Medical image enhancement analysis method for urology department

Abstract

The invention relates to the technical field of image analysis, and discloses a urology department medical image enhancement analysis method, which comprises the following steps: collecting multi-source heterogeneous data including ultrasonic, CT, MRI and other multi-modal image data, and patient electronic medical records, inspection reports and gene detection data; constructing an image enhancement model based on a quantum neural network, and performing feature extraction and enhancement strategy optimization on the image data by using the strong computing power and parallel processing characteristics of quantum computing; and designing a multi-modal space-time fusion enhancement module, wherein the module adopts an attention-guided spatial feature fusion mechanism to integrate key structures and lesion part features of different modal images in a spatial dimension. By constructing an image enhancement model based on a quantum neural network and designing a multi-modal space-time fusion enhancement module, efficient feature extraction and accurate enhancement of multi-modal image data of the urology department are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and specifically to a method for enhanced analysis of urological medical images. Background Art

[0002] In the field of urological medicine, accurate and clear images play a crucial role in disease diagnosis, treatment plan formulation, and surgical implementation. However, existing urological medical image enhancement technologies have deficiencies in many aspects. Traditional image enhancement methods, such as filter-based enhancement, histogram-based enhancement, and curve adjustment-based enhancement, can improve image quality to a certain extent. However, when dealing with multi-modal image data, they often fail to fully mine and integrate key information in different modal images. These methods are limited by computing power in feature extraction and enhancement effects, making it difficult to achieve in-depth processing and precise enhancement of image data. In addition, existing technologies also have limitations in dynamic image processing, unable to effectively capture and utilize the motion information of organ tissues in dynamic image sequences, resulting in difficulties in ensuring spatio-temporal consistency in dynamic enhancement. At the same time, most existing technologies ignore the physiological structure and functional characteristics of urological organs and fail to organically combine the physiological parameters of patients with the image enhancement process, resulting in deficiencies in the enhanced images in reflecting the true physiological functions of organs.

[0003] Therefore, those skilled in the art have proposed a method for enhanced analysis of urological medical images to solve the above problems. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides a method for enhanced analysis of urological medical images, which solves the problems raised in the above background art.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for enhanced analysis of urological medical images, comprising the following steps: Collect multi-source heterogeneous data, where the multi-source heterogeneous data includes multi-modal image data such as ultrasound, CT, MRI, as well as patient electronic medical records, test reports, and gene detection data; Construct an image enhancement model based on a quantum neural network, and utilize the powerful computing power and parallel processing characteristics of quantum computing to extract features from image data and optimize enhancement strategies; Design a multi-modal spatio-temporal fusion enhancement module. The module uses an attention-guided spatial feature fusion mechanism in the spatial dimension to integrate the key structures and lesion site features of different modal images, and uses optical flow estimation and temporal convolutional network in the temporal dimension to capture the motion information of organ tissues in dynamic image sequences to achieve spatio-temporal consistency under dynamic enhancement; Introduce an adaptive enhancement mechanism based on a physiological model, and adaptively adjust the image enhancement process according to the physiological structure and function model of urinary system organs and the patient's physiological parameters to highlight key image features related to physiological functions; Construct an interpretable enhancement decision-making mechanism to demonstrate the roles and impacts of various factors during the image enhancement process; Quantitatively evaluate the enhanced images, including calculating the contrast enhancement factor, the degree of signal-to-noise ratio improvement, and the edge sharpness index of key structures of the images, ensuring that the enhancement effect meets the diagnostic requirements. The quantitative evaluation also includes analyzing the texture features of the images, evaluating the smoothness and uniformity of the images by calculating the gray-level co-occurrence matrix of the images, and using this as a supplementary evaluation index for the enhancement effect; Based on the results of the quantitative evaluation, iteratively optimize the image enhancement model, adjust the parameters of the quantum neural network and the fusion strategy of the multi-modal spatio-temporal fusion enhancement module to continuously improve the enhancement effect. Use a genetic algorithm to globally optimize the search for the parameters of the quantum neural network, and at the same time use the gradient descent method to finely adjust the local parameters of the multi-modal spatio-temporal fusion enhancement module to achieve continuous improvement of the enhancement effect.

[0006] Preferably, in the step of constructing an image enhancement model based on a quantum neural network, it includes: Preprocess the multi-modal image data to unify different modal image data into a standard format and size suitable for the input of the quantum neural network; Use quantum gate circuits to construct the network structure of the quantum neural network, and achieve parallel exploration of various image enhancement strategies through quantum superposition states; Use the training data set to train the quantum neural network, and optimize the model parameters through the measurement and adjustment of quantum states to achieve the best image enhancement effect.

[0007] Preferably, in the step of designing the multi-modal spatio-temporal fusion enhancement module, the attention-guided spatial feature fusion mechanism includes: Calculate the spatial feature vectors of each region in different modal images; Based on the spatial feature vectors, use attention weights to determine the diagnostically valuable regions in different modal images; Fuse the key region features of different modal images according to the attention weights to highlight the key structures and lesion sites of urinary organs.

[0008] Preferably, in the step of designing the multi-modal spatio-temporal fusion enhancement module, the combination process of the optical flow estimation and the temporal convolutional network includes: The optical flow estimation algorithm calculates the motion vectors of pixel points between adjacent frames in the dynamic image sequence; Use the motion vector as the input of the temporal convolutional network, and capture the motion information of the organ tissue in the image sequence in the time dimension through the convolutional operation of the temporal convolutional network; Perform spatio-temporal fusion enhancement on the dynamic image sequence according to the motion information to ensure the spatio-temporal consistency of the image under dynamic enhancement.

[0009] Preferably, in the step of introducing the adaptive enhancement mechanism based on the physiological model, the physiological model includes the glomerular filtration rate model of the kidney, the volume-pressure relationship model of the bladder, etc., and the adaptive enhancement mechanism includes: Extract the patient's physiological parameters, such as glomerular filtration rate value, bladder volume and pressure value, etc.; Determine the key physiological states of the urinary system organs according to the physiological parameters and the corresponding physiological models; According to the key physiological states, adjust the parameters and strategies of image enhancement so that the enhanced image can more accurately reflect the physiological function of the organs.

[0010] Preferably, in the step of constructing the interpretable enhancement decision-making mechanism, it includes: Record key information such as the network structure, parameter changes of the quantum neural network, and fusion weights of the multi-modal spatio-temporal fusion enhancement module during the image enhancement process; Display the influence of different enhancement strategies on each region of the image and the role of the physiological model in adaptive enhancement through visualization techniques; Combine the patient's multi-source heterogeneous data to generate a report containing the basis for enhancement decision-making and result interpretation to assist doctors in making diagnostic decisions.

[0011] Preferably, after the step of collecting multi-source heterogeneous data, it also includes the step of performing quality assessment and cleaning on the data to ensure the accuracy and reliability of the input data. The quality assessment and cleaning include: Check the integrity and clarity of the image data, and eliminate images with serious artifacts or missing data; Unify the formats of data such as the patient's electronic medical records and test reports, and correct error information to ensure the consistency and usability of the data.

[0012] Preferably, after the interpretable enhancement decision-making mechanism, it also includes the step of comparing and analyzing the enhanced image with the original image to verify the enhancement effect and provide a reference for further optimization of the model. The comparison and analysis include: Quantitatively evaluate from multiple aspects such as the contrast, clarity, noise level, and prominence of key structures of the image; Subjectively evaluate the enhanced image by professional doctors and give feedback on the diagnostic value and clinical application effect.

[0013] Preferably, it further includes the step of customizing the image enhancement model for different patient groups, to meet the differences in the urinary system anatomical structures, physiological functions, and disease characteristics of different patients. The customization includes: Collecting and classifying a large amount of multi-source heterogeneous data of patients and analyzing the characteristics of different patient groups; Adjusting and optimizing the parameters of the quantum neural network and the multi-modal spatio-temporal fusion enhancement module according to the characteristics of the patient groups; Generating corresponding image enhancement models for each patient group to achieve personalized enhancement.

[0014] The present invention provides a method for enhancing and analyzing urological medical images. It has the following beneficial effects: 1. By constructing an image enhancement model based on a quantum neural network and designing a multi-modal spatio-temporal fusion enhancement module, the present invention realizes efficient feature extraction and precise enhancement of urological multi-modal image data. During the data processing, the QNN utilizes the powerful computing power and parallel processing characteristics of quantum computing to quickly mine the key information in the images, effectively avoiding the problems of insufficient feature extraction or poor enhancement effect caused by the computational power limitation of traditional enhancement methods. At the same time, the multi-modal spatio-temporal fusion enhancement module adopts an attention-guided spatial feature fusion mechanism in the spatial dimension to accurately integrate the key structure and lesion site features of different modal images, and uses optical flow estimation and temporal convolutional network in the temporal dimension to capture the motion information of organ tissues in the dynamic image sequence, ensuring spatio-temporal consistency under dynamic enhancement.

[0015] 2. The present invention introduces an adaptive enhancement mechanism based on a physiological model, which closely conforms to the physiological structure and function characteristics of the urinary system organs. During the data processing stage, by extracting the physiological parameters of the patients and combining them with corresponding physiological models, such as the glomerular filtration rate model of the kidney and the volume-pressure relationship model of the bladder, the image enhancement process can be adaptively adjusted according to the physiological state of the organs. This innovative data fusion and processing method not only enables the enhanced images to more accurately reflect the true physiological function of the organs, but also highlights the key image features closely related to the physiological function, providing more physiologically significant visual information for clinical diagnosis.

[0016] 3. The present invention constructs an interpretable enhanced decision-making mechanism to comprehensively and meticulously record and display the image enhancement process. During data processing and enhancement, by recording key information such as the network structure and parameter changes of the quantum neural network, as well as the fusion weights of the multi-modal spatio-temporal fusion enhancement module, and presenting the specific impacts of different enhancement strategies on each region of the image with the aid of visualization technology, a report containing the basis for enhancement decisions and result explanations is generated in combination with the multi-source heterogeneous data of the patient, providing a clear and intuitive way for doctors to understand the enhancement results. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] FIG. 1 is a flowchart of data collection and preprocessing according to the present invention; FIG. 2 is a flowchart of constructing an image enhancement model according to the present invention; FIG. 3 is a flowchart of multi-modal spatio-temporal fusion enhancement according to the present invention; FIG. 4 is a flowchart of an interpretable enhanced decision-making mechanism according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0019] Embodiment 1: Referring to FIGS. 1-4, an embodiment of the present invention provides a method for enhancing and analyzing urological medical images, including the following steps: Collect multi-source heterogeneous data, where the multi-source heterogeneous data includes multi-modal image data such as ultrasound, CT (Computed Tomography), MRI (Magnetic Resonance Imaging), as well as patient electronic medical records, test reports, and gene detection data; Construct an image enhancement model based on a quantum neural network, and use the powerful computing power and parallel processing characteristics of quantum computing to extract features and optimize enhancement strategies for image data; Constructing an image enhancement model based on a quantum neural network includes: Preprocess the multi-modal image data to unify different modal image data into a standard format and size suitable for input to the quantum neural network; Use quantum gate circuits to construct the network structure of the quantum neural network, and realize parallel exploration of various enhancement strategies for images through quantum superposition states; Train the quantum neural network using a training data set, and optimize the model parameters by measuring and adjusting quantum states to achieve the best image enhancement effect.

[0020] Specifically, first, it is necessary to preprocess the multi-modal image data. The purpose of this step is to convert the image data of different modalities (such as ultrasound, CT, MRI, etc.) into a standard format and size suitable for the input of the quantum neural network, ensuring the unity and compatibility of the data. Next, use quantum gate circuits to construct the network structure of the quantum neural network. Utilizing the characteristics of quantum superposition states, multiple image enhancement strategies can be explored simultaneously, thereby achieving parallel processing and improving the computational efficiency and enhancement effect. Finally, use the training dataset to train the constructed quantum neural network, evaluate the output of the network through the measurement of quantum states, and adjust and optimize the model parameters according to the evaluation results. This process is continuously iterated until the model achieves the best image enhancement effect, can accurately extract the key features in the image, and perform effective enhancement.

[0021] Design a multi-modal spatio-temporal fusion enhancement module. The module adopts an attention-guided spatial feature fusion mechanism in the spatial dimension to integrate the key structures and lesion site features of different modality images, and uses optical flow estimation and temporal convolutional network in the temporal dimension to capture the motion information of organ tissues in the dynamic image sequence to achieve spatio-temporal consistency under dynamic enhancement; Specifically, the design of the multi-modal spatio-temporal fusion enhancement module aims to achieve the efficient integration and dynamic enhancement of different modality image data to improve the diagnostic value of urological medical images. The module adopts an attention-guided spatial feature fusion mechanism in the spatial dimension. By calculating the spatial feature vectors of each region in different modality images and using attention weights to determine the regions valuable for diagnosis, the key structures and lesion site features can be accurately integrated, making the important information in the image more prominent. In the temporal dimension, the module combines optical flow estimation and temporal convolutional network. First, use the optical flow estimation algorithm to calculate the motion vectors of adjacent frame pixels in the dynamic image sequence, and then input this motion vector into the temporal convolutional network to capture the motion information of organ tissues in time, thereby ensuring the spatio-temporal consistency of the image under dynamic enhancement, enabling doctors to more accurately observe and analyze the dynamic changes of organ tissues.

[0022] Introduce an adaptive enhancement mechanism based on physiological models. According to the physiological structure and function models of urological organs and the patient's physiological parameters, the image enhancement process is adaptively adjusted to highlight the key image features related to physiological functions; In the step of introducing the adaptive enhancement mechanism based on physiological models, the physiological models include the glomerular filtration rate model of the kidney, the volume-pressure relationship model of the bladder, etc. The adaptive enhancement mechanism includes: Extract the patient's physiological parameters, such as glomerular filtration rate value, bladder volume and pressure value, etc.; Determine the key physiological states of urological organs according to the physiological parameters and the corresponding physiological models; Adjust the parameters and strategies of image enhancement according to the key physiological states, so that the enhanced image can more accurately reflect the physiological function of organs.

[0023] The attention-guided spatial feature fusion mechanism includes: Calculate the spatial feature vectors of each region in different modality images; Based on the spatial feature vectors, use attention weights to determine the diagnostically valuable regions in different modality images; Fuse the key region features of different modality images according to the attention weights to highlight the key structures and lesion sites of the urinary organs.

[0024] Construct an interpretable enhancement decision-making mechanism to show the roles and impacts of various factors in the image enhancement process, assist doctors in understanding the enhancement results, and provide references for subsequent scientific research and technological innovation.

[0025] In the steps of constructing an interpretable enhancement decision-making mechanism, it includes: Record the key information such as the network structure, parameter changes of the quantum neural network, and the fusion weights of the multi-modal spatio-temporal fusion enhancement module in the image enhancement process; Through visualization techniques, show the impacts of different enhancement strategies on each region of the image and the role of the physiological model in adaptive enhancement; Combine the multi-source heterogeneous data of the patient to generate a report containing the basis for enhancement decision-making and the interpretation of the results to assist doctors in making diagnostic decisions.

[0026] Specifically, constructing an interpretable enhancement decision-making mechanism is mainly to help doctors better understand the process and results of image enhancement, so as to make more accurate diagnoses and provide support for subsequent research. Firstly, it will record in detail the key information in the image enhancement process, such as the structure, parameter changes of the quantum neural network, and the fusion weights of the multi-modal spatio-temporal fusion enhancement module. These information are the basis for understanding the enhancement results. Then, using visualization techniques, visually show the impacts of different enhancement strategies on each region of the image and the role of the physiological model in adaptive enhancement. Finally, combine the multi-source heterogeneous data of the patient to generate a report containing the basis for enhancement decision-making and the interpretation of the results. This report can assist doctors in making more accurate diagnostic decisions and also provides valuable references for subsequent scientific research and technological innovation.

[0027] The combination process of optical flow estimation and temporal convolutional network includes: The optical flow estimation algorithm calculates the motion vectors of pixel points between adjacent frames in the dynamic image sequence; Take the motion vectors as the input of the temporal convolutional network, and capture the motion information of organ tissues in the time dimension through the convolutional operation of the temporal convolutional network.

[0028] Perform spatio-temporal fusion enhancement on the dynamic image sequence according to the motion information to ensure the spatio-temporal consistency of the image under dynamic enhancement.

[0029] Specifically, the optical flow estimation algorithm is used to calculate the motion vectors of pixel points between adjacent frames in the dynamic image sequence, which can accurately capture the motion direction and speed of pixel points in the time series. Input these motion vectors into the temporal convolutional network, and through convolutional operations, the motion feature information of organ tissues in the image sequence can be effectively extracted. According to the obtained motion information, perform spatio-temporal fusion enhancement processing on the dynamic image sequence, which can ensure the spatio-temporal consistency of the image during the dynamic enhancement process. This process makes the enhanced dynamic image maintain coherence in the time dimension and clarity and accuracy in the space dimension, providing higher-quality dynamic image data for medical diagnosis.

[0030] It also includes steps for quality assessment and cleaning of the data to ensure the accuracy and reliability of the input data. Quality assessment and cleaning include: Check the integrity and clarity of the image data, and eliminate images with serious artifacts or missing data; Unify the formats of data such as the patient's electronic medical records and test reports and correct error information to ensure the consistency and usability of the data.

[0031] Specifically, during the data processing process, the steps of quality assessment and cleaning are crucial to ensure the accuracy and reliability of the input data. For image data, it is necessary to strictly check its integrity and clarity, and eliminate those images with serious artifacts or data missing to avoid interference with subsequent analysis. At the same time, for structured and unstructured data such as the patient's electronic medical records and test reports, perform format unification and error information correction to ensure the consistency and usability of the data, so as to provide a high-quality data basis for subsequent image enhancement analysis.

[0032] It also includes steps for comparing and analyzing the enhanced image with the original image to verify the enhancement effect and provide a reference for further optimization of the model. The comparison and analysis include: Quantitatively evaluate from multiple aspects such as the contrast, clarity, noise level, and prominence of key structures of the image; Have professional doctors subjectively evaluate the enhanced image and give feedback on the diagnostic value and clinical application effect.

[0033] Specifically, to verify the image enhancement effect and provide a reference for model optimization, it is necessary to compare and analyze the enhanced image with the original image. This process includes two parts: quantitative evaluation and subjective evaluation. The quantitative evaluation is carried out from multiple dimensions such as image contrast, clarity, noise level, and prominence of key structures, and the enhancement effect is measured by objective indicators. At the same time, professional doctors subjectively evaluate the enhanced images and give feedback on the diagnostic value and clinical application effect based on their clinical experience to comprehensively evaluate the practical application value of the enhanced images.

[0034] It also includes the step of customizing the image enhancement model for different patient groups to meet the differences in the urinary system anatomical structures, physiological functions, and disease characteristics of different patients. The personalized customization includes: Collecting multi-source heterogeneous data of a large number of patients and conducting classification and statistics to analyze the characteristics of different patient groups; According to the characteristics of the patient group, adjusting and optimizing the parameters of the quantum neural network and the multi-modal spatio-temporal fusion enhancement module; Conducting quantitative evaluation on the enhanced images, including calculating the contrast enhancement factor of the images, the degree of signal-to-noise ratio improvement, and the edge clarity index of key structures, to ensure that the enhancement effect meets the diagnostic requirements. The above-mentioned quantitative evaluation also includes analyzing the texture features of the images. By calculating the gray-level co-occurrence matrix of the images, the smoothness and uniformity of the images are evaluated, which is used as a supplementary evaluation index for the enhancement effect. Specifically, the calculation of the contrast enhancement factor is an important index to measure the degree of gray-scale difference change between different tissues or structures in the image. By comparing the gray-scale distribution differences of the corresponding regions of the images before and after enhancement, it can intuitively reflect the improvement effect of the enhancement process on the image contrast. In urological image diagnosis, such as the detection of renal tumors, if the enhanced image has higher contrast, doctors can more clearly distinguish the tumor tissue from the normal renal parenchyma, detect small tumors at an early stage, and accurately evaluate their scope and invasion depth.

[0035] Secondly, the degree of signal-to-noise ratio improvement is directly related to the clarity and detail presentation ability of the image. The signal-to-noise ratio reflects the ratio of the effective signal (organizational structure information) to the noise signal in the image. After enhancing the images of urinary system organs such as the bladder, a higher signal-to-noise ratio means that the fine hierarchical structures of the bladder wall, potential superficial tumors or inflammatory lesions, etc. can be highlighted from the background noise, thus providing more accurate diagnostic basis for doctors and reducing the possibility of misdiagnosis and missed diagnosis. The edge clarity index of key structures focuses on the sharpening and highlighting effect of the enhancement process on the boundaries of key regions such as organs and lesions. In the scenario of imaging diagnosis of prostate hyperplasia, clear edges of the prostate capsule and the boundaries between the internal glands and the surrounding tissues can accurately judge the degree of prostate hyperplasia, the situation of protruding into the bladder, and the degree of compression on the urethra, providing key morphological basis for formulating personalized treatment plans in the future.

[0036] In addition, the quantitative evaluation also delves into the analysis level of the texture features of the images. By calculating the gray-level co-occurrence matrix (GLCM) of the images, the smoothness and uniformity of the images are quantified from a mathematical statistical perspective. Smoothness reflects the intensity of gray-scale changes in the images, while uniformity reflects the uniformity of the gray-scale distribution of the images.

[0037] In the urological imaging diagnosis, for the evaluation of diffuse kidney lesions, the normal renal parenchyma usually has relatively uniform texture features. When lesions such as glomerulonephritis occur, the texture of the renal parenchyma will show non-uniform changes. The smoothness and uniformity indexes calculated by GLCM can assist doctors in more accurately identifying and analyzing such abnormal changes in image texture, further exploring the potential information related to diseases in the images, so as to more comprehensively evaluate the enhancement effect, ensure that the enhanced images reach the best state in reflecting the physiological functions and pathological features of urinary system organs, meet the high standards of clinical diagnosis, and provide strong support for doctors to make accurate treatment decisions.

[0038] Based on the results of the quantitative evaluation, the image enhancement model is iteratively optimized, and the parameters of the quantum neural network and the fusion strategy of the multi-modal spatio-temporal fusion enhancement module are adjusted to continuously improve the enhancement effect. The genetic algorithm is used to globally optimize the search for the parameters of the quantum neural network, and at the same time, the gradient descent method is used to finely adjust the local parameters of the multi-modal spatio-temporal fusion enhancement module to achieve continuous improvement of the enhancement effect.

[0039] Specifically, through quantitative evaluation means such as calculating the contrast enhancement factor, the degree of signal-to-noise ratio improvement, the edge sharpness index of key structures of the images, and analyzing the texture features of the images, the quality of the enhanced images is comprehensively measured. These quantitative evaluation results provide a clear basis and direction for the optimization of the model. In the present invention, using the genetic algorithm to globally optimize the search for the parameters of the quantum neural network can avoid the problem that traditional optimization methods are prone to falling into local optima, thereby more effectively improving the performance of the quantum neural network, enabling it to better extract and enhance key features in image enhancement.

[0040] The multi-modal spatio-temporal fusion enhancement module is responsible for integrating the key structures and lesion site features of different modal images and capturing the motion information of organ tissues in the dynamic image sequence. In order to achieve fine adjustment of the local parameters of this module, the gradient descent method is used. The gradient descent method is a local optimization algorithm based on gradient information. It updates the parameters along the direction of the gradient by calculating the gradient of the loss function with respect to the parameters, thereby gradually reducing the value of the loss function and achieving the optimization of the parameters. Its core formula is:

[0041] Among them, Parameters representing the multi-modal spatio-temporal fusion enhancement module; n represents the learning rate, controlling the step size of each parameter update; ∇J(θ) represents the loss function. By finely adjusting the local parameters of the multi-modal spatio-temporal fusion enhancement module, it can better adapt to the characteristics of different modal images and the requirements of dynamic image sequences, further improving the effect of image enhancement.

[0042] By combining the genetic algorithm with the gradient descent method, globally optimizing and searching the parameters of the quantum neural network, and at the same time finely adjusting the local parameters of the multi-modal spatio-temporal fusion enhancement module, continuous optimization of the image enhancement model can be achieved, continuously improving the enhancement effect, and meeting the high requirements of clinical diagnosis for image quality.

[0043] Generate corresponding image enhancement models for each patient group to achieve personalized enhancement.

[0044] Example 2: Application in the diagnosis of renal tumors Data collection: Select the multi-modal image data of 50 renal tumor patients in the urology department of a certain tertiary hospital. Including ultrasound, CT, and MRI images, and at the same time collect the electronic medical records, test reports, and gene detection data of the patients.

[0045] Model construction and training: Construct an image enhancement model based on the quantum neural network. After preprocessing the multi-modal image data, use quantum gate circuits to construct the network structure, and use the training data set to train and optimize the model.

[0046] Multi-modal spatio-temporal fusion enhancement: Process the image data through the multi-modal spatio-temporal fusion enhancement module. In the spatial dimension, use the attention-guided spatial feature fusion mechanism to integrate the key structures of different modal images and the features of tumor lesions. In the time dimension, use optical flow estimation and temporal convolutional network to capture the motion information of the kidneys in the dynamic image sequence.

[0047] Adaptive enhancement and decision-making: Introduce an adaptive enhancement mechanism based on the glomerular filtration rate model of the kidneys, and adjust the parameters and strategies of image enhancement according to the physiological parameters such as the glomerular filtration rate of the patients. Finally, construct an interpretable enhancement decision-making mechanism, record key information, and generate reports to assist doctors in diagnosis.

[0048] Results: The enhanced images can more clearly show the size, location, boundary of renal tumors and their relationship with surrounding tissues, and details such as necrosis and bleeding inside the tumors are also more obvious, providing strong support for doctors to formulate precise treatment plans.

[0049] Example 3: Application in the pre-operative evaluation of bladder tumors Data collection: Collect the ultrasound, CT, and MRI multimodal imaging data and relevant clinical information of 30 patients with bladder tumors from another top - tier hospital.

[0050] Model construction and training: Construct a quantum neural network image enhancement model using the method of the present invention, and train and optimize the model according to the characteristics of bladder tumors.

[0051] Multimodal spatio - temporal fusion enhancement: Use the multimodal spatio - temporal fusion enhancement module to process the images, highlighting the morphology, size, location of bladder tumors and the hierarchical structure of the bladder wall, while capturing the dynamic change information of the bladder during the filling and emptying processes.

[0052] Adaptive enhancement and decision - making: Combine physiological models such as the volume - pressure relationship model of the bladder, and achieve adaptive enhancement according to physiological parameters such as the bladder volume and pressure of the patient, so that the images can more accurately reflect the impact of bladder tumors on bladder function. Provide detailed diagnostic basis for doctors through an interpretable enhancement decision - making mechanism.

[0053] Results: Doctors can more accurately evaluate the stage, invasion depth of bladder tumors and the feasibility of surgical resection, improve the accuracy of preoperative evaluation, formulate personalized surgical plans, and reduce surgical risks.

[0054] Example 4: Application data collection in the analysis of imaging features of benign prostatic hyperplasia Select the multimodal imaging data and relevant examination reports of 40 patients with benign prostatic hyperplasia.

[0055] Model construction and training: Construct an image enhancement model based on a quantum neural network, and train it according to the characteristics of benign prostatic hyperplasia so that the model can better identify and enhance the relevant imaging features of the prostate.

[0056] Multimodal spatio - temporal fusion enhancement: Use the multimodal spatio - temporal fusion enhancement module to integrate information on the morphology, size, location of the prostate and the surrounding tissue structures in different - modality images, and capture the changes of the prostate under different physiological states in the time dimension.

[0057] Adaptive enhancement and decision - making: According to physiological parameters such as the prostate - specific antigen (PSA) level of the patient, introduce corresponding physiological models to achieve adaptive enhancement, making the internal glandular structure, capsule and other features of the prostate clearer, and providing more accurate imaging basis for doctors to analyze the degree and type of benign prostatic hyperplasia.

[0058] Results: The enhanced images enable doctors to more precisely judge the severity of benign prostatic hyperplasia, improve the accuracy and reliability of diagnosis, and provide strong support for subsequent treatment options.

[0059] Comparative example: Performance of traditional methods in urological imaging enhancement Data collection: Select multimodal images of the same number and type of renal tumor patients as in Example 1 data.

[0060] Traditional enhancement method: Use the traditional image enhancement method based on filtering and histogram equalization to process the image data.

[0061] Results: The enhanced images are not as good as the method of the present invention in terms of detail display and functional reflection. The tumor margins are not clear enough to accurately judge the infiltration range of the tumor; the improvement degree of the contrast and noise level of the images is limited, and it is impossible to highlight the key image features closely related to physiological functions like the method of the present invention, providing relatively low value for clinical diagnosis.

[0062] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for enhanced analysis of urological medical images, characterized in that, It includes the following steps: Collect multi-source heterogeneous data, where the multi-source heterogeneous data includes multi-modal image data such as ultrasound, CT, MRI, etc., as well as patient electronic medical records, test reports, and gene detection data; Construct an image enhancement model based on a quantum neural network, and use the powerful computing power and parallel processing characteristics of quantum computing to extract features and optimize enhancement strategies for the image data; Design a multi-modal spatio-temporal fusion enhancement module. This module uses an attention-guided spatial feature fusion mechanism in the spatial dimension to integrate the key structures and lesion site features of different modal images, and uses optical flow estimation and temporal convolutional network in the temporal dimension to capture the motion information of organ tissues in the dynamic image sequence to achieve spatio-temporal consistency under dynamic enhancement; Introduce an adaptive enhancement mechanism based on a physiological model. According to the physiological structure and function model of the urinary system organs and the patient's physiological parameters, the image enhancement process is adaptively adjusted to highlight the key image features related to physiological functions; Construct an interpretable enhancement decision-making mechanism to show the roles and influences of various factors in the image enhancement process; Quantitatively evaluate the enhanced images, including calculating the contrast enhancement factor, signal-to-noise ratio improvement degree, and edge sharpness index of key structures of the images to ensure that the enhancement effect meets the diagnostic requirements. The quantitative evaluation also includes analyzing the texture features of the images, and evaluating the smoothness and uniformity of the images by calculating the gray-level co-occurrence matrix of the images, which is used as a supplementary evaluation index for the enhancement effect; Based on the quantitative evaluation results, iteratively optimize the image enhancement model, adjust the parameters of the quantum neural network and the fusion strategy of the multi-modal spatio-temporal fusion enhancement module to continuously improve the enhancement effect. Use a genetic algorithm to globally optimize the search for the parameters of the quantum neural network, and at the same time use the gradient descent method to fine-tune the local parameters of the multi-modal spatio-temporal fusion enhancement module to continuously improve the enhancement effect. Its characteristics lie in that in the step of constructing the image enhancement model based on a quantum neural network, it includes: Preprocess the multi-modal image data to unify different modal image data into a standard format and size suitable for the input of the quantum neural network; Use quantum gate circuits to construct the network structure of the quantum neural network, and realize the parallel exploration of various enhancement strategies for images through quantum superposition states; 2. The urological medical image enhancement analysis method according to claim 1, Use the training data set to train the quantum neural network, and optimize the model parameters through the measurement and adjustment of quantum states to achieve the best image enhancement effect. Its characteristics lie in that in the step of designing the multi-modal spatio-temporal fusion enhancement module, the attention-guided spatial feature fusion mechanism includes: Calculate the spatial feature vectors of each region in different modal images; Based on the spatial feature vectors, use attention weights to determine the regions of different modal images that are valuable for diagnosis; Fuse the key region features of different modal images according to the attention weights to highlight the key structures and lesion sites of the urinary organs. Its characteristics lie in 3. A method for enhanced analysis of urological medical images according to claim 1, that in the step of designing the multi-modal spatio-temporal fusion enhancement module, the combination process of the optical flow estimation and the temporal convolutional network includes: The optical flow estimation algorithm calculates the motion vectors of pixel points between adjacent frames in the dynamic image sequence; ​ ​ ​ ​ ​ ​ 4. A method for enhanced analysis of urological medical images according to claim 1, ​ ​ ​ ​ Take the motion vector as the input of the temporal convolutional network, and capture the motion information of the organ tissues in the image sequence in the time dimension through the convolutional operation of the temporal convolutional network; Perform spatio-temporal fusion enhancement on the dynamic image sequence according to the motion information to ensure the spatio-temporal consistency of the image under dynamic enhancement. In the step of introducing the adaptive enhancement mechanism based on the physiological model, the physiological model includes the glomerular filtration rate model of the kidney, the volume-pressure relationship model of the bladder, etc., and the adaptive enhancement mechanism includes:

5. A method for enhanced analysis of urological medical images according to claim 1, Extract the patient's physiological parameters, such as glomerular filtration rate value, bladder volume and pressure value, etc.; According to the physiological parameters and the corresponding physiological model, determine the key physiological state of the urinary system organs; Based on the key physiological state, adjust the parameters and strategies of image enhancement so that the enhanced image can more accurately reflect the physiological function of the organs. In the step of constructing the interpretable enhancement decision-making mechanism, it includes: Record the key information such as the network structure, parameter changes of the quantum neural network, and the fusion weights of the multi-modal spatio-temporal fusion enhancement module during the image enhancement process; Display the influence of different enhancement strategies on each region of the image and the role of the physiological model in adaptive enhancement through visualization technology; Combine the patient's multi-source heterogeneous data to generate a report containing the basis for enhancement decision-making and result interpretation to assist doctors in making diagnostic decisions.

6. A method for enhanced analysis of urological medical images according to claim 1, In the step of collecting multi-source heterogeneous data, it also includes the steps of quality assessment and cleaning of the data to ensure the accuracy and reliability of the input data. The quality assessment and cleaning include: Check the integrity and clarity of the image data, and eliminate the images with serious artifacts or missing data; Unify the formats of the data such as the patient's electronic medical records and test reports, and correct the error information to ensure the consistency and usability of the data. In the step of comparing and analyzing the enhanced image with the original image after the interpretable enhancement decision-making mechanism, to verify the enhancement effect and provide a reference for the further optimization of the model. The comparison and analysis includes: Quantitatively evaluate from multiple aspects such as the contrast, clarity, noise level, and prominence of key structures of the image; 7. A method for enhanced analysis of urological medical images according to claim 1, Subjectively evaluate the enhanced image by professional doctors and give feedback on the diagnostic value and clinical application effect. In the step of customizing the image enhancement model for different patient groups to meet the differences in the urinary system anatomical structures, physiological functions, and disease characteristics of different patients. The customization includes: Collect a large amount of multi-source heterogeneous data of patients and conduct classification and statistics to analyze the characteristics of different patient groups; According to the characteristics of the patient group, adjust and optimize the parameters of the quantum neural network and the multi-modal spatio-temporal fusion enhancement module; Generate corresponding image enhancement models for each patient group to achieve personalized enhancement.

8. A method for enhanced analysis of urological medical images according to claim 1, ​ ​ ​ ​ ​ ​ ​ ​ 9. A method for enhanced analysis of urological medical images according to claim 1, ​ ​ ​ ​ ​ ​ ​ ​ ​

Images