Kidney physiological age prediction method and device based on multi-modal MRI (Magnetic Resonance Imaging) image

Abstract

The invention provides a kidney physiological age prediction method and device based on a multi-mode MRI image, and the method comprises the steps: carrying out the semantic segmentation of a kidney MRI image, and obtaining a cortical medullary segmentation mask of the kidney image; performing high-throughput radiomics feature extraction on the cortical medullary part of the kidney MRI image, and performing combined screening on radiomics parameters to obtain a radiomics parameter with the highest correlation with kidney physiological age prediction in a combined result as a final radiomics parameter; and inputting the final omics parameters as features into a neural network model to obtain a predicted age. According to the method, different tissue areas of the kidney are automatically segmented, high-throughput radiomics characteristics are extracted, and on the basis, the kidney physiological age is predicted.

Description

Technical Field

[0001] The present invention relates to the fields of image semantic segmentation, multimodal image analysis and artificial intelligence technology, and in particular to a method and device for predicting kidney physiological age based on multimodal MRI images. Background Art

[0002] Existing technologies mainly rely on two major methods in kidney health assessment: biomarkers and imaging examinations. Biomarkers provide information on kidney function status by measuring serum creatinine (sCr) and estimated glomerular filtration rate (eGFR). sCr is a metabolite that reflects the kidney's filtration capacity, while eGFR combines factors such as age, gender and body surface area to more accurately estimate kidney function status. Imaging examinations include ultrasound, CT scans and MRI. Ultrasound uses real-time dynamic imaging, is non-invasive and radiation-free, and is often used for preliminary assessment of kidney morphology and abnormalities (such as stones, cysts, and masses); CT scans provide high-resolution cross-sectional images, which are suitable for the analysis of complex anatomical structures, especially in the qualitative and quantitative assessment of kidney tumors; MRI uses strong magnetic fields and radio frequency waves to generate images with high soft tissue contrast, which can detect subtle differences between tissues and evaluate renal blood flow and lesion characteristics. These methods evaluate kidney health from multiple levels through their own technical means, and are usually used in combination to ensure the accuracy and comprehensiveness of diagnosis.

[0003] Insufficient timeliness of biomarkers: Although sCr and eGFR are commonly used markers of renal function, they usually do not change significantly until significant renal damage occurs. This lag means that in many cases, especially in the early stages of chronic kidney disease, patients may have unknown renal damage. In addition, because these markers are affected by muscle mass, diet, and other individual differences, they may lead to false negative or false positive results in some populations, limiting their effectiveness in individualized assessments.

[0004] Limitations of imaging examinations: Ultrasound, a traditional imaging method, has low resolution, which limits the observation of fine structures. Although CT scanning provides high anatomical details, its radiation risk limits its application in young patients or pregnant women who require long-term follow-up. In addition, CT is limited in assessing renal function and is more used for the detection of structural abnormalities. Although MRI does not involve radiation and can provide excellent soft tissue contrast, traditional MRI analysis methods often rely on subjective evaluation and lack quantitative functional evaluation. Current artificial intelligence technology focuses on the simple segmentation of renal images, and the analysis of renal health status through renal structure is still based on manual analysis.

[0005] Lack of ability to identify early lesions: Existing assessment methods can only provide reliable information in the middle and late stages of kidney disease, and lack the ability to actively monitor and identify subtle pathological changes in the kidneys. This leads to insufficient intervention in the early stages of the disease and missed opportunities for optimal treatment.

[0006] Limited integrated analysis capabilities: Biomarkers and imaging examinations in existing technologies are usually performed separately, without effectively integrating information from different data sources to provide a comprehensive kidney health status. This lack of integration leads to an insufficient overall understanding of the dynamic changes in kidney lesions, which limits the comprehensive management and decision-making in complex pathological conditions, especially in the case of coexistence of multiple diseases (such as diabetes, hypertension and chronic kidney disease).

[0007] Specifically, the prior art has proposed a method for predicting kidney age based on CT images, which predicts age by using the renal cortex and medulla volume determined by CT images. However, the radiation risk in acquiring CT images limits the application scenarios of this technology. Therefore, there is an urgent need for a method for predicting kidney age based on non-radiation images, such as a method for predicting kidney age based on MRI images. However, if MRI images are input into this method for predicting kidney age, the following three technical problems will arise: First, MRI images are not the first choice for measuring the renal cortex and medulla volume of the kidney, because CT images have high spatial resolution and can clearly display the subtle structure and morphology of the kidney, which is convenient for accurate volume measurement. CT images can perform detailed data analysis through multi-plane reconstruction and three-dimensional imaging. Therefore, inputting MRI images into the method for predicting kidney age based on CT images will result in a decrease in prediction accuracy; second, the amount of information in MRI images is much greater than that in CT images. For example, MRI images have richer soft tissue contrast, multi-parameter imaging, and functional and metabolic information, while CT images can only provide limited soft tissue contrast and focus mainly on tissue density. MRI's various imaging sequences and functional imaging can provide more comprehensive and diverse biological tissue information for imaging omics analysis, while CT is relatively lacking in soft tissue detail identification and diverse information. Therefore, directly applying the kidney age prediction method based on CT images will result in information waste, but MRI images have a lot of information, not all of which is related to kidney age prediction. That is, how to integrate and screen the information of MRI images to improve the accuracy of kidney age prediction of artificial intelligence models and make the kidney age output by artificial intelligence models interpretable is the current technical challenge. Summary of the invention

[0008] This application proposal aims to solve the problem that existing kidney health assessment technology cannot effectively identify early kidney damage, and it is difficult to integrate kidney structure and function information for comprehensive assessment. Traditional biomarkers such as sCr and eGFR often only show changes when kidney function decreases significantly, and lack sensitivity to early, microscopic kidney lesions. Although existing imaging methods can provide structural information, they fail to effectively quantify and evaluate the functional status of the kidneys. The present invention proposes an automated model that completes the entire process of "automatic segmentation-omics extraction-kidney health assessment".

[0009] In view of the shortcomings of existing technologies, such as Figure 5 As shown, the present invention proposes a method for predicting kidney physiological age based on multimodal MRI images, which includes:

[0010] In the initial step, multiple renal MRI image pairs of healthy people are obtained, each pair of renal MRI images includes a renal T1-weighted image and a renal T1-mapped image, and each pair of renal MRI images has been annotated with the corresponding actual age;

[0011] A semantic segmentation step, performing semantic segmentation on each pair of kidney MRI images to obtain a first cortex-medullary segmentation mask of the kidney weighted image and a second cortex-medullary segmentation mask of the kidney mapping image;

[0012] a feature extraction step, performing radiomics acquisition on the renal T1 weighted image and the renal T1 mapping image according to the first cortex-medullary segmentation mask and the second cortex-medullary segmentation mask, to obtain a first set of radiomics parameters of the renal T1 weighted image and a second set of radiomics parameters of the renal T1 mapping image;

[0013] A combined training step, combining the first group of imaging omics parameters and the second group of imaging omics parameters, and screening the combined results to obtain the imaging omics parameters with the highest correlation with the prediction of kidney physiological age in the combined results as the final omics parameters; inputting the final omics parameters into the neural network model as the features of the kidney MRI image pair to obtain the predicted age, constructing a loss function with the predicted age and the actual age marked by the kidney MRI image pair, training the neural network model, and obtaining a kidney physiological age prediction model;

[0014] In the age prediction step, the kidney MRI image pair to be predicted for physiological age is semantically segmented and the final omics parameters are extracted. The kidney physiological age prediction model predicts the kidney physiological age according to the final omics parameters of the kidney MRI image pair to be predicted for physiological age. When the difference between the predicted kidney physiological age and the actual age of the kidney MRI image to be predicted for physiological age is greater than a threshold, it is determined that there is kidney damage.

[0015] The method for predicting kidney physiological age based on multimodal MRI images, wherein the first group of radiomics parameters and the second group of radiomics parameters both include first-order radiomics parameters and high-order radiomics parameters, and the first group of radiomics parameters also includes shape radiomics parameters;

[0016] Among them, the first-order radiomics parameters include: the mean and median of the grayscale values ​​of the voxels in the renal cortex and medulla; the high-order radiomics parameters include: the grayscale co-occurrence matrix, grayscale dependency matrix, grayscale run matrix, grayscale size area matrix, and neighborhood grayscale difference matrix of the renal cortex and medulla. Shape-omics parameters include: the volume of the renal cortex and medulla, the long diameter of the kidney, surface flatness, surface area-volume ratio, etc.

[0017] In the method for predicting kidney physiological age based on multimodal MRI images, the process of screening the combined results in the combined training step includes:

[0018] The first group of imaging omics parameters and the second group of imaging omics parameters are classified to obtain a kidney shape parameter data set, a T1 weighted image feature parameter data set, and a T1 mapping image feature parameter data set, and the above three are combined to obtain multiple data sets;

[0019] Taking multiple data sets as input and predicting the physiological age of kidney as output, the neural network model is trained by interpretable machine learning methods, and the data set with the best prediction performance is found as the final omics parameter;

[0020] The age prediction step also includes outputting the final omics parameters of the kidney MRI image to be predicted for physiological age as interpretable features for predicting the kidney physiological age.

[0021] The method for predicting kidney physiological age based on multimodal MRI images, wherein the multiple data sets include:

[0022] The kidney shape parameter data set, the T1 weighted image feature parameter data set and the T1 mapping image feature parameter data set are combined in pairs and / or in triplicate to obtain the multiple data sets. The first group of imaging omics parameters and the second group of imaging omics parameters also include a corticomedullary differentiation (CMD) feature, and the more obvious the CMD feature is, the more it represents that the kidney is normal and healthy.

[0023] like Figure 6 As shown, the present invention also proposes a kidney physiological age prediction device based on multimodal MRI images, which includes:

[0024] The initial module obtains multiple renal MRI image pairs of healthy people, each pair of renal MRI images includes a renal T1-weighted image and a renal T1-mapped image, and each pair of renal MRI images has been annotated with the corresponding physiological age;

[0025] A semantic segmentation module performs semantic segmentation on each pair of kidney MRI images to obtain a first cortex-medullary segmentation mask of the kidney weighted image and a second cortex-medullary segmentation mask of the kidney mapping image;

[0026] a feature extraction module, performing radiomics acquisition on the renal T1 weighted image and the renal T1 mapping image according to the first cortex-medullary segmentation mask and the second cortex-medullary segmentation mask, to obtain a first set of radiomics parameters of the renal T1 weighted image and a second set of radiomics parameters of the renal T1 mapping image;

[0027] A combined training module combines the first group of imaging omics parameters and the second group of imaging omics parameters, and screens the combined results to obtain the imaging omics parameters with the highest correlation with the prediction of kidney physiological age in the combined results as the final omics parameters; the final omics parameters are input into the neural network model as the features of the kidney MRI image pair to obtain the predicted age, a loss function is constructed with the predicted age and the physiological age annotated by the kidney MRI image pair, and the neural network model is trained to obtain a kidney physiological age prediction model;

[0028] The age prediction module performs semantic segmentation on the kidney MRI image pair to be predicted for physiological age and extracts the final omics parameters. The kidney physiological age prediction model predicts the kidney physiological age according to the final omics parameters of the kidney MRI image pair to be predicted for physiological age.

[0029] The device for predicting kidney physiological age based on multimodal MRI images, wherein the first group of radiomics parameters and the second group of radiomics parameters both include first-order radiomics parameters and high-order radiomics parameters, and the first group of radiomics parameters also includes shape radiomics parameters;

[0030] Among them, the first-order radiomics parameters include: the mean and median of the grayscale values ​​of the voxels in the renal cortex and medulla; the high-order radiomics parameters include: the grayscale co-occurrence matrix, grayscale dependency matrix, grayscale run matrix, grayscale size area matrix, and neighborhood grayscale difference matrix of the renal cortex and medulla. Shape-omics parameters include: the volume of the renal cortex and medulla, the long diameter of the kidney, surface flatness, surface area-volume ratio, etc.

[0031] In the device for predicting kidney physiological age based on multimodal MRI images, the process of screening the combined results in the combined training module includes:

[0032] The first group of imaging omics parameters and the second group of imaging omics parameters are classified to obtain a kidney shape parameter data set, a T1 weighted image feature parameter data set, and a T1 mapping image feature parameter data set, and the above three are combined to obtain multiple data sets;

[0033] Taking multiple data sets as input and predicting the physiological age of kidney as output, the neural network model is trained by interpretable machine learning methods, and the data set with the best prediction performance is found as the final omics parameter;

[0034] The age prediction module also includes outputting the final omics parameters of the kidney MRI image to be predicted for physiological age as interpretable features for predicting the kidney physiological age.

[0035] The present invention also proposes an electronic device, which includes the above-mentioned kidney physiological age prediction device. The electronic device may be connected to an information display device, and the information display device is used to display the kidney physiological age according to display parameters and attributes set by the user or through an artificial intelligence model.

[0036] The present invention also proposes a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the kidney physiological age prediction methods are implemented.

[0037] The present invention also proposes a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the steps of any of the kidney physiological age prediction methods are implemented.

[0038] It can be seen from the above scheme that the advantages of the present invention are:

[0039] Compared with the prior art, this invention has significant beneficial effects in kidney health assessment:

[0040] 1. Improved early detection capabilities: Advanced imaging genomics and deep learning technologies can capture microscopic changes in early renal injury, providing an opportunity to intervene before clinical symptoms appear.

[0041] 2. Comprehensive assessment and personalized diagnosis: Combining structural and functional information, it provides a comprehensive assessment of kidney health status, supporting more accurate clinical diagnosis and personalized treatment decisions.

[0042] 3. No radiation risk and safety: Using MRI to obtain imaging data, there is no radiation risk and it can be repeatedly used for users who need frequent monitoring, especially suitable for sensitive groups such as children and pregnant women.

[0043] 4. Improve diagnostic accuracy and efficiency: Automated image analysis methods reduce reliance on manual experience, reduce subjective errors, and improve diagnostic efficiency and consistency.

[0044] 5. Promote clinical practice of prevention and early treatment: By providing a comparison of the physiological age of the kidney with the actual age of the user (KAG), it helps to identify high-risk groups and guide early treatment measures, thereby effectively delaying further deterioration of renal function. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a structural diagram of the model of the present invention;

[0046] Figure 2 This is the result of semantic segmentation of kidney image;

[0047] Figure 3 This is a comparison chart of data prediction performance;

[0048] Figure 4 This is a schematic diagram of the risk of cardiovascular events caused by the change of renal age difference over time;

[0049] Figure 5 is a flow chart of the method of the present invention;

[0050] Figure 6 It is a module diagram of the device of the present invention;

[0051] Figure 7 This is a schematic diagram of the structure of a first electronic device of the present invention;

[0052] Figure 8 This is a schematic diagram of the application environment structure of the first electronic device of the present invention;

[0053] Fig. 9 It is a schematic diagram of the structure of a second electronic device of the present invention.

[0054] Reference numerals:

[0055] A-First electronic device;

[0056] B-Renal physiological age prediction device based on multimodal MRI images;

[0057] C-data acquisition equipment;

[0058] D-information display device;

[0059] 1000 - second electronic device;

[0060] Ⅰ-computational unit;

[0061] II-ROM;

[0062] III-RAM;

[0063] IV-bus;

[0064] V-interface;

[0065] VI - input unit;

[0066] VII-output unit;

[0067] VIII- Storage medium;

[0068] Ⅸ-Communication unit. DETAILED DESCRIPTION

[0069] It should be noted that, in this application, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0070] Without more constraints, an element defined by the phrase "comprising a..." does not exclude the existence of other identical elements in the process, method, article or apparatus comprising the element.

[0071] The processor described in the present invention is the control center of the electronic device, which can be a processor or a general term for multiple processing elements. For example, it can be one or more central processing units (CPUs), or application specific integrated circuits (ASICs), or one or more integrated circuits configured to implement the embodiments of the present invention, such as one or more microprocessors (digital signal processors, DSPs), or one or more field programmable gate arrays (FPGAs).

[0072] Optionally, the processor can perform various functions of the electronic device by running or executing a software program stored in the memory, and calling data stored in the memory.

[0073] In a specific implementation, as an embodiment, the processor may include one or more CPUs. Each of these processors may be a single-core processor (single-CPU) or a multi-core processor (multi-CPU). The processor here may refer to one or more devices, circuits, and / or processing cores for processing data (such as computer program instructions). Electronic devices may include: servers, desktop computers, laptops, smart phones, tablet computers, embedded computers, etc., wherein the embedded computers include vehicles and robots, etc.

[0074] The memory is used to store the software program for executing the solution of the present invention, and the execution is controlled by the processor. The specific implementation method can refer to the above method embodiment, which will not be repeated here.

[0075] It should be noted that the structure of the electronic device shown in the drawings of the present invention does not constitute a limitation thereto, and the actual knowledge structure recognition device may include more or fewer components than shown in the drawings, or a combination of certain components, or a different arrangement of components.

[0076] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0077] It should also be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0078] In the present invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0079] It should also be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0080] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0081] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0082] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0083] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0084] When conducting research on kidney health assessment, the present inventors found that the existing technology has significant defects in accurately detecting early renal damage, which is mainly due to the fact that traditional methods only rely on biomarkers such as serum creatinine (sCr) and estimated glomerular filtration rate (eGFR). These markers are usually not able to effectively reflect early or subtle damage before the patient has obvious renal function impairment. In addition, although imaging detection methods can provide certain structural information, they lack combined analysis with functional status (biomarkers). However, these methods fail to fully consider the microscopic changes in kidney structure, which is particularly important for the early diagnosis and prevention of chronic kidney disease.

[0085] After in-depth research, the inventors realized that the key to solving this problem is to combine the structural and functional information of the kidneys and obtain a more comprehensive assessment through the integrated analysis of multimodal imaging technology and machine learning algorithms. Specifically, the inventors found that multimodal MRI images can provide rich structural and tissue information, which can be converted into quantitative data through imaging omics feature extraction to reflect the microscopic changes in kidney health.

[0086] Therefore, on this basis, the present invention uses deep learning technology to develop a new solution: the present invention trains a complex neural network model KAGE-NET, extracts detailed imaging genomics features by automatically segmenting different tissue regions of the kidney, and predicts the "kidney physiological age" on this basis. These features include shape features and first-order statistical features (such as volume, length, mean and variance of brightness, etc.), and high-order features (grayscale co-occurrence matrix, grayscale run matrix, etc.), so as to more accurately reflect the physiological age of the kidney (K-AGE). By combining with traditional biomarkers, this model can transcend the limitations of current diagnostic methods and provide a perspective on the health status of the kidney. The concept of kidney physiological age gap (KAG) is proposed, and the difference between "kidney physiological age" and actual age is used to reflect the acceleration of kidney aging: positive values ​​represent accelerated kidney aging, and vice versa. Finally, a comprehensive diagnostic tool integrating structural and functional evaluation is realized, which improves the accuracy of kidney health evaluation and early detection ability.

[0087] In order to achieve the above technical effects, the present invention proposes the following key technical points:

[0088] Key point 1: Multimodal MRI images combined with automatic segmentation and omics feature extraction by deep learning. By automatically segmenting various tissue regions of the kidney (such as the cortex and medulla) and extracting relevant radiomic features (such as volume, surface area, and pixel value statistical indicators), the structural changes and functional status of the kidney can be quantitatively evaluated, and the ability to detect early renal damage can be improved.

[0089] Key point 2: Prediction model of kidney physiological age (K-AGE). Multiple imaging omics features are integrated through machine learning algorithms to construct a model for predicting kidney physiological age. The predicted kidney physiological age can be compared with the user's biological age (actual age), and the kidney age difference (KAG) indicator is used to achieve a personalized assessment of the degree of kidney aging and health status. It should be noted that the actual kidney age of the user to be tested involved in the present invention is the user's biological age. For example, if the biological age of a user with normal and healthy kidneys is 20 years old, then the biological age of his kidneys should also be 20 years old, and KAG is 20 minus 20, which is 0; if the age of a user with kidney disease is 20 years old, then the biological age of his kidneys is 30 years old, and KAG is 30 minus 20, which is 10.

[0090] Key point 3: Potential tools for personalized medicine. A comprehensive set of tools that can be used for personalized diagnosis and treatment in clinical settings has been established to quantify the complex impact of different risk factors on kidney health. This has promoted personalized treatment plans and efficacy monitoring for patients with chronic kidney disease (CKD) and metabolic syndrome.

[0091] In order to make the above features and effects of the present invention more clearly and understandably described, embodiments are given below and described in detail with reference to the accompanying drawings. This specification discloses one or more embodiments that include the features of the present invention. The disclosed embodiments are only for illustration. The scope of protection of the present invention is not limited to the disclosed embodiments, and the present invention is defined by the attached claims.

[0092] This invention introduces an automated radiomics analysis method based on multimodal MRI images and artificial intelligence technology, and achieves a comprehensive assessment of kidney health by creating a model of kidney "physiological age (K-AGE)". This innovative technology can reflect subtle changes in kidney structure and function early and accurately, surpassing the limitations of traditional methods, and providing information about the rate of kidney aging and potential lesions. Using this method, medical staff can monitor and manage kidney disease more accurately, develop more personalized treatment strategies for users, and significantly improve the effectiveness and timeliness of clinical intervention. Figure 1 As shown, specifically the method of the present invention comprises:

[0093] Step 1:

[0094] Obtain the renal T1 weighted and T1 mapping datasets (DICOM format) of renal healthy users and their actual ages, and convert the renal T1 weighted and T1 mapping datasets (DICOM format) into NIFTI format suitable for machine reading and processing; Figure 2 As shown, renal T1-weighted and renal T1-mapping data are two different imaging methods under MRI imaging.

[0095] Step 2:

[0096] The T1-weighted data were divided into a training set and a test set. The corticomedullary regions of the left and right kidneys in the training set were manually annotated to establish a renal corticomedullary model for automatic segmentation of renal T1-weighted images.

[0097] Step 3:

[0098] Automatically segment the images of the test set to obtain the automatic segmentation masks of the renal cortex and medulla;

[0099] Step 4:

[0100] The T1-weighted image mask is matched with the T1 mapping data set one by one, and a radiologist performs manual inspection to see whether the two overlap. If so, execute step 6, otherwise, execute step 5;

[0101] Step 5:

[0102] Manually annotate and correct the images that do not meet the requirements, so that the T1 weighted image mask and the weighted image mask of the T1 mapping dataset can establish a corresponding relationship, so that the T1 weighted image mask can correctly describe the renal cortex and medulla areas of the T1 mapping dataset, improve the accuracy of subsequent image feature extraction, and improve the accuracy of renal physiological age prediction;

[0103] Step 6:

[0104] The value range of T1-weighted images will fluctuate with changes in the instrument, so it is necessary to standardize the T1-weighted data, and finally obtain the T1-weighted standardized data set and the T1 mapping data set;

[0105] Step 7:

[0106] Radiomics acquisition was performed for the T1 mapping dataset, including first-order radiomics parameters (mean and median of the grayscale values ​​of the voxels in the renal cortex and medulla) and higher-order radiomics parameters (including the grayscale co-occurrence matrix, grayscale dependency matrix, grayscale run matrix, grayscale size region matrix, neighborhood grayscale difference matrix, etc. in the renal cortex and medulla);

[0107] Step 8:

[0108] Radiomics were collected for T1-weighted standardized data sets, including shape features (kidney volume, long diameter, surface area-to-volume ratio, etc.), first-order radiomics parameters (mean and median of voxel grayscale values ​​in the kidney cortex and medulla regions, etc.) and higher-order radiomics parameters (including grayscale co-occurrence matrix, grayscale dependency matrix, grayscale run matrix, grayscale size region matrix, neighborhood grayscale difference matrix, etc. in the kidney cortex and medulla regions);

[0109] Step 9:

[0110] After screening and combining, cleaning, dimension reduction and standardization of the two groups of T1 radiomics parameters, a standard data set of multiple radiomics parameters was established;

[0111] Combination of kidney datasets: The features of kidney images are divided into three types of imaging datasets according to their specific physical meanings: shape parameter imaging dataset (kidney volume, thickness, long diameter, etc.) (Shape), T1 weighted image feature parameter imaging dataset (T1W), T1 mapping influence feature parameter imaging dataset (T1M); and four other imaging datasets obtained by free combination: WShape (T1W+Shape), MShape (T1M+Shape), T1 (T1W+T1M), ALL (T1W+T1M+Shape).

[0112] Step 10:

[0113] Obtain a model set including random forest, support vector machine, XGboost and other neural network models; obtain the above-mentioned imaging omics dataset and the actual age, input the imaging omics dataset into the neural network in the model set to predict kidney physiological age; it should be noted that the actual age is not input into the neural network, but plays a supervisory role, and iterative ensemble learning of each neural network model is performed through supervision of multiple different machine learning methods, and by comparing functional indicators, the neural network model with the best prediction performance (best functional indicators) in the model set is found, that is, the final "kidney physiological age" prediction model is obtained.

[0114] By using the SHAP interpretable machine learning method based on the Shapely algorithm, it was found that the data prediction performance of the MShape and WShape data sets was the best. Among them, some features were screened out and had obvious implications for the prediction of "kidney physiological age" ( Figure 3 ). The median T1 intensity value in the cortical region was one of the most significant features, showing a wide range of SHAP values. Higher eigenvalues ​​(red) increased the model prediction, while lower eigenvalues ​​(blue) decreased the prediction. The difference in median T1 intensity values ​​between the medulla and cortex also showed a significant effect, with a wide range of SHAP values. Medulla volume and kidney surface-to-volume ratio showed different patterns of effect.

[0115] Clinical studies have shown that an increase in the median T1 in the renal cortex usually means a higher degree of renal fibrosis. This fibrotic process not only increases the rigidity and density of renal tissue, but also impairs the glomerular filtration function, thereby accelerating renal aging. Therefore, an increase in the median T1 is logically associated with an increase in the model's predicted value. In addition, on T1 images of normal healthy kidneys, the cortex and medulla can be clearly distinguished, a feature called corticomedullary differentiation (CMD). A decrease in the difference in T1 median intensity values ​​between the medulla and cortex means that CMD is not obvious, which is another key signal of decreased renal function. In the case of renal injury, the structural boundary between the cortex and medulla is no longer obvious, reflecting pathological changes in the internal structure of the kidney. In terms of renal shape parameters, a decrease in medullary volume and an increase in the renal surface area-to-volume ratio mean renal atrophy and increased surface irregularity, both of which are biological markers of renal aging and disease progression. A decrease in renal volume is usually accompanied by a decrease in function, while an increase in the surface area-to-volume ratio means irregularities in the renal surface that may be caused by fibrosis or other pathological changes.

[0116] After excluding participants from the training set, we recorded 254 cardiovascular events during a median follow-up of 13.7 years (IQR 12.9-14.4 years). Figure 4As shown in the study, participants who were in the top 10% or bottom 10% of the renal age difference (KAG) were defined as "extreme KAG", and these individuals showed a significantly higher cumulative risk of cardiovascular events. Compared with individuals with normal KAG, the hazard ratio for cardiovascular events in the extreme KAG group was 1.45 (95% CI 1.10-1.90), after adjusting for age, sex, and further adjusting for traditional cardiovascular risk factors such as hypertension, diabetes, and overweight (defined as BMI over 25 kg / m 2 ), the hazard ratio was 1.37 (95% CI 1.04-1.80). This effect remained statistically significant even after adjustment for baseline eGFR.

[0117] The following is a system embodiment corresponding to the above method embodiment. This embodiment can be implemented in conjunction with the above embodiment. The relevant technical details mentioned in the above embodiment are still valid in this embodiment. In order to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied in the above embodiment.

[0118] The present invention also proposes a kidney physiological age prediction device based on multimodal MRI images, which includes:

[0119] The initial module obtains multiple renal MRI image pairs of healthy people, each pair of renal MRI images includes a renal T1-weighted image and a renal T1-mapped image, and each pair of renal MRI images has been annotated with the corresponding physiological age;

[0120] A semantic segmentation module performs semantic segmentation on each pair of kidney MRI images to obtain a first cortex-medullary segmentation mask of the kidney weighted image and a second cortex-medullary segmentation mask of the kidney mapping image;

[0121] a feature extraction module, performing radiomics acquisition on the renal T1 weighted image and the renal T1 mapping image according to the first cortex-medullary segmentation mask and the second cortex-medullary segmentation mask, to obtain a first set of radiomics parameters of the renal T1 weighted image and a second set of radiomics parameters of the renal T1 mapping image;

[0122] A combined training module combines the first group of imaging omics parameters and the second group of imaging omics parameters, and screens the combined results to obtain the imaging omics parameters with the highest correlation with the prediction of kidney physiological age in the combined results as the final omics parameters; the final omics parameters are input into the neural network model as the features of the kidney MRI image pair to obtain the predicted age, a loss function is constructed with the predicted age and the physiological age annotated by the kidney MRI image pair, and the neural network model is trained to obtain a kidney physiological age prediction model;

[0123] The age prediction module performs semantic segmentation on the kidney MRI image pair to be predicted for physiological age and extracts the final omics parameters. The kidney physiological age prediction model predicts the kidney physiological age according to the final omics parameters of the kidney MRI image pair to be predicted for physiological age.

[0124] The device for predicting kidney physiological age based on multimodal MRI images, wherein the first group of radiomics parameters and the second group of radiomics parameters both include first-order radiomics parameters and high-order radiomics parameters, and the first group of radiomics parameters also includes shape radiomics parameters;

[0125] Among them, the first-order radiomics parameters include: the mean and median of the grayscale values ​​of the voxels in the renal cortex and medulla; the high-order radiomics parameters include: the grayscale co-occurrence matrix, grayscale dependency matrix, grayscale run matrix, grayscale size area matrix, and neighborhood grayscale difference matrix of the renal cortex and medulla. Shape-omics parameters include: the volume of the renal cortex and medulla, the long diameter of the kidney, surface flatness, surface area-volume ratio, etc.

[0126] In the device for predicting kidney physiological age based on multimodal MRI images, the process of screening the combined results in the combined training module includes:

[0127] The first group of imaging omics parameters and the second group of imaging omics parameters are classified to obtain a kidney shape parameter data set, a T1 weighted image feature parameter data set, and a T1 mapping image feature parameter data set, and the above three are combined to obtain multiple data sets;

[0128] Taking multiple data sets as input and predicting the physiological age of kidney as output, the neural network model is trained by interpretable machine learning methods, and the data set with the best prediction performance is found as the final omics parameter;

[0129] The age prediction module also includes outputting the final omics parameters of the kidney MRI image to be predicted for physiological age as interpretable features for predicting the kidney physiological age.

[0130] like Figure 7 As shown, the present invention further proposes a first electronic device A in another embodiment, comprising the aforementioned kidney physiological age prediction device.

[0131] like Figure 8 As shown, the first electronic device A can also be connected to the data acquisition device C and the information display device D through a wired or wireless information transmission scheme, the data acquisition device C is used to collect and obtain the kidney MRI image to be predicted for physiological age, and the information display device D is used to display the kidney physiological age prediction results, interpretable features, semantic segmentation results, etc. obtained by the analysis of the present invention.

[0132] The information display device D can process the data output by the first electronic device A based on the information display mechanism to improve the readability of the data output by the first electronic device A. The information display mechanism can be manually preset, for example, the data output by the first electronic device A is visually displayed, which can be based on the display parameters and / or attributes set by the user. The display parameters can be, for example, the display data range, and the display attributes can be, for example, the display font, color, whether to scroll and play, etc. The user is presented with the key information specified by the user, such as the prediction results of the physiological age of the kidney, the interpretable features, the semantic segmentation results, etc. The user can understand this information more timely without having to access the secondary page or scroll the page, saving the user's operation. Or the information display mechanism can be an artificial intelligence AI display model, which can learn the user's key information based on the user's previous usage habits, such as viewing time, number of clicks, number of edits, etc., and then automatically present rich and necessary key information to the user.

[0133] The present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a readable storage medium. When the computer program is executed by a processor, the computer can execute the kidney physiological age prediction method provided by the above methods.

[0134] The present invention also proposes a storage medium VIII in another embodiment for storing a computer program for executing the method for predicting the physiological age of the kidney. It should be understood that the storage medium in the embodiment of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DRRAM).

[0135] Fig. 9 A schematic block diagram of a second electronic device 1000 that can be used to implement an embodiment of the present invention is shown. The second electronic device 1000 electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The second electronic device 1000 can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples, and are not intended to limit the implementation of the present invention described and / or required herein. The second electronic device 1000 may be the same or different from the first electronic device A.

[0136] The second electronic device 1000 includes a computing unit I, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory II (ROM) or a computer program loaded from a storage medium VIII into a random access memory (RAM) III. In RAM III, various programs and data required for the operation of the device 1000 can also be stored. The computing unit I, ROM II, and RAM III are connected to each other via a bus IV. An input / output (I / O) interface V is also connected to the bus IV.

[0137] A plurality of components in the second electronic device 1000 are connected to the I / O interface V, including: an input unit VI, such as a keyboard, a mouse, etc.; an output unit VII, such as various types of displays, speakers, etc.; a storage medium VIII, such as a disk, an optical disk, etc.; and a communication unit IX, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit IX allows the second electronic device 1000 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0138] The computing unit I may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of computing unit I include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit I performs the various methods and processes described above, such as method steps S1-S5. For example, in some embodiments, the method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage medium VIII. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 1000 via ROM II and / or a communication unit IX. When the computer program is loaded into RAM III and executed by the computing unit I, one or more steps of the method described above may be performed. Alternatively, in other embodiments, the computing unit I may be configured to execute the method in any other appropriate manner (e.g., by means of firmware).

[0139] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes, and they can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the illustrations shown and described herein.

Claims

1. A method for predicting kidney physiological age based on multimodal MRI images, characterized in that: include: In the initial step, multiple modality renal MRI image pairs of healthy people are obtained, each pair of renal MRI images includes a renal T1-weighted image and a renal T1-mapped image, and each pair of renal MRI images has been annotated with the corresponding actual age; A semantic segmentation step, performing semantic segmentation on each pair of kidney MRI images to obtain a first cortex-medullary segmentation mask of the kidney weighted image and a second cortex-medullary segmentation mask of the kidney mapping image; a feature extraction step, performing multimodal feature extraction on the renal T1 weighted image and the renal T1 mapping image according to the first cortex-medullary segmentation mask and the second cortex-medullary segmentation mask, to obtain a first set of imaging omics parameters of the renal T1 weighted image and a second set of imaging omics parameters of the renal T1 mapping image; A combined training step, combining the first group of imaging omics parameters and the second group of imaging omics parameters, and screening the combined results to obtain the imaging omics parameters with the highest correlation with the prediction of kidney physiological age in the combined results as the final omics parameters; inputting the final omics parameters into the neural network model as the features of the kidney MRI image pair to obtain the predicted age; constructing a loss function with the predicted age and the actual age marked by the kidney MRI image pair, training the neural network model, and obtaining a kidney physiological age prediction model; In the age prediction step, the kidney MRI image pair to be predicted for physiological age is semantically segmented and then the final omics parameters are extracted. The kidney physiological age prediction model predicts the kidney physiological age according to the final omics parameters of the kidney MRI image pair to be predicted for physiological age.

2. The method for predicting kidney physiological age based on multimodal MRI images according to claim 1, characterized in that: The first group of radiomics parameters and the second group of radiomics parameters both include first-order radiomics parameters and high-order radiomics parameters, and the first group of radiomics parameters also includes shape-omics parameters; Among them, the first-order imaging omics parameters include: the mean and median of the grayscale values ​​of the voxels in the renal cortex and medulla; the high-order imaging omics parameters include: the grayscale co-occurrence matrix, grayscale dependency matrix, grayscale run matrix, grayscale size area matrix, and neighborhood grayscale difference matrix of the renal cortex and medulla. The shape omics parameters include: the volume of the renal cortex and medulla, the long diameter of the kidney, surface flatness, surface area-volume ratio, etc.

3. The method for predicting kidney physiological age based on multimodal MRI images according to claim 1, characterized in that: The process of screening the combination results in the combination training step includes: The first group of imaging omics parameters and the second group of imaging omics parameters are classified to obtain a kidney shape parameter data set, a T1 weighted image feature parameter data set, and a T1 mapping image feature parameter data set, and the above three are combined to obtain multiple data sets; Taking multiple data sets and the actual age as input, predicting the physiological age of the kidney as output, training the neural network model through interpretable machine learning methods, and finding the data set with the best prediction performance as the final omics parameter; The age prediction step also includes outputting the final omics parameters of the kidney MRI image to be predicted for physiological age as interpretable features for predicting the kidney physiological age.

4. The method for predicting kidney physiological age based on multimodal MRI images as claimed in claim 3, characterized in that: The multiple data sets include: The kidney shape parameter data sets, the T1 weighted image feature parameter data sets and the T1 mapping image feature parameter data sets are combined in pairs and / or in triplicate to obtain the multiple data sets.

5. A device for predicting kidney physiological age based on multimodal MRI images, characterized in that: include: The initial module obtains multiple modality renal MRI image pairs of healthy people. Each pair of renal MRI images includes renal T1-weighted images and renal T1-mapped images, and each pair of renal MRI images has been labeled with the corresponding actual age. A semantic segmentation module performs semantic segmentation on each pair of kidney MRI images to obtain a first cortex-medullary segmentation mask of the kidney weighted image and a second cortex-medullary segmentation mask of the kidney mapping image; a feature extraction module, performing multimodal feature extraction on the renal T1 weighted image and the renal T1 mapping image according to the first cortex-medullary segmentation mask and the second cortex-medullary segmentation mask, to obtain a first set of imaging omics parameters of the renal T1 weighted image and a second set of imaging omics parameters of the renal T1 mapping image; A combined training module combines the first group of radiomics parameters with the second group of radiomics parameters, and screens the combined results to obtain the radiomics parameters with the highest correlation with the prediction of kidney physiological age in the combined results as the final radiomics parameters; The final omics parameters are input into a neural network model as features of the renal MRI image pair to obtain a predicted age, a loss function is constructed using the predicted age and the actual age annotated by the renal MRI image pair, and the neural network model is trained to obtain a renal physiological age prediction model; The age prediction module performs semantic segmentation on the kidney MRI image pair to be predicted for physiological age and extracts the final omics parameters. The kidney physiological age prediction model predicts the kidney physiological age according to the final omics parameters of the kidney MRI image pair to be predicted for physiological age.

6. The device for predicting kidney physiological age based on multimodal MRI images according to claim 5, characterized in that: The first group of radiomics parameters and the second group of radiomics parameters both include first-order radiomics parameters and high-order radiomics parameters, and the first group of radiomics parameters also includes shape-omics parameters; Among them, the first-order radiomics parameters include: the mean and median of the grayscale values ​​of the voxels in the renal cortex and medulla; the high-order radiomics parameters include: the grayscale co-occurrence matrix, grayscale dependency matrix, grayscale run matrix, grayscale size area matrix, and neighborhood grayscale difference matrix of the renal cortex and medulla. Shape-omics parameters include: the volume of the renal cortex and medulla, the long diameter of the kidney, surface flatness, surface area-volume ratio, etc.

7. The device for predicting kidney physiological age based on multimodal MRI images according to claim 5, characterized in that: The process of screening the combination results in the combination training module includes: The first group of imaging omics parameters and the second group of imaging omics parameters are classified to obtain a kidney shape parameter data set, a T1 weighted image feature parameter data set, and a T1 mapping image feature parameter data set, and the above three are combined to obtain multiple data sets; Taking multiple data sets as input and predicting the physiological age of kidney as output, the neural network model is trained by interpretable machine learning methods, and the data set with the best prediction performance is found as the final omics parameter; The age prediction module also includes outputting the final omics parameters of the kidney MRI image to be predicted for physiological age as interpretable features for predicting the kidney physiological age.

8. An electronic device, characterized in that: It includes a kidney physiological age prediction device as described in claims 5-7, and the electronic device is connected to an information display device, which is used to display the kidney physiological age according to display parameters and attributes set by the user or through an artificial intelligence model.

9. A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of the method for predicting the physiological age of kidney according to any one of claims 1 to 4 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for predicting kidney physiological age described in any one of claims 1 to 4 are implemented.

Images