostatitis ultrasonic drug treatment planning method has been described in the above-mentioned embodiment, it will not be repeated here.

[0107] The prostate ultrasound drug treatment planning system provided in the embodiment of the present invention can be implemented on the terminal side or the server side. As for the hardware structure of the electronic terminal, please refer to Figure 5, is an optional hardware structure diagram of an electronic terminal 1000 provided in an embodiment of the present invention. The terminal 1000 may be a mobile phone, a computer device, a tablet device, a personal digital processing device, a factory background processing device, etc. The terminal 1000 includes: at least one processor 1001, a memory 1002, at least one network interface 10010 and a user interface 1009. The various components in the device are coupled together through a bus system 1005. It can be understood that the bus system 1005 is used to realize the connection and communication between these components. In addition to the data bus, the bus system 1005 also includes a power bus, a control bus and a status signal bus. However, for the sake of clarity, in Figure 5 In the specification, various buses are labeled as bus systems.

[0108] The user interface 1009 may include a display, a keyboard, a mouse, a trackball, a click gun, keys, buttons, a touch pad or a touch screen.

[0109] It is understood that the memory 1002 can be a volatile memory or a non-volatile memory, and can also include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), which is used as an external cache. By way of exemplary but not limiting explanation, many forms of RAM are available, such as static random access memory (SRAM), synchronous static random access memory (SSRAM). The memory described in the embodiments of the present invention is intended to include but is not limited to these and any other suitable categories of memory.

[0110] The memory 1002 in the embodiment of the present invention is used to store various types of data to support the operation of the terminal 1000. Examples of these data include: any executable program used to operate on the terminal 1000, such as an operating system 10021 and an application 10022; the operating system 10021 includes various system programs, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application 10022 can include various applications, such as a media player (MediaPlayer), a browser (Browser), etc., for implementing various application services. The prostatitis ultrasound drug treatment planning system provided by the embodiment of the present invention can be included in the application 10022.

[0111] The method disclosed in the above embodiment of the present invention can be applied to the processor 1001, or implemented by the processor 1001. The processor 1001 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit in the processor 1001 or the instruction in the form of software. The above processor 1001 may be a general processor, a digital signal processor (DSP, Digital Signal Processor), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The processor 1001 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiment of the present invention. The general processor 1001 can be a microprocessor or any conventional processor, etc. In combination with the steps of the accessory optimization method provided in the embodiment of the present invention, it can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium, which is located in a memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0112] In an exemplary embodiment, the terminal 1000 may be implemented by one or more application specific integrated circuits (ASIC), DSP, programmable logic device (PLD), complex programmable logic device (CPLD) to execute the aforementioned method.

[0113] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

[0114] In the embodiments provided in the present application, the computer readable and writable storage medium may include a read-only memory, a random access memory, an EEPROM, a CD-ROM or other optical disk storage device, a disk storage device or other magnetic storage device, a flash memory, a USB flash drive, a mobile hard disk, or any other medium that can be used to store a desired program code in the form of an instruction or data structure and can be accessed by a computer. In addition, any connection can be appropriately referred to as a computer-readable medium. For example, if the instruction is sent from a website, a server or other remote source using a coaxial cable, an optical fiber cable, a twisted pair, a digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwaves, the coaxial cable, optical fiber cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwaves are included in the definition of the medium. However, it should be understood that computer readable and writable storage media and data storage media do not include connections, carriers, signals, or other temporary media, but are intended to be non-temporary, tangible storage media. Disk and disc, as used in this application, includes compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk and Blu-ray disc where disks usually reproduce data magnetically, while discs reproduce data optically with lasers.

[0115] In summary, the prostatitis ultrasonic drug treatment planning system, method and terminal of the present invention extracts a multivariate prostate inflammation discrimination feature tensor from the collected photoacoustic RF signal, and then discriminates whether the prostate is inflammatory and the degree of inflammation based on the multivariate prostate inflammation discrimination feature tensor. Once it is determined that inflammation exists, the system will immediately generate a discrimination result of the degree of inflammation, and customize a personalized inflammation treatment plan based on this result and the feature tensor. The present invention extracts the feature tensor in the photoacoustic RF signal through an algorithm, and then based on the constructed discrimination network and decision network, it can achieve accurate and intelligent evaluation of the patient's prostate inflammation degree, and outputs the ultrasound-drug synergistic treatment planning plan in a personalized manner, aiming to maximize the efficacy of the drug, thereby effectively shortening the treatment cycle. Therefore, the present invention effectively overcomes the various shortcomings in the prior art and has a high industrial utilization value.

[0116] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the technology may modify or change the above embodiments without violating the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by a person of ordinary skill in the art without departing from the spirit and technical ideas disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A prostatitis ultrasonic drug treatment planning system, characterized in that: The system comprises: A photoacoustic RF signal acquisition module, used to acquire the collected photoacoustic RF signal; A feature extraction module, connected to the photoacoustic RF signal acquisition module, for extracting a multivariate prostate inflammation discrimination feature tensor from the photoacoustic RF signal; A discriminant network module, connected to the feature extraction module, is used to discriminate whether the prostate is inflammatory and the degree of inflammation according to the multivariate prostate inflammation discriminant feature tensor, and obtain a discriminant result of the degree of inflammation if it is judged to be inflammatory; The decision network module is connected to the feature extraction module and the discrimination network module, and is used to obtain an inflammation treatment plan based on the inflammatory degree discrimination result and the prostate inflammation discrimination feature tensor.

2. The prostatitis ultrasonic drug treatment planning system according to claim 1, characterized in that: The feature extraction module is used to extract five types of prostate inflammation discrimination feature tensors from the photoacoustic RF signal based on the feature tensor extraction algorithm; wherein the five types of prostate inflammation discrimination feature tensors include: a first feature tensor PAABTF, a second feature tensor PAAIB, a third feature tensor PAFIAB, a fourth feature tensor PAFSAB and a fifth feature tensor PASCS.

3. The prostatitis ultrasonic drug treatment planning system according to claim 2, characterized in that: The algorithm for extracting feature tensors includes: The first characteristic tensor PAABTF extraction algorithm includes: using the periodogram method to perform classical power spectrum density estimation on the photoacoustic RF signal, and then extracting the first characteristic tensor PAABTF; The second characteristic tensor PAAIB extraction algorithm includes: calculating the integral of the first characteristic tensor PAABTF within the effective frequency band, and extracting the second characteristic tensor PAAIB; The third characteristic tensor PAAIB extraction algorithm includes: extracting the third characteristic tensor PAFIAB by extrapolating the best fitting straight line of the second characteristic tensor PAAIB within the set frequency range to the value at zero frequency; The fourth feature tensor PAFSAB extraction algorithm comprises: extracting the fourth feature tensor PAFSAB from the value of the slope of the best fitting straight line of the second feature tensor PAAIB within a set frequency range; The fifth characteristic tensor PASCS extraction algorithm includes: obtaining the fifth characteristic tensor PASCS based on the photoacoustic RF signal.

4. The prostatitis ultrasonic drug treatment planning system according to claim 3, characterized in that: The discriminant network module adopts a three-classification deep neural network based on a multi-stream fusion architecture, which is used to input five types of prostate inflammation discriminant feature tensors and output three-classification results corresponding to mild inflammation, moderate inflammation, and severe inflammation.

5. The prostatitis ultrasonic drug treatment planning system according to claim 4, characterized in that: The three-classification deep neural network based on the multi-stream fusion architecture includes: A parameter tensor standardization module, used to standardize the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB and the fifth feature tensor PASCS respectively by using a standard deviation method; A multi-stream pyramid parameter tensor distillation and fusion module is connected to the parameter tensor standardization module, and is used to shrink the dimensions of the standardized first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB, and the fifth feature tensor PASCS into a pyramid structure on each data flow line to perform distillation and selection of data tensors, and to fuse multiple data streams on different tower layers during the distillation and selection to obtain fused feature tensors corresponding to the first feature tensor PAABTF, the second feature tensor PAAIB, the third feature tensor PAFIAB, the fourth feature tensor PAFSAB, and the fifth feature tensor PASCS respectively; The multi-layer perceptron module is connected to the multi-stream pyramid parameter tensor distillation and fusion module, and is used to merge the fused feature tensors and classify them to obtain the prediction probabilities of three classification results of mild inflammation, moderate inflammation, and severe inflammation.

6. The prostatitis ultrasonic drug treatment planning system according to claim 5, characterized in that: The decision network module is an iterative linear regression deep neural network based on multimodal data fusion, which is used to output a corresponding prostate inflammation treatment plan based on the input three modal data; wherein the input three modal data include: five types of prostate inflammation discrimination feature tensors, classification results and original treatment plans.

7. The prostatitis ultrasonic drug treatment planning system according to claim 6, characterized in that: The iterative linear regression deep neural network based on multimodal data fusion includes: The parameter tensor normalization module is used to normalize the data of the three modes using the maximum and minimum value methods; A heterogeneous multimodal tensor feature alignment and extraction module, connected to the parameter tensor standardization module, is used to perform feature tensor dimension alignment and data feature extraction on the standardized three modal data using their respective corresponding heterogeneous networks to obtain feature data corresponding to the three modal data; A multimodal fusion module, connected to the heterogeneous multimodal tensor feature alignment and extraction module, is used to preliminarily fuse the feature data of the three modalities by using a tensor splicing method, and then use a linear layer cascade softmax activation function to perform weight fusion to obtain a multimodal fusion tensor; A multi-layer perceptron module, connected to the multi-modal fusion module, for performing linear regression on the fused multi-modal fusion tensor to iteratively update the treatment plan tensor; The tensor inverse normalization module is connected to the multi-layer perceptron module and is used to obtain the inflammation treatment plan by performing an inverse normalization operation on the iteratively updated treatment plan tensor.

8. The prostatitis ultrasonic drug treatment planning system according to claim 1, characterized in that: The inflammation treatment program includes: ultrasonic acoustic parameters, duration of stimulation mode, administration method and drug dosage.

9. A method for planning ultrasonic drug treatment of prostatitis, characterized in that: The method comprises: acquiring the collected photoacoustic RF signal; Extracting a multivariate prostate inflammation discriminant feature tensor from the photoacoustic RF signal; According to the multivariate prostate inflammation discrimination feature tensor, whether the prostate is inflammatory and the degree of inflammation are judged, and when it is judged to be inflammatory, the discrimination result of the degree of inflammation is obtained; The inflammation treatment plan is obtained based on the inflammation degree discrimination results and the prostate inflammation discrimination feature tensor.

10. An electronic terminal, characterized in that: include: one or more memories and one or more processors; The one or more memories are used to store computer programs; The one or more processors, connected to the memory, are configured to run the computer program to perform the method of the system according to any one of claims 1 to 8.

Images