Dynamic compensation system for wavelength-dependent light flux attenuation in gynecological tumor photoacoustic imaging

Abstract

The application relates to the technical field of biomedical engineering, and discloses a dynamic compensation system for wavelength-dependent light flux attenuation in gynecological tumor photoacoustic imaging. The system comprises a data acquisition module, a double-layer iterative compensation module, a depth normalization compensation module and an image reconstruction module. The data acquisition module is used for acquiring original radio frequency data of gynecological tumor photoacoustic imaging. The double-layer iterative compensation module is used for double-layer iterative spectral compensation on the original radio frequency data. The depth normalization compensation module is used for depth normalization compensation on the original radio frequency data. The depth normalization compensation module decomposes illumination and absorption spectrum through characteristic spectral multi-spectral photoacoustic tomography, layers each image according to depth, calculates a two-wavelength signal ratio for each depth layer, acquires an average exponential attenuation model of the corresponding depth layer, and normalizes the wavelength signal of the corresponding depth layer. The image reconstruction module is used for image reconstruction on the original radio frequency data after double-layer iterative compensation and depth normalization compensation, and adjusts the reconstructed image by minimizing reconstruction error.

Description

Technical Field

[0002] This application relates to the field of biomedical engineering technology, and in particular to a dynamic compensation system for wavelength-dependent light flux attenuation in photoacoustic imaging of gynecological tumors. Background Technology

[0003] Photoacoustic imaging (PAI), as an emerging non-invasive imaging technology, has shown potential in the diagnosis of gynecological tumors due to its molecular functional imaging capabilities, especially in areas such as blood oxygen saturation (SO2) estimation and microvascular density analysis, where it possesses irreplaceable advantages over traditional techniques. However, the pathological characteristics of gynecological tumors (such as invasive epithelial ovarian cancer, borderline and sex cord-stromal tumors) (high vascularization, low oxygen saturation, and deep tissue spectral staining effects) lead to core challenges for PAI, including a severe underestimation of deep tissue hemoglobin concentration, low signal-to-noise ratio, and incomplete presentation of functional information. Existing technologies mainly offer two solutions to address the wavelength-dependent light flux attenuation problem: 1. Single-spectral unmixing technique: It only uses linear spectral unmixing and assumes that the optical properties of the tissue are uniform, without considering the nonlinear characteristics of light flux attenuation in deep tissues, which leads to significant errors in deep tissue estimation. 2. Single-layer normalization technique: Although it corrects the spectral coloring effect through deep layering, it has high computational complexity and is difficult to meet the requirements of real-time imaging.

[0004] Therefore, a system is urgently needed to solve at least one of the above problems. Summary of the Invention

[0006] This application provides a dynamic compensation system for wavelength-dependent light flux attenuation in photoacoustic imaging of gynecological tumors, aiming to solve the problem that traditional methods either sacrifice accuracy for speed (such as single-layer linear demixing) or sacrifice speed for accuracy (such as single-layer modeling), and do not involve the synergistic optimization of the two compensation mechanisms.

[0007] In a first aspect, this application provides a dynamic compensation system for wavelength-dependent optical flux attenuation in photoacoustic imaging of gynecological tumors, comprising: The data acquisition module is used to acquire raw radio frequency data of photoacoustic imaging of gynecological tumors, wherein the raw radio frequency data includes photoacoustic signals of a first wavelength and a second wavelength. The dual-layer iterative compensation module is used to perform dual-layer iterative spectral compensation on the original radio frequency data. The first layer uses linear spectral demixing, assuming approximate uniformity, and directly demixes to obtain the estimated value. The second layer uses Monte Carlo or finite element forward model to calculate the iterative model, and uses the reflectivity signal of the second wavelength to compensate the reflectivity signal of the first wavelength of the first layer. The depth normalization compensation module is used to perform depth normalization compensation on the original radio frequency data. It decomposes the illumination and absorption spectrum through characteristic spectral multispectral photoacoustic tomography, divides each image into layers according to depth, calculates the ratio of two wavelength signals for each depth layer, obtains the average exponential decay model of the corresponding depth layer, normalizes the wavelength signal of the corresponding depth layer, and corrects the spectral coloring effect at each imaging position. The image reconstruction module is used to reconstruct images from the original radio frequency data after double-layer iterative compensation and depth normalization compensation, and to adjust the reconstructed image by minimizing the reconstruction error.

[0008] In some embodiments, the corresponding first layer employs linear spectral demixing, assuming approximate uniformity, to directly demix and obtain an estimated value, including: preprocessing the original radio frequency data to remove noise and interference signals, and obtaining a preprocessed radio frequency signal; based on the preprocessed radio frequency signal, assuming that the optical characteristics within the imaging area are approximately uniform, applying a linear spectral demixing algorithm to demix the photoacoustic signals of the first wavelength and the second wavelength, and calculating an initial estimated value of the hemoglobin concentration.

[0009] In some embodiments, the corresponding second layer employs a Monte Carlo or finite element forward model to calculate an iterative model, using the reflectivity signal of the second wavelength to compensate for the reflectivity signal of the first layer at the first wavelength. This includes: establishing a Monte Carlo model or a finite element forward model; inputting the initial estimate of the hemoglobin concentration obtained from the first layer and the reflectivity signal of the second wavelength into the Monte Carlo model or the finite element forward model; simulating the propagation process of light in the tissue using the Monte Carlo model or the finite element forward model to calculate an iterative model for light flux attenuation; and compensating for the reflectivity signal of the first layer at the first wavelength according to the iterative model to correct deviations caused by illumination effects.

[0010] In some embodiments, the step of decomposing illumination and absorption spectra by characteristic spectral multispectral photoacoustic tomography and layering each image by depth includes: using characteristic spectral multispectral photoacoustic tomography to perform spectral decomposition on the original radio frequency data to separate illumination signals and absorption spectrum signals; dividing each image into multiple depth layers according to the imaging depth, with each depth layer corresponding to a different depth range, so as to perform layered processing on signals at different depths.

[0011] In some embodiments, the step of calculating the ratio of two wavelength signals for each depth layer to obtain the average exponential decay model of the corresponding depth layer includes: for each depth layer, extracting photoacoustic signals of a first wavelength and a second wavelength respectively, and calculating the ratio of the photoacoustic signals of the first wavelength and the second wavelength; performing statistical analysis on the ratio of the photoacoustic signals of the first wavelength and the second wavelength in each depth layer, and fitting to obtain the average exponential decay model of the depth layer, wherein the average exponential decay model is used to characterize the decay characteristics of luminous flux with depth in the depth layer.

[0012] In some embodiments, normalizing the wavelength signals of the corresponding depth layer to correct the spectral coloring effect at each imaging location includes: normalizing all wavelength signals within the depth layer according to the average exponential decay model of each depth layer, adjusting the signal intensity at each imaging location; and compensating for the influence of light flux attenuation within the depth layer on the signal through normalization, thereby correcting the spectral coloring effect at each imaging location due to different depths.

[0013] In some embodiments, the image reconstruction of the original radio frequency data after double-layer iterative compensation and depth normalization compensation includes: processing the original radio frequency data after double-layer iterative compensation and depth normalization compensation according to a preset image reconstruction algorithm; the image reconstruction algorithm includes a back projection algorithm or a filtered back projection algorithm; and converting the processed radio frequency data into a two-dimensional or three-dimensional photoacoustic image to achieve visualization imaging of gynecological tumors.

[0014] In some embodiments, adjusting the reconstructed image by minimizing the reconstruction error includes: establishing a reconstruction error evaluation function, calculating the error between the reconstructed image and the original radio frequency data; and using an optimization algorithm to adjust the parameters in the image reconstruction process to minimize the value of the reconstruction error evaluation function and optimize the quality of the reconstructed image.

[0015] Secondly, this application provides a dynamic compensation method for wavelength-dependent luminous flux attenuation in photoacoustic imaging of gynecological tumors, applicable to the dynamic compensation system for wavelength-dependent luminous flux attenuation in photoacoustic imaging of gynecological tumors provided in any embodiment of this application. The method includes: Acquire raw radio frequency data for photoacoustic imaging of gynecological tumors, wherein the raw radio frequency data includes photoacoustic signals of a first wavelength and a second wavelength; The original radio frequency data is subjected to two-layer iterative spectral compensation. The first layer uses linear spectral demixing, assuming approximate uniformity, and directly demixes to obtain the estimated value. The second layer uses Monte Carlo or finite element forward model to calculate the iterative model, and uses the reflectivity signal of the second wavelength to compensate the reflectivity signal of the first wavelength of the first layer. The original radio frequency data is subjected to depth normalization compensation. Illumination and absorption spectra are decomposed by characteristic spectral multispectral photoacoustic tomography. Each image is layered by depth, and the ratio of two wavelength signals is calculated for each depth layer to obtain the average exponential decay model of the corresponding depth layer. This model is then used to normalize the wavelength signals of the corresponding depth layer and correct the spectral coloring effect at each imaging position. The original radio frequency data after double-layer iterative compensation and depth normalization compensation is then used for image reconstruction. The reconstructed image is adjusted by minimizing the reconstruction error.

[0016] Thirdly, embodiments of this application provide a computer device, the computer device including a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program and, when executing the computer program, implement the method provided in any embodiment of this application.

[0017] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the method provided in any embodiment of this application.

[0018] This application designs a detection system targeting the non-enzymatic functions of metabolic enzymes (such as stabilizing PCBP2 and activating the cGAS-STING pathway), overcoming the limitations of traditional enzyme activity detection and providing direct evidence for ferroptosis mechanism research. By linking multi-dimensional data from NGS, mass spectrometry, and metabolite detection, a composite scoring model is constructed to eliminate interference from classical metabolic pathways, significantly improving the specificity and sensitivity of ferroptosis validity assessment. It supports cross-platform validation of non-invasive samples (serum) and tissue samples, and expands to multi-gene panel analysis by combining disease-specific databases, making it suitable for personalized diagnosis and treatment of various solid tumors such as liver cancer, lung cancer, and ovarian cancer. Embedding AI analysis software and a remote cloud platform, it automates the detection process and enables multi-center data validation, complying with dynamic compensation (CDx) regulations and providing a standardized evaluation tool for combination immunotherapy.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1This is a schematic diagram of the structure of a dynamic compensation system for wavelength-dependent light flux attenuation in photoacoustic imaging of gynecological tumors provided in an embodiment of this application; Figure 2 This is a schematic flowchart illustrating the steps of a dynamic compensation method provided in an embodiment of this application; Figure 3 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application.

[0022] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0025] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0026] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0027] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0028] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0029] Photoacoustic imaging (PAI), as an emerging non-invasive imaging technology, has shown potential in the diagnosis of gynecological tumors due to its molecular functional imaging capabilities, especially in areas such as blood oxygen saturation (SO2) estimation and microvascular density analysis, where it possesses irreplaceable advantages over traditional techniques. However, the pathological characteristics of gynecological tumors (such as invasive epithelial ovarian cancer, borderline and sex cord-stromal tumors) (high vascularization, low oxygen saturation, and deep tissue spectral staining effects) lead to core challenges for PAI, including a severe underestimation of deep tissue hemoglobin concentration, low signal-to-noise ratio, and incomplete presentation of functional information. Existing technologies mainly offer two solutions to address the wavelength-dependent light flux attenuation problem: 1. Single-spectral unmixing technique: It only uses linear spectral unmixing and assumes that the optical properties of the tissue are uniform, without considering the nonlinear characteristics of light flux attenuation in deep tissues, which leads to significant errors in deep tissue estimation. 2. Single-layer normalization technique: Although it corrects the spectral coloring effect through deep layering, it has high computational complexity and is difficult to meet the requirements of real-time imaging.

[0030] Therefore, a method is urgently needed to solve at least one of the above problems.

[0031] To solve the above problem, please refer to Figure 1 This application provides a dynamic compensation system for wavelength-dependent luminous flux attenuation in photoacoustic imaging of gynecological tumors, comprising: a data acquisition module for acquiring raw radio frequency (RF) data of photoacoustic imaging of gynecological tumors, the raw RF data including photoacoustic signals of a first wavelength and a second wavelength; and a two-layer iterative compensation module for performing two-layer iterative spectral compensation on the raw RF data, wherein the first layer employs linear spectral demixing, assuming approximate uniformity, and directly demixes to obtain an estimated value; the second layer employs Monte Carlo or finite element forward model calculation of the iterative model, using the reflectivity signal of the second wavelength to calculate the reflectivity of the first wavelength in the first layer. The signal is compensated; the depth normalization compensation module is used to perform depth normalization compensation on the original radio frequency data. It decomposes the illumination and absorption spectrum through characteristic spectral multispectral photoacoustic tomography, divides each image into layers according to depth, calculates the ratio of two wavelength signals for each depth layer, obtains the average exponential decay model of the corresponding depth layer, and normalizes the wavelength signals of the corresponding depth layer to correct the spectral coloring effect at each imaging position; the image reconstruction module is used to reconstruct the image from the original radio frequency data after double-layer iterative compensation and depth normalization compensation, and adjusts the reconstructed image by minimizing the reconstruction error.

[0032] Specifically, the core advantage of photoacoustic imaging (PAI) in the diagnosis of gynecological tumors (such as ovarian cancer) lies in its ability to non-invasively provide functional information related to tumor angiogenesis and metabolism (such as blood oxygen saturation SO2). However, PAI presents unique challenges: different wavelengths of light attenuate differently in tissues (wavelength-dependent luminous flux attenuation), causing PA signals in deep tissues to be not only affected by the concentration of the target substance (such as hemoglobin) but also distorted by severe spectral coloration effects. The two key wavelengths used to estimate SO2 (such as ~750nm for deoxyhemoglobin sensitivity and ~850nm for oxyhemoglobin sensitivity) have different penetration depths, rendering simple ratio calculations ineffective. The signal-to-noise ratio (SNR) decreases sharply with depth: deep signals are weak, making the extraction of functional information extremely difficult. However, the special pathophysiological characteristics of gynecological tumors (such as deep location, high vascularization, and hypoxic microenvironment) present unique challenges. Existing single solutions each have their limitations: single spectral unmixing ignores the nonlinear attenuation caused by depth, resulting in inaccurate quantification of deep areas; single depth-level normalization can correct this, but its high computational complexity contradicts the requirements of real-time imaging.

[0033] This system proposes a dynamic compensation strategy that aims to combine the advantages of two approaches to control computational load while ensuring correction accuracy, thereby achieving more reliable and faster deep functional imaging of gynecological tumors.

[0034] This system is a software / hardware integrated system that integrates data acquisition, signal processing, model compensation, and image reconstruction. Its core workflow is as follows (conceptual illustration): [Data Acquisition] → [Two-layer Iterative Compensation Module] → [Depth Normalization Compensation Module] → [Image Reconstruction and Optimization Module] → [Final Corrected Image].

[0035] The data acquisition module is responsible for receiving raw radio frequency (RF) data streams from PAI devices (such as multispectral photoacoustic tomography systems). This involves acquiring time-domain RF signals from raw ultrasonic transducer arrays excited by lasers at different wavelengths (at least λ750nm and 850nm). Two key wavelengths, λ1 and λ2, are used. The hardware interface connects to a multi-channel data acquisition card, synchronously recording laser trigger signals and ultrasonic RF signals. The software performs preliminary preprocessing on the raw RF data, including bandpass filtering (filtering out noise outside the transducer's frequency response range) and time gain compensation (TGC, compensating for sound wave attenuation during tissue propagation). The preprocessed, wavelength-labeled RF data matrix is ​​then transmitted to the subsequent processing module.

[0036] The dual-layer iterative compensation module employs a two-step strategy of "coarse estimation followed by fine adjustment" to compensate for the amplitude of the original signal used for spectral unmixing in response to deep optical flux attenuation.

[0037] First layer (linear fast unmixing layer): It is initially assumed that the optical properties (scattering, absorption) of the tumor region are approximately uniform. The PA signals (A1 and A2) from the data acquisition module are processed. 0 A2 0 ), directly applying the linear spectral unmixing model. For example, solving the system of equations [A1 0 A2 0 ] = [ε_HbR(λ1), ε_HbO2(λ1); ε_HbR(λ2), ε_HbO2(λ2)]* [C_HbR; C_HbO2] * Φ0, where ε is the extinction coefficient, C is the concentration, and Φ0 is the initially assumed constant luminous flux. This yields a preliminary, depth-biased deoxyhemoglobin concentration C_HbR. （1） Oxyhemoglobin concentration (C_HbO2) （1） Estimates, and preliminary SO2 levels （1） Total blood volume (TBV) （1） Distribution map.

[0038] Layer 2 (Model-Driven Iterative Compensation Layer): Model Construction: Based on the optical property parameters of the tissue (approximate values ​​can be obtained from the results of the first layer or prior knowledge), a forward optical transmission model is established. Monte Carlo simulation or the finite element method (FEM) is typically used. This model can accurately calculate the spatiotemporal distribution Φ(λ, z) of photons propagating in complex multilayer tissue structures (such as skin, fat, muscle, and tumors), where z is the depth.

[0039] The compensation is implemented using the second wavelength λ2 (typically with stronger penetration and more stable signal) PA signal A2 as a reference. Since A2 ∝ C_total * ε_avg(λ2) * Φ(λ2, z), the distribution of Φ(λ2, z) at depth z can be calculated using a model. The compensation relationship is established as: A1_corrected(z) = A1 0 (z) * [Φ_model(λ2, z) / Φ_model(λ1, z)] * k0; where k is a correction factor related to the optical properties of the tissue, which can be obtained by least-squares fitting of the A2 signal and the model-predicted Φ(λ2, z).

[0040] Essentially, it uses a more precise two-wavelength luminous flux ratio Φ(λ2) / Φ(λ1) calculated by a model to correct the actual measured two-wavelength PA signal ratio A2. / A1 The bias introduced by different light attenuation is addressed. The compensated A1_corrected and A2 are then demixed again to obtain the updated C_HbR. （2） C_HbO2（2） This result can be fed back into the model to fine-tune the tissue optical parameters, and 1-2 iterations can be performed to optimize the results. This results in a more accurate distribution of hemoglobin concentrations (C_HbR', C_HbO2') and SO2' after depth-dependent luminous flux attenuation compensation.

[0041] The depth normalization compensation module further corrects the spectral coloring space inhomogeneity caused by tissue heterogeneity (such as vascular clusters and necrotic areas) in the two-dimensional / three-dimensional image space, achieving pixel-level normalization.

[0042] The initially reconstructed (or from Module 2) multi-wavelength PA image (B-scan or volumetric data) is divided into multiple thin layers along the depth direction (z-axis) (e.g., one layer every 100 μm). For each depth layer L, the ratio image R_L(x,y) = I_λ1(x,y) / I_λ2(x,y) of the PA signal intensity values ​​of the two wavelengths (λ1, λ2) at all pixels in that layer is calculated.

[0043] The average attenuation model is established by assuming that the luminous flux attenuation patterns are similar within the same thin layer L. The spatial average value μ_R_L of the ratio profile R_L(x,y) for this layer is calculated. This average value reflects the overall level of the attenuation difference between the two wavelengths at depth z_L.

[0044] Spectral color correction: For each pixel (x,y) within layer L, its original dual-wavelength signal [I_λ(x,y),I_λ(x,y)] is normalized: [I_λ1_norm, I_λ2_norm] = [I_λ1(x,y) / μ_R_L, I_λ2(x,y)] (or other equivalent mathematical form, the purpose of which is to eliminate the average ratio bias of this layer). This step is equivalent to removing the spectral baseline shift common to this depth layer caused by pure photophysical attenuation, so that subsequent unmixing reflects more the true differences in hemoglobin composition. Output: The multi-wavelength PA image data after depth-layer normalization has a "purer" spectral curve at different locations, which more directly corresponds to the composition of the absorber. 12 The image reconstruction and optimization module integrates the multi-wavelength PA signal after the above dynamic compensation processing to reconstruct high-fidelity, quantitative functional images (such as SO2 images and hemoglobin concentration images), and optimizes the image quality through post-processing.

[0045] Reconstruction algorithms can employ temporal backprojection (BP), time-delay stacking (DAS), or model-based iterative reconstruction algorithms (such as the Simultaneous Algebraic Reconstruction Technique (SART)) to reconstruct images from compensated RF data. The system sets an objective function, such as minimizing the difference (L2 norm) between the reconstructed image and the synthetic data predicted based on the forward model. Global optimum is achieved by fine-tuning reconstruction parameters (such as sound velocity distribution and initial pressure distribution) or a small number of parameters backpropagated to the compensation module using optimization algorithms such as gradient descent.

[0046] Ultimately, a series of core images are generated, including: a high-precision oxygen saturation (SO2) distribution map, a deoxygenated / oxygenated hemoglobin concentration map, and a total hemoglobin distribution map reflecting vascular density. These images have all undergone dynamic compensation for wavelength-dependent light flux attenuation, significantly improving quantitative reliability in both depth and space.

[0047] The core innovation of this dynamic compensation system lies in the synergy between "two-layer iterative model compensation" and "deep hierarchical spatial normalization." By introducing a physical model (Monte Carlo / finite element method), it corrects the systematic bias of linear unmixing in the depth dimension, improving the accuracy of deep quantitative analysis. Through simple hierarchical statistical processing, it efficiently smooths out spectral coloring noise caused by heterogeneity in the spatial dimension, improving the uniformity and comparability of images.

[0048] Compared to traditional full-model iteration or complex hierarchical inversion, this system offers fast first-layer unmixing for real-time preview, parallelizable second-layer model computation with background optimization, and low-computation-complexity depth normalization. The combination of these three features ensures accuracy while better meeting the needs of real-time or near-real-time clinical imaging.

[0049] In summary, this system provides a powerful signal processing solution that balances accuracy and efficiency for the precise functional diagnosis of deep and complex tissue structures such as gynecological tumors using photoacoustic imaging.

[0050] In some embodiments, the corresponding first layer employs linear spectral demixing, assuming approximate uniformity, to directly demix and obtain an estimated value, including: preprocessing the original radio frequency data to remove noise and interference signals, and obtaining a preprocessed radio frequency signal; based on the preprocessed radio frequency signal, assuming that the optical characteristics within the imaging area are approximately uniform, applying a linear spectral demixing algorithm to demix the photoacoustic signals of the first wavelength and the second wavelength, and calculating an initial estimated value of the hemoglobin concentration.

[0051] This embodiment constructs the first-layer processing framework of a two-layer iterative compensation module. Its core is to obtain an initial estimate of hemoglobin concentration through fast linear computation, providing benchmark data for subsequent precise compensation. This layer adopts the "approximately uniform" assumption, simplifying the complex nonlinear optical transmission problem into a system of linear equations that can be solved in real time, ensuring sufficient accuracy of physiological information while maintaining computational speed.

[0052] The preprocessing stage includes: noise removal: an adaptive wavelet threshold denoising algorithm is used to decompose the original radio frequency data into multiple scales. The threshold λ = σ√(2lnN) is set, where σ is the noise standard deviation and N is the signal length, effectively separating physiological signals and high-frequency electronic noise near the center frequency (usually 5-15MHz) of the ultrasonic transducer.

[0053] Interference suppression employs bandpass filters (bandwidth 3-20MHz) to eliminate low-frequency tissue motion artifacts and high-frequency electromagnetic interference. For gynecological tumor imaging, notch filtering is specifically added for periodic physiological movements (such as respiration and intestinal peristalsis), with the center frequency set at 0.2-0.5Hz.

[0054] Signal calibration eliminates amplitude deviations caused by differences in array units by performing transducer array sensitivity normalization and obtaining the response function of each channel using a standard phantom (containing Indian ink and lipid microspheres of known concentration).

[0055] The mathematical model is constructed by establishing a linear relationship between the photoacoustic signal amplitude A and the absorber concentration: A(λ) = Φ0 * [ε_HbR(λ) * C_HbR + ε_HbO2(λ) * C_HbO2], where Φ0 is the assumed uniform luminous flux, and ε is the molar extinction coefficient (at 750 nm: ε_HbR = 0.78 mm). -1 mM -1 ε_HbO2=0.38mm -1 mM -1 At 850 nm: ε_HbR = 0.32 mm -1 mM -1 ε_HbO2=0.58mm -1 mM -1 Matrix solution: Establish the following equation system for the dual-wavelength data: [A750] = Φ0*[0.78 0.38][C_HbR]; [A850] = Φ0*[0.32 0.58][C_HbO2]; Solve using the least squares method to obtain the initial concentration estimate C_HbR. （1） and C_HbO2 （1） .

[0056] Parameter initialization is achieved through total hemoglobin concentration (TBV). （1） =C_HbR （1） +C_HbO2（1） Blood oxygen saturation SO2 （1） =C_HbO2 （1） / TBV （1） This serves as the input prior for the two-layer model. A signal-to-noise ratio (SNR) threshold of >10dB is set, and the unmixing results for low SNR regions (depth >3cm) are marked as "to be corrected" to prevent erroneous estimations from propagating to the two-layer model.

[0057] In some embodiments, the corresponding second layer employs a Monte Carlo or finite element forward model to calculate an iterative model, using the reflectivity signal of the second wavelength to compensate for the reflectivity signal of the first layer at the first wavelength. This includes: establishing a Monte Carlo model or a finite element forward model; inputting the initial estimate of the hemoglobin concentration obtained from the first layer and the reflectivity signal of the second wavelength into the Monte Carlo model or the finite element forward model; simulating the propagation process of light in the tissue using the Monte Carlo model or the finite element forward model to calculate an iterative model for light flux attenuation; and compensating for the reflectivity signal of the first layer at the first wavelength according to the iterative model to correct deviations caused by illumination effects.

[0058] This embodiment constructs a precision compensation layer based on a physical optical transmission model. By forward simulating the actual propagation path of photons in the gynecological tumor microenvironment, it quantitatively corrects the depth-dependent bias caused by the linear assumption of the first layer. This layer achieves a leap from "empirical estimation" to "physical calculation" and is the core guarantee of the system's accuracy.

[0059] The Monte Carlo model configuration includes: tissue optical parameters: dynamically updated based on the first-layer results, with an absorption coefficient μa of 0.15-0.35 mm in the tumor region. -1 (Corresponding to TBV 50-120μM), reduced scattering coefficient μs' = 1.0-1.5mm -1 Normal ovarian tissue μ =0.05mm -1 μs'=0.8mm -1 .

[0060] The geometric structure was constructed using a three-layer cylindrical model (40mm in diameter): a surface layer (skin + fat, 5mm thick), a stromal layer (normal ovarian tissue, 15mm thick), and a tumor layer (irregular shape, automatically segmented based on the first layer's TBV > 70μM region). Photon tracing was performed by emitting 10 6 -10 7 Each photon packet records the luminous flux Φ(λ,z) ​​at each depth z, with a statistical error of <2%.

[0061] The finite element model configuration uses the diffusion approximation equation: - *[D Φ(r,λ)] + μ (r,λ)Φ(r,λ)=S(r,λ), where the diffusion coefficient D=1 / [3(μa+μs')]. The mesh generation uses a fine 0.2mm mesh for the tumor region and a coarse 0.5mm mesh for the outer region, with a total of approximately 50,000-80,000 nodes.

[0062] Boundary conditions are set by extrapolating boundary conditions to account for the tissue-air refractive index mismatch.

[0063] The iterative compensation mechanism includes: Reference signal selection: 850nm wavelength is chosen as the reference because its signal-to-noise ratio is better than 750nm in deep tissues (>2cm) (typically 3-5dB higher). The compensation algorithm calculates the luminous flux ratio at depth z using a model: R_model(z) = Φ_model(850nm,z) / Φ_model(750nm,z). The actual measured signal ratio is calculated: R_measured(z) = A850 （1） (z) / A750 （1） (z). Corrected 750nm signal A750_corrected(z)=A750 （1） (z)*[R_model(z) / R_measured(z)]^α, where α is an adaptive weighting factor (0.6-0.8) used to balance model accuracy and measurement noise. Iterative convergence is performed for 2-3 iterations, terminating when the change between two adjacent SO2 estimates is <5%. Typical computation time is 30-60 seconds (GPU accelerated).

[0064] During the phantom validation phase, Intralipid phantoms (μs'=1.0mm) containing different concentrations of hemoglobin (20-100μM) were used. -1 Within a depth range of 1-4 cm, the SO2 estimation error after compensation decreased from ±15% to ±5%.

[0065] In some embodiments, the step of decomposing illumination and absorption spectra by characteristic spectral multispectral photoacoustic tomography and layering each image by depth includes: using characteristic spectral multispectral photoacoustic tomography to perform spectral decomposition on the original radio frequency data to separate illumination signals and absorption spectrum signals; dividing each image into multiple depth layers according to the imaging depth, with each depth layer corresponding to a different depth range, so as to perform layered processing on signals at different depths.

[0066] This embodiment realizes the transformation from the original signal to a deep hierarchical data structure. It separates the illumination and absorption contributions through feature spectral decomposition and establishes a computational layer corresponding to the physical depth, providing standardized input for subsequent layer-by-layer attenuation modeling. This layer serves as a bridge connecting the signal space and the image space.

[0067] Feature spectral multispectral photoacoustic tomography (FE-MSOT) decomposition includes: Spectral separation: Singular value decomposition (SVD) is used to reduce the dimensionality of the multi-wavelength (≥5 wavelengths, 700-900nm range) dataset, extracting the first three principal components: PC1 (total hemoglobin distribution), PC2 (blood oxygen saturation distribution), and PC3 (illuminance distribution). Illuminance-absorption spectrum separation is achieved by establishing a spectral mixing model: S(λ)=a*S_illumination(λ)+b*S_HbR(λ)+c*S_HbO2(λ), and solving for the coefficients a, b, and c using the non-negative least squares (NNLS) algorithm to achieve physical separation of the illumination and absorption components.

[0068] The gynecologic tumor-specific optimization targets the high lipid content of ovarian cancer (lipid absorption peak at 930 nm) by adding a lipid baseline to the spectral library to avoid illumination estimation bias caused by lipid interference.

[0069] The depth stratification process includes: physical depth calibration: using the ultrasonic flight time t and the sound velocity c (approximately 1540 m / s in soft tissue) to calculate the depth z = c * t / 2, and converting the time-domain RF signal into a depth-domain signal.

[0070] The layer thickness is dynamically adjusted based on signal quality: Shallow layers (0-1cm): 0.5mm thickness, requiring fine layering due to strong and rapidly changing signals. Middle layers (1-2.5cm): 1.0mm thickness, balancing computational efficiency and accuracy. Deep layers (2.5-4cm): 1.5mm thickness, allowing for coarse layering due to weak and gradually changing signals. A 50% inter-layer overlap (sliding window) is set to avoid layer boundary artifacts and ensure the continuity of the attenuation model. A depth-layered data cube is generated, where each depth layer L contains the dual-wavelength signal pairs (A750, A850) of all pixels within that layer and their corresponding depth coordinates, forming a standardized input matrix for subsequent use.

[0071] In some embodiments, the step of calculating the ratio of two wavelength signals for each depth layer to obtain the average exponential decay model of the corresponding depth layer includes: for each depth layer, extracting photoacoustic signals of a first wavelength and a second wavelength respectively, and calculating the ratio of the photoacoustic signals of the first wavelength and the second wavelength; performing statistical analysis on the ratio of the photoacoustic signals of the first wavelength and the second wavelength in each depth layer, and fitting to obtain the average exponential decay model of the depth layer, wherein the average exponential decay model is used to characterize the decay characteristics of luminous flux with depth in the depth layer.

[0072] In this embodiment, a mathematical characterization of luminous flux attenuation is established within each depth layer. By statistically analyzing the signal ratio distribution and fitting an exponential attenuation model, the complex physical attenuation process is abstracted into calculable mathematical parameters, providing layer-specific correction coefficients for normalization.

[0073] Signal ratio calculation includes: pixel-level ratio: for each pixel i within the depth layer L, calculate the ratio r_i = A750(i) / A850(i), and generate a ratio distribution histogram H_L. Statistical feature extraction calculates the mean μ_L, standard deviation σ_L, and median M_L of the intra-layer ratio. For normal ovarian tissue, μ_L is typically in the range of 0.8–1.2; for hyperoxic tumors, μ_L may decrease to 0.5–0.7. Outlier removal uses the 3σ criterion to remove outlier ratios (|r_i-μ_L|>3σ_L). These outliers usually correspond to vessel edges or noisy pixels, avoiding interference with model fitting.

[0074] The fitting of the average exponential decay model includes the following model assumptions: The decay of luminous flux with depth z conforms to the modified Beer-Lambert law: Φ(λ,z)=Φ0(λ)·exp[-μ_eff(λ)*z], where the effective decay coefficient μ_eff=√[3μa(μa+μs')].

[0075] Dual-wavelength ratio model: The theoretical ratio R(z) = Φ(750,z) / Φ(850,z) = k·exp{-[μ_eff(750)-μ_eff(850)]·z}, which simplifies to R(z) = k·exp(-Δμ·z). For depth layer L, with the layer center depth z_L as the independent variable and the average ratio μ_L of this layer as the dependent variable, nonlinear least squares fitting is used to solve for the parameters k (initial ratio offset) and Δμ (attenuation coefficient difference): min Σ[μ_L - k*exp(-Δμ*z_L)] 2 ; Calculate the goodness of fit R 2 Only R is retained 2 Layers with a value >0.85 are suitable for low Ro. 2 Layers (usually due to weak signals) are interpolated using adjacent layer models. Layer-specific parameter outputs for each effective depth layer L consist of output parameter pairs (k_L, Δμ_L), forming a depth-dependent attenuation lookup table.

[0076] In some embodiments, normalizing the wavelength signals of the corresponding depth layer to correct the spectral coloring effect at each imaging location includes: normalizing all wavelength signals within the depth layer according to the average exponential decay model of each depth layer, adjusting the signal intensity at each imaging location; and compensating for the influence of light flux attenuation within the depth layer on the signal through normalization, thereby correcting the spectral coloring effect at each imaging location due to different depths.

[0077] This embodiment applies the obtained attenuation model to normalize and correct all pixel signals in each depth layer, eliminating spectral baseline drift caused by different depths, making pixels at different depth positions comparable, and achieving true depth-invariant imaging.

[0078] The correction factor is calculated for each pixel i at depth L as: C_i = exp[Δμ_L*(z_i - z_ref)], where z_ref is the reference depth (usually 0 cm, i.e., the surface), and Δμ_L is the difference in attenuation coefficients at that layer. The 750nm signal is multiplied by the correction factor: A750_norm(i) = A750(i)*C_i, while the 850nm signal remains unchanged (as a reference). This is equivalent to "flattening" all pixels to the same depth reference plane. This operation compensates for the additional attenuation of 750nm light relative to 850nm light at depth z_i, restoring the ratio of the two wavelengths to the surface level.

[0079] The spectral coloring effect manifests as the difference between the pseudo oxygen saturation SO2_measured(z) measured at depth z and the true SO2: ΔSO2(z) = SO2_measured(z) - SO2_true. Correction implementation: Recalculate the ratio r_i_norm = A750_norm(i) / A850(i) using the normalized signal, and then solve for the corrected oxygen saturation. SO2_corrected(i) = [ε_HbR(750) - r_i_norm*ε_HbR(850)] / [r_i_norm*(ε_HbO2(850)-ε_HbR(850)) - (ε_HbO2(750)-ε_HbR(750))] In the phantom validation, the SO2 estimation error at a depth of 3cm decreased from ±18% to ±6%, significantly improving the spatial continuity of deep vascular structures. For pixels with a normalized signal-to-noise ratio below 6dB, intra-layer neighborhood (3×3 window) mean filtering was used to avoid excessive noise amplification.

[0080] In some embodiments, the image reconstruction of the original radio frequency data after double-layer iterative compensation and depth normalization compensation includes: processing the original radio frequency data after double-layer iterative compensation and depth normalization compensation according to a preset image reconstruction algorithm; the image reconstruction algorithm includes a back projection algorithm or a filtered back projection algorithm; and converting the processed radio frequency data into a two-dimensional or three-dimensional photoacoustic image to achieve visualization imaging of gynecological tumors.

[0081] This embodiment converts the radio frequency signal, after double-layer compensation and depth normalization, into a visualized image. An analytical reconstruction algorithm is used to achieve fast and stable image generation, meeting clinical real-time requirements. This layer is the final stage of the entire system from signal processing to image output.

[0082] The reconstruction algorithm is selected through filtered back projection (FBP): For a ring array transducer (typically 256 elements, center frequency 7MHz), the standard FBP procedure is used: Hilbert transform is performed on each channel of the RF signal to obtain the envelope. A ramp filter (|ω| filter) is applied to enhance the high-frequency components and compensate for sound propagation loss. Back-projection integration is performed along the sound path to reconstruct the initial pressure distribution p0(x,y). Delay stacking (DAS) is used for linear arrays, calculating the delay τ from each element to pixel (x,y) as τ = √[(x-xn)]. 2 +(y-yn) 2 ] / c, after alignment and stacking: p0(x,y) = Σwn*An(t-τn); where wn is the apodization weight (such as Hamming window) to suppress sidelobes.

[0083] Multi-wavelength fusion reconstruction includes: Independent reconstruction: Reconstructing the compensated 750nm and 850nm data separately to obtain p0. 750 (x,y) and p0 850 (x,y). In iterative reconstruction, the structural similarity (identical vascular morphology) of the two wavelength signals is utilized to apply joint total variation (JTV) regularization: min ||p0 750 - H*A750|| 2 + |||p0 850 - H*A850|| 2 + λ*JTV(p0 750 , p0 850 ); where H is the system matrix, λ=0.1-0.3, effectively suppressing noise and preserving the edge.

[0084] Gynecologic tumor parameter mapping includes: functional image generation: in the reconstructed p0 750 and p0 850 Based on this, calculate pixel-by-pixel: Total hemoglobin concentration: TBV(x,y) ∝ p0 750 (x,y) / ε_eff(750) Blood oxygen saturation: SO2(x,y) Calculated according to the formula in the above embodiment. Microvessel density (MVD) analysis involves thresholding TBV images (>50 μM), calculating the vessel area ratio, and establishing a regression model (R²) with the pathological MVD score. 2 =0.78). Utilizing GPU parallel acceleration (NVIDIA RTX 4090), the reconstruction time for 256×256 pixels is <50ms, meeting the requirements for real-time imaging (>10 frames / second).

[0085] In some embodiments, adjusting the reconstructed image by minimizing the reconstruction error includes: establishing a reconstruction error evaluation function, calculating the error between the reconstructed image and the original radio frequency data; and using an optimization algorithm to adjust the parameters in the image reconstruction process to minimize the value of the reconstruction error evaluation function and optimize the quality of the reconstructed image.

[0086] This embodiment establishes a closed-loop feedback mechanism from image quality assessment to adaptive parameter adjustment. By quantifying reconstruction error and driving optimization algorithms, it achieves dynamic calibration of system parameters and continuous improvement of image quality, ensuring long-term stability and accuracy.

[0087] Construction of Reconstruction Error Evaluation Function: Data Fidelity Term: Calculates the residual between the forward projection of the reconstructed image and the actual measured RF data: E_data = ||A_measured - H*p0_reconstructed||2 2 / N; where N is the number of data points, and H is the system matrix (considering transducer response and sound velocity distribution). The physical consistency term uses the optical transmission model to constrain the rationality of the functional parameters: E_phys = Σ[SO2(x,y) -SO2_model(x,y; C_HbR, C_HbO2)] 2 ; where SO2_model is predicted by the Monte Carlo model.

[0088] Total error function: E_total = w1*E_data + w2*E_phys + w3*TV(p0), weights w1=0.5, w2=0.3, w3=0.2, and TV is the total variation regularization term to suppress noise.

[0089] Adjustable parameters include tissue optical parameters (μa, μs'), sound velocity c, and transducer position deviation Δx. Calculation E_total / θ (where θ is the parameter to be optimized) is iteratively updated using the Adam optimizer (learning rate 0.001, momentum 0.9): θ_{k+1} = θ_k - η· E_total(θ_k); Convergence criterion: Stop when |E_total(k+1)-E_total(k)|<0.1% or the number of iterations>50.

[0090] In clinical applications, a "calibration scan" is inserted every 5 minutes. Using a uniform phantom with known optical parameters, system parameters are rapidly optimized to compensate for transducer temperature drift and tissue sound velocity changes. The E_data curve is displayed in real time. If the error suddenly increases (e.g., >15%), an alarm is triggered and automatic recalibration is performed to avoid misdiagnosis. In data from 20 ovarian cancer patients, after optimization, the SO2 percentage in deep layers (>2.5cm) decreased to ±4.1% (9.2%), and the image quality score (out of 5) improved from 3.2 to 4.5, significantly enhancing clinical reliability. The repeated measures standard deviation decreased from ± It should be noted that the acquisition of any information mentioned in the system is in accordance with relevant regulations and with the user's consent, and will not infringe on the user's privacy or violate relevant laws and regulations.

[0091] Please see Figure 2 , Figure 2 This is a schematic flowchart illustrating a dynamic compensation method for wavelength-dependent luminous flux attenuation in photoacoustic imaging of gynecological tumors, provided in one embodiment of this application. The execution device for the method is a computer device deployed in the dynamic compensation system for wavelength-dependent luminous flux attenuation in photoacoustic imaging of gynecological tumors provided in any embodiment of this application.

[0092] like Figure 2 As shown, the provided method includes steps S101 to S103. The computer device can be a handheld terminal, a laptop computer, a wearable device, or a robot, etc. This is used to implement steps S101 to S103 and their corresponding embodiments.

[0093] It should be noted that the acquisition of any information mentioned in the provided methods is in compliance with relevant regulations and is carried out with the user's consent, and will not infringe on the user's privacy or violate relevant laws and regulations.

[0094] Step S101. Obtain raw radio frequency data of photoacoustic imaging of gynecological tumors, wherein the raw radio frequency data includes photoacoustic signals of a first wavelength and a second wavelength; Step S102. Perform two-layer iterative spectral compensation on the original radio frequency data. The first layer uses linear spectral demixing, assuming approximate uniformity, and directly demixes to obtain the estimated value. The second layer uses Monte Carlo or finite element forward model to calculate the iterative model, and uses the reflectivity signal of the second wavelength to compensate the reflectivity signal of the first wavelength of the first layer. Step S103. Perform depth normalization compensation on the original radio frequency data. Decompose the illumination and absorption spectrum using characteristic spectral multispectral photoacoustic tomography. Divide each image into depth layers and calculate the ratio of two wavelength signals for each depth layer to obtain the average exponential decay model of the corresponding depth layer. Normalize the wavelength signals of the corresponding depth layer to correct the spectral coloring effect at each imaging position. Perform image reconstruction on the original radio frequency data after double-layer iterative compensation and depth normalization compensation, and adjust the reconstructed image by minimizing the reconstruction error.

[0095] In some embodiments, the corresponding first layer employs linear spectral demixing, assuming approximate uniformity, to directly demix and obtain an estimated value, including: preprocessing the original radio frequency data to remove noise and interference signals, and obtaining a preprocessed radio frequency signal; based on the preprocessed radio frequency signal, assuming that the optical characteristics within the imaging area are approximately uniform, applying a linear spectral demixing algorithm to demix the photoacoustic signals of the first wavelength and the second wavelength, and calculating an initial estimated value of the hemoglobin concentration.

[0096] In some embodiments, the corresponding second layer employs a Monte Carlo or finite element forward model to calculate an iterative model, using the reflectivity signal of the second wavelength to compensate for the reflectivity signal of the first layer at the first wavelength. This includes: establishing a Monte Carlo model or a finite element forward model; inputting the initial estimate of the hemoglobin concentration obtained from the first layer and the reflectivity signal of the second wavelength into the Monte Carlo model or the finite element forward model; simulating the propagation process of light in the tissue using the Monte Carlo model or the finite element forward model to calculate an iterative model for light flux attenuation; and compensating for the reflectivity signal of the first layer at the first wavelength according to the iterative model to correct deviations caused by illumination effects.

[0097] In some embodiments, the step of decomposing illumination and absorption spectra by characteristic spectral multispectral photoacoustic tomography and layering each image by depth includes: using characteristic spectral multispectral photoacoustic tomography to perform spectral decomposition on the original radio frequency data to separate illumination signals and absorption spectrum signals; dividing each image into multiple depth layers according to the imaging depth, with each depth layer corresponding to a different depth range, so as to perform layered processing on signals at different depths.

[0098] In some embodiments, the step of calculating the ratio of two wavelength signals for each depth layer to obtain the average exponential decay model of the corresponding depth layer includes: for each depth layer, extracting photoacoustic signals of a first wavelength and a second wavelength respectively, and calculating the ratio of the photoacoustic signals of the first wavelength and the second wavelength; performing statistical analysis on the ratio of the photoacoustic signals of the first wavelength and the second wavelength in each depth layer, and fitting to obtain the average exponential decay model of the depth layer, wherein the average exponential decay model is used to characterize the decay characteristics of luminous flux with depth in the depth layer.

[0099] In some embodiments, normalizing the wavelength signals of the corresponding depth layer to correct the spectral coloring effect at each imaging location includes: normalizing all wavelength signals within the depth layer according to the average exponential decay model of each depth layer, adjusting the signal intensity at each imaging location; and compensating for the influence of light flux attenuation within the depth layer on the signal through normalization, thereby correcting the spectral coloring effect at each imaging location due to different depths.

[0100] In some embodiments, the image reconstruction of the original radio frequency data after double-layer iterative compensation and depth normalization compensation includes: processing the original radio frequency data after double-layer iterative compensation and depth normalization compensation according to a preset image reconstruction algorithm; the image reconstruction algorithm includes a back projection algorithm or a filtered back projection algorithm; and converting the processed radio frequency data into a two-dimensional or three-dimensional photoacoustic image to achieve visualization imaging of gynecological tumors.

[0101] In some embodiments, adjusting the reconstructed image by minimizing the reconstruction error includes: establishing a reconstruction error evaluation function, calculating the error between the reconstructed image and the original radio frequency data; and using an optimization algorithm to adjust the parameters in the image reconstruction process to minimize the value of the reconstruction error evaluation function and optimize the quality of the reconstructed image.

[0102] It should be noted that those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the dynamic compensation method and each step of the wavelength-dependent light flux attenuation in photoacoustic imaging of gynecological tumors described above can be referred to the corresponding process in the embodiments of the dynamic compensation system for wavelength-dependent light flux attenuation in photoacoustic imaging of gynecological tumors described above, and will not be repeated here.

[0103] Please see Figure 3 , Figure 3 This is a schematic block diagram of the structure of a computer device provided in an embodiment of this application. The computer device includes a processor, a memory, and a network interface connected via a device bus, wherein the memory may include a storage medium and internal memory.

[0104] The storage medium may store operating devices and computer programs. The computer program includes program instructions that, when executed, cause the processor to perform an embodiment of any dynamic compensation method for wavelength-dependent luminous flux attenuation in photoacoustic imaging of gynecological tumors.

[0105] The processor provides computing and control capabilities, supporting the operation of the entire computer device.

[0106] Internal memory provides an environment for the execution of computer programs in non-volatile storage media. When executed by a processor, the computer program enables the processor to perform any dynamic compensation system method for wavelength-dependent light flux attenuation in photoacoustic imaging of gynecological tumors.

[0107] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the terminal to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0108] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0109] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps: Acquire raw radio frequency data for photoacoustic imaging of gynecological tumors, wherein the raw radio frequency data includes photoacoustic signals of a first wavelength and a second wavelength; The original radio frequency data is subjected to two-layer iterative spectral compensation. The first layer uses linear spectral demixing, assuming approximate uniformity, and directly demixes to obtain the estimated value. The second layer uses Monte Carlo or finite element forward model to calculate the iterative model, and uses the reflectivity signal of the second wavelength to compensate the reflectivity signal of the first wavelength of the first layer. The original radio frequency data is subjected to depth normalization compensation. Illumination and absorption spectra are decomposed by characteristic spectral multispectral photoacoustic tomography. Each image is layered by depth, and the ratio of two wavelength signals is calculated for each depth layer to obtain the average exponential decay model of the corresponding depth layer. This model is then used to normalize the wavelength signals of the corresponding depth layer and correct the spectral coloring effect at each imaging position. The original radio frequency data after double-layer iterative compensation and depth normalization compensation is then used for image reconstruction. The reconstructed image is adjusted by minimizing the reconstruction error.

[0110] In some embodiments, the corresponding first layer employs linear spectral demixing, assuming approximate uniformity, to directly demix and obtain an estimated value, including: preprocessing the original radio frequency data to remove noise and interference signals, and obtaining a preprocessed radio frequency signal; based on the preprocessed radio frequency signal, assuming that the optical characteristics within the imaging area are approximately uniform, applying a linear spectral demixing algorithm to demix the photoacoustic signals of the first wavelength and the second wavelength, and calculating an initial estimated value of the hemoglobin concentration.

[0111] In some embodiments, the corresponding second layer employs a Monte Carlo or finite element forward model to calculate an iterative model, using the reflectivity signal of the second wavelength to compensate for the reflectivity signal of the first layer at the first wavelength. This includes: establishing a Monte Carlo model or a finite element forward model; inputting the initial estimate of the hemoglobin concentration obtained from the first layer and the reflectivity signal of the second wavelength into the Monte Carlo model or the finite element forward model; simulating the propagation process of light in the tissue using the Monte Carlo model or the finite element forward model to calculate an iterative model for light flux attenuation; and compensating for the reflectivity signal of the first layer at the first wavelength according to the iterative model to correct deviations caused by illumination effects.

[0112] In some embodiments, the step of decomposing illumination and absorption spectra by characteristic spectral multispectral photoacoustic tomography and layering each image by depth includes: using characteristic spectral multispectral photoacoustic tomography to perform spectral decomposition on the original radio frequency data to separate illumination signals and absorption spectrum signals; dividing each image into multiple depth layers according to the imaging depth, with each depth layer corresponding to a different depth range, so as to perform layered processing on signals at different depths.

[0113] In some embodiments, the step of calculating the ratio of two wavelength signals for each depth layer to obtain the average exponential decay model of the corresponding depth layer includes: for each depth layer, extracting photoacoustic signals of a first wavelength and a second wavelength respectively, and calculating the ratio of the photoacoustic signals of the first wavelength and the second wavelength; performing statistical analysis on the ratio of the photoacoustic signals of the first wavelength and the second wavelength in each depth layer, and fitting to obtain the average exponential decay model of the depth layer, wherein the average exponential decay model is used to characterize the decay characteristics of luminous flux with depth in the depth layer.

[0114] In some embodiments, normalizing the wavelength signals of the corresponding depth layer to correct the spectral coloring effect at each imaging location includes: normalizing all wavelength signals within the depth layer according to the average exponential decay model of each depth layer, adjusting the signal intensity at each imaging location; and compensating for the influence of light flux attenuation within the depth layer on the signal through normalization, thereby correcting the spectral coloring effect at each imaging location due to different depths.

[0115] In some embodiments, the image reconstruction of the original radio frequency data after double-layer iterative compensation and depth normalization compensation includes: processing the original radio frequency data after double-layer iterative compensation and depth normalization compensation according to a preset image reconstruction algorithm; the image reconstruction algorithm includes a back projection algorithm or a filtered back projection algorithm; and converting the processed radio frequency data into a two-dimensional or three-dimensional photoacoustic image to achieve visualization imaging of gynecological tumors.

[0116] In some embodiments, adjusting the reconstructed image by minimizing the reconstruction error includes: establishing a reconstruction error evaluation function, calculating the error between the reconstructed image and the original radio frequency data; and using an optimization algorithm to adjust the parameters in the image reconstruction process to minimize the value of the reconstruction error evaluation function and optimize the quality of the reconstructed image.

[0117] It should be noted that those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the processor described above can be referred to the corresponding process in the method embodiments of the above embodiments, and will not be repeated here.

[0118] The embodiments of this application also provide a computer-readable storage medium storing a computer program, the computer program including program instructions, and the processor executing the program instructions to implement the steps of the dynamic compensation method for wavelength-dependent light flux attenuation in photoacoustic imaging of gynecological tumors provided in the above embodiments of this application.

[0119] The computer-readable storage medium may be an internal storage unit of the computer device described in the foregoing embodiments, such as the hard disk or memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card equipped on the computer device.

[0120] It should be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. It should be understood that when an element or layer is referred to as “on,” “adjacent to,” “connected to,” or “coupled to” other elements or layers, it may be directly on, adjacent to, connected to, or coupled to other elements or layers, or there may be intervening elements or layers. Conversely, when an element is referred to as “directly on,” “directly adjacent to,” “directly connected to,” or “directly coupled to” other elements or layers, there are no intervening elements or layers. It should be understood that although the terms first, second, third, etc., may be used to describe various elements, components, areas, layers, and / or portions, these elements, components, areas, layers, and / or portions should not be limited by these terms. These terms are merely used to distinguish one element, component, area, layer, or portion from another element, component, area, layer, or portion. Therefore, without departing from the teachings of this application, the first element, component, area, layer, or portion discussed below may be referred to as a second element, component, area, layer, or portion.

[0121] Spatial relation terms such as “below,” “under,” “below,” “under,” “above,” “above,” etc., are used herein for convenience of description to describe the relationship between one element or feature shown in the figure and other elements or features. It should be understood that, in addition to the orientation shown in the figure, spatial relation terms are intended to also include different orientations of the device in use and operation. For example, if the device in the figure is flipped, then the element or feature described as “below,” “under,” or “below” other elements or features will be oriented “above” other elements or features. Therefore, the exemplary terms “below” and “under” can include both above and below orientations. The device may be otherwise oriented (rotated 90 degrees or otherwise) and the spatial descriptive terms used herein will be interpreted accordingly.

[0122] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of this application. When used herein, the singular forms “a,” “an,” and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the terms “comprising” and / or “including,” when used in this specification, identify the presence of the stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups. When used herein, the term “and / or” includes any and all combinations of the associated listed items.

[0123] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0124] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A dynamic compensation system for wavelength-dependent luminous flux attenuation in photoacoustic imaging of gynecological tumors, characterized in that, include: The data acquisition module is used to acquire raw radio frequency data of photoacoustic imaging of gynecological tumors, wherein the raw radio frequency data includes photoacoustic signals of a first wavelength and a second wavelength. The dual-layer iterative compensation module is used to perform dual-layer iterative spectral compensation on the original radio frequency data. The first layer uses linear spectral demixing, assuming approximate uniformity, and directly demixes to obtain the estimated value. The second layer uses Monte Carlo or finite element forward model to calculate the iterative model, and uses the reflectivity signal of the second wavelength to compensate the reflectivity signal of the first wavelength of the first layer. The depth normalization compensation module is used to perform depth normalization compensation on the original radio frequency data. It decomposes the illumination and absorption spectrum through characteristic spectral multispectral photoacoustic tomography, divides each image into layers according to depth, calculates the ratio of two wavelength signals for each depth layer, obtains the average exponential decay model of the corresponding depth layer, normalizes the wavelength signal of the corresponding depth layer, and corrects the spectral coloring effect at each imaging position. The image reconstruction module is used to reconstruct images from the original radio frequency data after double-layer iterative compensation and depth normalization compensation, and to adjust the reconstructed image by minimizing the reconstruction error.

2. The system according to claim 1, characterized in that, The corresponding first layer uses linear spectral unmixing, assuming approximate homogeneity, to directly obtain the estimated value through unmixing, including: The raw radio frequency data is preprocessed to remove noise and interference signals, and the preprocessed radio frequency signal is obtained. Based on the preprocessed radio frequency signal, assuming that the optical properties in the imaging area are approximately uniform, a linear spectral unmixing algorithm is applied to unmix the photoacoustic signals of the first and second wavelengths to calculate the initial estimate of the hemoglobin concentration.

3. The system according to claim 1, characterized in that, The corresponding second layer uses a Monte Carlo or finite element forward model to calculate an iterative model, and uses the reflectivity signal of the second wavelength to compensate for the reflectivity signal of the first wavelength of the first layer, including: A Monte Carlo model or a finite element forward model is established. The initial estimate of the hemoglobin concentration obtained from the first layer and the reflectivity signal of the second wavelength are input into the Monte Carlo model or the finite element forward model. The propagation process of light in the tissue is simulated by the Monte Carlo model or the finite element forward model, and an iterative model of light flux attenuation is calculated. Based on the iterative model, the reflectivity signal of the first wavelength of the first layer is compensated to correct the deviation caused by the influence of illumination.

4. The system according to claim 1, characterized in that, The method of decomposing illumination and absorption spectra through characteristic spectral multispectral photoacoustic tomography, and layering each image by depth, includes: The original radio frequency data was spectrally decomposed using characteristic spectral multispectral photoacoustic tomography to separate the illumination signal and the absorption spectrum signal. Each image is divided into multiple depth layers based on the imaging depth, with each depth layer corresponding to a different depth range, so that signals at different depths can be processed in layers.

5. The system according to claim 1, characterized in that, The step of calculating the ratio of the two wavelength signals for each depth layer to obtain the average exponential decay model for the corresponding depth layer includes: For each depth layer, the photoacoustic signals of the first wavelength and the second wavelength are extracted respectively, and the ratio of the photoacoustic signals of the first wavelength and the second wavelength is calculated. Statistical analysis was performed on the ratio of the first wavelength and the second wavelength photoacoustic signal in each depth layer, and an average exponential decay model of the depth layer was obtained by fitting. The average exponential decay model is used to characterize the decay characteristics of light flux with depth in the depth layer.

6. The system according to claim 1, characterized in that, The normalization of the wavelength signal at the corresponding depth layer to correct the spectral coloring effect at each imaging location includes: Based on the average exponential decay model of each depth layer, all wavelength signals within the depth layer are normalized, and the signal intensity at each imaging position is adjusted. By normalizing the process, the effect of light flux attenuation within the depth layer on the signal is compensated, and the spectral coloration effect caused by different depths at each imaging location is corrected.

7. The system according to claim 1, characterized in that, The image reconstruction of the original radio frequency data after double-layer iterative compensation and depth normalization compensation includes: Based on the preset image reconstruction algorithm, the original radio frequency data after double-layer iterative compensation and depth normalization compensation is processed; the image reconstruction algorithm includes back projection algorithm or filtered back projection algorithm; The processed radio frequency data is converted into two-dimensional or three-dimensional photoacoustic images to achieve visual imaging of gynecological tumors.

8. The system according to claim 1, characterized in that, The adjustment of the reconstructed image by minimizing the reconstruction error includes: A reconstruction error evaluation function is established to calculate the error between the reconstructed image and the original radio frequency data. An optimization algorithm is used to adjust the parameters in the image reconstruction process to minimize the value of the reconstruction error evaluation function and optimize the quality of the reconstructed image.

9. A dynamic compensation method, characterized in that, A dynamic compensation system for wavelength-dependent luminous flux attenuation in photoacoustic imaging of gynecological tumors, as described in any one of claims 1-8, comprising: Acquire raw radio frequency data for photoacoustic imaging of gynecological tumors, wherein the raw radio frequency data includes photoacoustic signals of a first wavelength and a second wavelength; The original radio frequency data is subjected to two-layer iterative spectral compensation. The first layer uses linear spectral demixing, assuming approximate uniformity, and directly demixes to obtain the estimated value. The second layer uses Monte Carlo or finite element forward model to calculate the iterative model, and uses the reflectivity signal of the second wavelength to compensate the reflectivity signal of the first wavelength of the first layer. The original radio frequency data is subjected to depth normalization compensation. Illumination and absorption spectra are decomposed by characteristic spectral multispectral photoacoustic tomography. Each image is layered by depth, and the ratio of two wavelength signals is calculated for each depth layer to obtain the average exponential decay model of the corresponding depth layer. This model is then used to normalize the wavelength signals of the corresponding depth layer and correct the spectral coloring effect at each imaging position. The original radio frequency data after double-layer iterative compensation and depth normalization compensation is then used for image reconstruction. The reconstructed image is adjusted by minimizing the reconstruction error.

10. A computer device, characterized in that, The computer device includes a memory and a processor; The memory is used to store computer programs; The processor is configured to execute the computer program and, in executing the computer program, implement the method as described in claim 9.

Images