Kidney status assessment apparatus, method, device, storage medium and program product

Abstract

Embodiments of the present application provide a kidney state evaluation device, method, equipment, storage medium and program product. In the device: a obtaining module obtains a kidney structure image, an RBF image and a T1Map; a registration module performs image registration on the kidney structure image, the RBF image and the T1Map to obtain a target kidney structure image, a target RBF image and a target T1Map; a segmentation module segments the target kidney structure image by using a trained kidney segmentation model to obtain a segmentation mask of the renal parenchyma, renal cortex and renal medulla respectively; a processing module determines the morphology of the renal parenchyma from the target kidney structure image, determines the blood perfusion values of the renal cortex and the renal medulla from the target RBF image, and determines the T1 relaxation times of the renal cortex and the renal medulla from the target T1Map based on the segmentation mask; and an evaluation module evaluates the kidney state according to the processing result. The scheme provided in the embodiments of the present application can improve the accuracy of kidney state evaluation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, and in particular, relates to a kidney state evaluation device, method, equipment, storage medium and program product. BACKGROUND

[0002] The kidney is one of the important organs of the human body, responsible for filtering waste and excess water from the blood, maintaining electrolyte balance, regulating blood pressure, and participating in the production of red blood cells.

[0003] The normality of kidney function is directly related to the health and quality of life of individuals. Among them, kidney fibrosis affects the metal ion transport, glucose metabolism, immune function, inflammation and cell mitosis of the organ, and is the main cause of organ lesions.

[0004] At present, the kidney evaluation scheme has the problem of poor accuracy. SUMMARY

[0005] In view of the above problems, the present application is proposed to provide an information recommendation method, a data processing method, an equipment and a storage medium which solve the above problems or at least partially solve the above problems.

[0006] The first aspect of the present application provides a kidney state evaluation device, comprising:

[0007] The acquisition module is configured to acquire a kidney structure map, a kidney blood perfusion map and a T1 relaxation time quantitative map of a user.

[0008] The registration module is configured to perform image registration on the kidney structure map, the kidney blood perfusion map and the T1 relaxation time quantitative map to obtain a target kidney structure map, a target kidney blood perfusion map and a target T1 relaxation time quantitative map.

[0009] The segmentation module is configured to segment the target kidney structure map by using a trained kidney segmentation model to obtain a first segmentation mask of kidney parenchyma, a second segmentation mask of kidney cortex and a third segmentation mask of kidney medulla.

[0010] The processing module is configured to determine the morphology of the kidney parenchyma from the target kidney structure map, determine the blood perfusion values of the kidney cortex and the kidney medulla from the target kidney blood perfusion map, and determine the T1 relaxation times of the kidney cortex and the kidney medulla from the target T1 relaxation time quantitative map based on the first segmentation mask, the second segmentation mask and the third segmentation mask.

[0011] The evaluation module is configured to evaluate the kidney state of the user according to the shape of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively.

[0012] In a second aspect of the present application, a kidney data processing method is provided, comprising:

[0013] obtaining a kidney structure image, a kidney blood perfusion image, and a T1 relaxation time quantitative image of a user;

[0014] performing image registration on the kidney structure image, the kidney blood perfusion image, and the T1 relaxation time quantitative image to obtain a target kidney structure image, a target kidney blood perfusion image, and a target T1 relaxation time quantitative image;

[0015] segmenting the target kidney structure image using the trained kidney segmentation model to obtain a first segmentation mask of renal parenchyma, a second segmentation mask of renal cortex, and a third segmentation mask of renal medulla;

[0016] determining the shape of the renal parenchyma from the kidney structure image based on the first segmentation mask, the second segmentation mask, and the third segmentation mask, determining the blood perfusion values of the renal cortex and the renal medulla respectively from the target kidney blood perfusion image, and determining the T1 relaxation times of the renal cortex and the renal medulla respectively from the target T1 relaxation time quantitative image.

[0017] In a third aspect of the present application, an electronic device is provided. The electronic device comprises a memory and a processor, wherein

[0018] The memory is configured to store a program.

[0019] The processor is coupled to the memory and is configured to execute the program stored in the memory to implement:

[0020] obtaining a kidney structure image, a kidney blood perfusion image, and a T1 relaxation time quantitative image of a user;

[0021] performing image registration on the kidney structure image, the kidney blood perfusion image, and the T1 relaxation time quantitative image to obtain a target kidney structure image, a target kidney blood perfusion image, and a target T1 relaxation time quantitative image;

[0022] segmenting the target kidney structure image using the trained kidney segmentation model to obtain a first segmentation mask of renal parenchyma, a second segmentation mask of renal cortex, and a third segmentation mask of renal medulla;

[0023] determine the morphology of the renal parenchyma from the renal structure image, determine the blood perfusion values of the renal cortex and the renal medulla respectively from the target renal blood perfusion image, and determine the T1 relaxation times of the renal cortex and the renal medulla respectively from the target T1 relaxation time quantitative image.

[0024] Optionally, the processor is further configured to implement:

[0025] evaluate the renal status of the user according to the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively.

[0026] In a fourth aspect, the present application provides a computer readable storage medium storing a computer program, which, when executed by a computer, can implement:

[0027] obtain a renal structure image, a renal blood perfusion image, and a T1 relaxation time quantitative image of a user;

[0028] perform image registration on the renal structure image, the renal blood perfusion image, and the T1 relaxation time quantitative image to obtain a target renal structure image, a target renal blood perfusion image, and a target T1 relaxation time quantitative image;

[0029] segment the target renal structure image using the trained renal segmentation model to obtain a first segmentation mask of the renal parenchyma, a second segmentation mask of the renal cortex, and a third segmentation mask of the renal medulla;

[0030] determine the morphology of the renal parenchyma from the renal structure image, determine the blood perfusion values of the renal cortex and the renal medulla respectively from the target renal blood perfusion image, and determine the T1 relaxation times of the renal cortex and the renal medulla respectively from the target T1 relaxation time quantitative image.

[0031] Optionally, the computer program, when executed by a computer, can further implement:

[0032] evaluate the renal status of the user according to the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively.

[0033] In a fifth aspect, the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements:

[0034] obtain a renal structure image, a renal blood perfusion image, and a T1 relaxation time quantitative image of a user;

[0035] perform image registration on the kidney structure map, the kidney blood perfusion map and the T1 relaxation time quantitative map to obtain a target kidney structure map, a target kidney blood perfusion map and a target T1 relaxation time quantitative map;

[0036] segment the target kidney structure map by using the trained kidney segmentation model to obtain a first segmentation mask of kidney parenchyma, a second segmentation mask of kidney cortex and a third segmentation mask of kidney medulla;

[0037] based on the first segmentation mask, the second segmentation mask and the third segmentation mask, determine the morphology of the kidney parenchyma from the kidney structure map, determine the blood perfusion values of the kidney cortex and the kidney medulla respectively from the target kidney blood perfusion map, and determine the T1 relaxation times of the kidney cortex and the kidney medulla respectively from the target T1 relaxation time quantitative map.

[0038] Optionally, the computer program, when executed by the processor, can further implement:

[0039] based on the morphology of the kidney parenchyma, the blood perfusion values of the kidney cortex and the kidney medulla respectively and the T1 relaxation times of the kidney cortex and the kidney medulla respectively, evaluate the kidney state of the user.

[0040] In the technical scheme provided by the embodiments of the present application, the segmentation model is used to automatically segment the kidney cortex and the kidney medulla in the kidney structure map, and based on the image registration technology and the model segmentation result, the evaluation indexes of the morphology (volume, surface area), blood perfusion and T1 relaxation time of the segmented different kidney regions are automatically obtained from the kidney blood perfusion map and the T1 relaxation time quantitative map, and then the kidney state is evaluated based on the extracted morphology, blood perfusion and T1 relaxation time. That is, the present scheme can comprehensively evaluate the kidney state from the morphology of the kidney, the blood perfusion and the T1 relaxation time of the kidney cortex and the kidney medulla respectively. Moreover, the image registration technology can ensure the accuracy of the quantitative information calculation and the kidney state evaluation. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical scheme in the embodiments of the present application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0042] Figure 1 the structure block diagram of the kidney state evaluation device provided by an embodiment of the present application;

[0043] Figure 2A flowchart of a kidney data processing method provided by an embodiment of the present application is shown in FIG. 1.

[0044] Figure 3 A flowchart of a kidney data processing method provided by an embodiment of the present application is shown in FIG. 1.

[0045] Figure 4 A structural block diagram of an electronic device provided by an embodiment of the present application is shown in FIG. 6. DETAILED DESCRIPTION

[0046] In order to enable persons skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by persons skilled in the art without creative work fall within the scope of protection of the present application.

[0047] In addition, in some of the processes described in the specification, claims, and accompanying drawings of the present application, a plurality of operations appear in a specific order, which can not be executed in the order in which they appear in the text or in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and the operations can be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in the text are used to distinguish different messages, devices, modules, etc., and do not represent the order of sequence, nor do "first" and "second" represent different types.

[0048] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation portal for user to choose authorization or refusal.

[0049] Structurally, the kidney is divided into two parts: the renal cortex and the renal medulla. The renal cortex is located in the outer layer of the kidney and is mainly composed of glomeruli and renal tubules, connected to the renal columns between the renal pyramids. The glomeruli are responsible for the initial filtration of blood, while the renal tubules further process the filtered liquid, recovering necessary substances and excreting waste. The renal medulla is a deep structure of the kidney, composed of more than ten renal pyramids, mainly responsible for filtering impurities and storing urine. In kidney imaging, morphological characteristics such as kidney volume play an important role in the monitoring and diagnosis of various diseases. For example, an increase in overall kidney volume is a key factor in the diagnosis of autosomal dominant polycystic kidney disease, and an disproportionate decrease in cortical volume relative to medullary volume is a manifestation of aging and chronic kidney disease.

[0050] The kidney is also richly supplied with blood. In a normal adult at rest, 1200 ml of blood flows through both kidneys per minute, which is 1 / 5 to 1 / 4 of the cardiac output. Of this, 80-90% of the blood is distributed in the renal cortex, provided by the afferent arteriole, and 10-20% is distributed in the outer and inner medulla, provided by the efferent arteriole. The commonly referred to as renal perfusion volume mainly refers to the renal cortical blood flow.

[0051] In addition, the normal or abnormal function of the kidney is directly related to the health and quality of life of an individual. Among them, kidney fibrosis affects the metal ion transport, glucose metabolism, immune function, inflammation, and cell mitosis of the organ, and is the main cause of organ disease.

[0052] Previous evaluation of the kidney mainly relies on laboratory tests (such as serum creatinine and urea nitrogen levels) and imaging examinations such as ultrasound, computed tomography (CT), and magnetic resonance imaging (MRI), but there are limitations such as single modality and insufficient sensitivity. Therefore, there is an urgent need for a comprehensive kidney status analysis / evaluation device that combines morphological, perfusion, and functional information to comprehensively evaluate the kidney status and improve the accuracy of kidney status evaluation.

[0053] Figure 1is a structural block diagram of a kidney state evaluation device provided by an embodiment of the present application. The kidney state evaluation device can include, but is not limited to, a device integrated on any terminal device such as a smart phone, a tablet computer, a PDA (Personal Digital Assistant), a smart television, a laptop computer, a desktop computer, a smart wearable device, etc. The device includes a transceiver module for receiving to-be-processed data (neck blood flow data as described below) and a processor for processing the to-be-processed data. The processor of the device can be carried in the terminal device. The processor of the device can be integrated in the same device as the transceiver module, or can be integrated in different devices, respectively, and the embodiments of the present application are not limited. Optionally, the device further includes a display module for displaying the processing result (for example, an evaluation result) of the device, for example, a screen in the terminal device.

[0054] As shown in Figure 1 The device includes an acquisition module 101, a registration module 102, a segmentation module 103, a processing module 104, and an evaluation module 105.

[0055] The acquisition module 101 is configured to acquire a kidney structure map, a renal blood flow perfusion map, and a T1 relaxation time quantitative map of a user.

[0056] The kidney structure map, the renal blood flow (RBF) map, and the T1 relaxation time quantitative map of the user can be acquired based on data collected by a computed tomography device and / or a magnetic resonance imaging device.

[0057] Optionally, the kidney structure map can be an image collected by a computed tomography device for the kidney of the user or an image collected by a magnetic resonance imaging device for the kidney of the user.

[0058] In actual applications, the clarity of the kidney anatomical structure embodied in the kidney structure map is higher than that of the kidney anatomical structure embodied in the renal blood flow perfusion map and the T1 relaxation time quantitative map.

[0059] The kidney structure image is an image used to show the structure of the kidney, and is mainly used to reflect anatomical information such as the shape, size, and position of the kidney tissue. In the process of obtaining the kidney structure image, a high-resolution image can be obtained by using a sequence such as VIBE (Volumetric Interpolated Breath-hold Examination), HASTE (Half-Fourier Acquisition Single-shot Turbo spin Echo), respiratory navigation T2-weighted imaging, and the like. The sequence has the characteristics of fast imaging and multi-tissue contrast, and can clearly show the anatomical structure of the kidney without breath holding or with short breath holding, thereby reducing the motion artifacts or chemical shift artifacts caused by respiration.

[0060] For example, the structure image is obtained by performing magnetic resonance scanning on the kidney by applying T1-weighted imaging to obtain morphological structure information of the kidney. That is, the structure image includes the morphological structure information of the kidney.

[0061] Optionally, the kidney blood perfusion image is generated based on an Arterial Spin Labeling (ASL) sequence collected by a magnetic resonance imaging device. The ASL technology can be used to achieve non-invasive evaluation. The ASL sequence can be one or more of a single-delay ASL sequence, a Multidelay ASL (MDASL) sequence, a Pseudo-Continuous ASL (PCASL) sequence, and a PCASL sequence under a Post Label Delay (PLD).

[0062] In an embodiment, the PCASL sequence under the PLD can be processed to obtain a kidney blood perfusion image that can quantify the blood perfusion value of the kidney. For example, the following formula (1) can be used to determine the kidney blood perfusion image under each delay time:

[0063]

[0064] where i is the i-th voxel in the kidney blood perfusion image, RBF i is the renal blood flow of the i-th voxel, R 1a = 0.61 is the longitudinal relaxation rate of blood, M0 is the equilibrium magnetization of the tissue, a = 0.8 is the labeling efficiency, τ = 1.5 s is the labeling duration, and w i= 0.5 / 1 / 1.5 / 2 / 2.5s, is the set-up post-label delay time, λ = 0.9 g / ml is the blood / tissue water partition coefficient, δ is the Arterial Transit Time (ATT) value in each voxel, and ΔM(i) refers to the signal change of the ith voxel.

[0065] Optionally, the T1 relaxation time map (T1Map) is obtained based on a T1Mapping sequence acquired by the magnetic resonance imaging device. The signal intensity value of each voxel in the T1 relaxation time map represents the T1 relaxation time of the voxel.

[0066] The inversion recovery sequence can be used for acquisition. According to the different signal intensities of the tissue in different inversion times (TI), the signal intensities of 4-5 TI values can be acquired in one repetition time (TR), and the T1 value of the tissue can be calculated according to the following formula (2):

[0067] S(TI) = M0[1 - 2exp(-TI / T1) + exp(-TR / T1) (2)

[0068] Wherein, M0 is the equilibrium magnetization of the tissue.

[0069] The registration module 102 is configured to perform image registration on the kidney structure map, the kidney perfusion map, and the T1 relaxation time map to obtain a target kidney structure map, a target kidney perfusion map, and a target T1 relaxation time map.

[0070] Image registration refers to the process of spatially aligning two or more images so that they have consistent anatomical structures or features in the same spatial coordinate system.

[0071] Image registration is achieved by finding a spatial transformation relationship to align the reference image and the image to be registered in the spatial position, so as to ensure that the same anatomical points or feature points in the image have the same position in the two images. This alignment is achieved by calculating the transformation parameters between the two images, which usually includes translation, rotation, scaling, etc.

[0072] The image registration technology is used to spatially align the kidney structure map, the kidney perfusion map, and the T1 relaxation time map to obtain a target kidney structure map, a target kidney perfusion map, and a target T1 relaxation time map. The same anatomical feature points of the kidney have the same position or coordinate in the target kidney structure map, the target kidney perfusion map, and the target T1 relaxation time map. The anatomical feature points of the kidney can be multiple, for example: blood vessel branch points, renal pelvis, etc.

[0073] In an alternative embodiment, the renal blood perfusion map and the T1 relaxation time quantitative map can be converted to the coordinate system of the renal structure map, to obtain the target renal blood perfusion map and the target T1 relaxation time quantitative map. In this embodiment, the target renal structure map is the renal structure map. That is, the image registration between the renal structure map, the renal blood perfusion map and the T1 relaxation time quantitative map can be achieved by only coordinate conversion of the renal blood perfusion map and the T1 relaxation time quantitative map.

[0074] Alternatively, the renal blood perfusion map and the T1 relaxation time quantitative map can be converted to the coordinate system of the renal structure map based on a feature point matching non-rigid registration algorithm, to obtain the target renal blood perfusion map and the target T1 relaxation time quantitative map. The feature point is an anatomical feature point. In the transformation process, non-rigid transformation is involved, that is, the distance between two points in one image will be transformed after being transformed into another image.

[0075] In an embodiment, the image registration process described above can include the following steps:

[0076] S11, determining a plurality of first anatomical feature points in a target map.

[0077] The target map is one of the renal blood perfusion map and the T1 relaxation time quantitative map.

[0078] The plurality of first anatomical feature points can include, but are not limited to, vessel branch points and renal pelvis. These points have obvious anatomical features, so as to be accurately identified in two images.

[0079] For example, the corresponding anatomical feature points are identified and marked in the target map.

[0080] S12, determining a plurality of second anatomical feature points in the renal structure map.

[0081] The plurality of second anatomical feature points can include, but are not limited to, vessel branch points and renal pelvis.

[0082] For example, the corresponding anatomical feature points are identified and marked in the target map.

[0083] S13, using the point-by-point mutual information method to pair the plurality of first anatomical feature points and the plurality of second anatomical feature points, to obtain a plurality of feature point pairs.

[0084] Pointwise Mutual Information (PMI) is a method used in probability theory and information theory to measure the correlation between two events or variables. The basic principle is that if two events x and y are independent, then the probability of both occurring P(x,y) is equal to the product of their individual probabilities P(x) and P(y). If x and y are not independent, i.e., there is a correlation between them, then P(x,y) will be greater than P(x)P(y).

[0085] In an alternative embodiment, the correlation between any first anatomical landmark in the plurality of first anatomical landmarks and any second anatomical landmark in the plurality of second anatomical landmarks is calculated using Pointwise Mutual Information, and according to the correlation, a plurality of landmark pairs is determined. One landmark in each landmark pair is from the plurality of first anatomical landmarks, and the other is from the second anatomical landmarks.

[0086] The implementation can be achieved by the following steps:

[0087] A. Local region selection:

[0088] A local region is defined for each landmark, which should be small enough to capture local features but large enough to contain sufficient information. The local region can be circular or square, and its size may need to be adjusted according to the resolution of the image and the local variation of the landmark.

[0089] B. Calculate local histogram:

[0090] For each landmark's local region, a histogram of its intensity values is calculated. This histogram will be used for subsequent mutual information calculation.

[0091] C. Estimate joint probability distribution:

[0092] For each pair of corresponding landmarks in the two images, the joint probability distribution of their local region intensity values is estimated by calculating the intersection of the two histograms.

[0093] D. Calculate mutual information:

[0094] The mutual information between each pair of landmarks is calculated using the following formula (3):

[0095] MI(x,y) = ∑ i,j P(x i ,y i ) log(P(x i ,y i ) / P(x i )P(y i ))(3)

[0096] where P(x i ,y i ) is the joint probability of two feature points on specific voxels i and j, P(x i ) and P(y i ) are the marginal probabilities.

[0097] E. Constructing the change matrix:

[0098] The calculated mutual information is composed into a mutual information change matrix.

[0099] F. Finding the best matching feature points:

[0100] In the mutual information change matrix, a mutual information threshold is set, and when the threshold is exceeded, the corresponding feature points are considered to be the most likely matching feature points.

[0101] S14. Determine the transformation matrix according to the plurality of feature point pairs.

[0102] According to the respective coordinate information of the two feature points in each of the plurality of feature point pairs, determine the transformation matrix.

[0103] According to the found plurality of feature point pairs, the transformation model between the images can be calculated through rigid transformation (translation, rotation, etc.) and affine transformation (scaling, shearing, etc.), an error function is constructed, and the transformation matrix is obtained by minimizing the error function.

[0104] S15. Convert the target image to the coordinate system of the kidney structure image according to the transformation matrix.

[0105] According to the transformation matrix, modify the target image to obtain an image registered with the kidney structure image.

[0106] Apply the calculated transformation matrix to all pixel points of the target image to convert the target image to the coordinate system of the kidney structure image. After applying the transformation, the registration quality is evaluated by calculating the overlap rate, mean square error, and other indicators of the corresponding anatomical structures of the two registered images. If the registration effect is not ideal, the matching point selection strategy, mutual information calculation parameters, and other methods are optimized.

[0107] For example: modify the kidney perfusion image according to the first transformation matrix to obtain a target kidney perfusion image registered with the kidney structure image.

[0108] For another example: modify the T1 relaxation time quantitative image according to the second transformation matrix to obtain a target T1 relaxation time quantitative image registered with the kidney structure image.

[0109] It should be noted that the first transformation matrix and the second transformation matrix can be determined in the manner provided in the above embodiments, and the first transformation matrix and the second transformation matrix can be different.

[0110] The segmentation module 103 is configured to segment the target kidney structure image by using the trained kidney segmentation model to obtain a first segmentation mask of kidney parenchyma, a second segmentation mask of kidney cortex, and a third segmentation mask of kidney medulla.

[0111] The kidney segmentation model can be a deep learning model. In an optional implementation, the kidney segmentation model can be a deep learning model with an Encoding-Decoding structure.

[0112] For example, the model can include a feature sharing and encoding module, a feature recognition module, and a decoding and up-sampling module. The input of the model is an image block obtained by cropping from a kidney structure image, and six types of segmentation results of left kidney parenchyma, right kidney parenchyma, left kidney cortex, right kidney cortex, left kidney medulla, and right kidney medulla are obtained by using three sub-segmentation networks.

[0113] The first segmentation mask of kidney parenchyma includes a first segmentation mask of left kidney parenchyma and a first segmentation mask of right kidney parenchyma; the second segmentation mask of kidney cortex includes a second segmentation mask of left kidney cortex and a second segmentation mask of right kidney cortex; and the third segmentation mask of kidney medulla includes a third segmentation mask of left kidney medulla and a third segmentation mask of right kidney medulla.

[0114] A training method of the above kidney segmentation model is introduced below, and the method includes the following steps:

[0115] S21, obtaining a training sample.

[0116] The training sample includes a sample kidney structure image and an expected segmentation result.

[0117] The expected segmentation result can be obtained by manual annotation, and the embodiments of the present application do not make specific limitations thereto.

[0118] S22, inputting the sample kidney structure image into the kidney segmentation model to obtain an actual segmentation result of the kidney segmentation model.

[0119] Optionally, the sample kidney structure image can be cropped to obtain a plurality of image blocks, and the plurality of image blocks can be input into the kidney segmentation model to obtain the actual segmentation result of the kidney segmentation model.

[0120] S23, according to the expected segmentation result and the actual segmentation result, using a cross-entropy loss function and a Dice loss function to optimize parameters of the kidney segmentation model.

[0121] To train the model, the embodiments of the present application adopt a composite loss function, including Cross-Entropy Loss and Dice Loss. Cross-Entropy Loss function is used to measure the difference between the probability distribution of model output and the real label, which is suitable for multi-class segmentation task, while Dice Loss function focuses on calculating the overlap between the predicted region and the real region, which is particularly suitable for dealing with class imbalance problem. The combination of such loss function can optimize the classification accuracy and segmentation precision of the model at the same time.

[0122] wherein the function definition of Cross-Entropy Loss is as follows:

[0123]

[0124] wherein C is the number of classes, y ic is an index variable, if sample i belongs to class c, then y ic = 1, otherwise y ic = 0, p ic is the probability of the model predicting that sample i belongs to class c. Any pixel point in the structure diagram is a sample.

[0125] The function definition of Dice Loss is as follows:

[0126]

[0127] wherein N is the number of samples, y i is the real label of the i-th sample, p i is the label of the i-th sample predicted by the model, and ε is a very small constant to prevent the denominator from being 0.

[0128] In addition, during the model training process, data augmentation strategies can be implemented, including rotation, scaling and flipping operations, to enhance the generalization ability of the model. In addition, small batch gradient descent method can also be used for training, combined with learning rate decay strategy and early stopping method, to prevent overfitting and accelerate the convergence of the model.

[0129] The processing module 104 is configured to determine the morphology of the renal parenchyma from the target renal structure diagram based on the first segmentation mask, the second segmentation mask and the third segmentation mask, determine the blood perfusion values of the renal cortex and the renal medulla respectively from the target renal blood perfusion diagram, and determine the T1 relaxation time of the renal cortex and the renal medulla respectively from the target T1 relaxation time quantitative diagram.

[0130] wherein the morphology of the renal parenchyma can include volume and surface area.

[0131] determine a morphology of the left renal cortex from the target T1 relaxation time quantitative map based on the third segmentation mask of the left renal medulla.

[0132] determine a morphology of the right renal cortex from the target T1 relaxation time quantitative map based on the third segmentation mask of the right renal medulla.

[0133] determine a blood perfusion value of the left renal cortex from the target renal blood perfusion map based on the second segmentation mask of the left renal cortex.

[0134] determine a T1 relaxation time of the left renal cortex from the target T1 relaxation time quantitative map based on the second segmentation mask of the left renal cortex.

[0135] determine a blood perfusion value of the right renal cortex from the target renal blood perfusion map based on the second segmentation mask of the right renal cortex.

[0136] determine a T1 relaxation time of the right renal cortex from the target T1 relaxation time quantitative map based on the second segmentation mask of the right renal cortex.

[0137] determine a blood perfusion value of the left renal medulla from the target renal blood perfusion map based on the third segmentation mask of the left renal medulla.

[0138] determine a T1 relaxation time of the left renal medulla from the target T1 relaxation time quantitative map based on the third segmentation mask of the left renal medulla.

[0139] determine a blood perfusion value of the right renal medulla from the target renal blood perfusion map based on the third segmentation mask of the right renal medulla.

[0140] determine a T1 relaxation time of the right renal medulla from the target T1 relaxation time quantitative map based on the third segmentation mask of the right renal medulla.

[0141] The evaluation module 105 is configured to evaluate the renal status of the user according to the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively.

[0142] In an optional implementation, the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively can be compared with prior information of renal disease states to evaluate the renal status.

[0143] The renal status can include a renal perfusion status.

[0144] In summary, the technical scheme provided by the embodiments of the present application utilizes the advantages of multi-modal magnetic resonance imaging to automatically segment the renal cortex and medulla, and automatically obtain the evaluation indexes of the shapes (volume, surface area), blood perfusion, and functional quantification of different renal regions after segmentation, establish the correlation between the evaluation indexes and disease characteristics, and thus realize non-invasive assessment of the state of the kidney.

[0145] Figure 2 A flowchart of a kidney data processing method provided by the embodiments of the present application is shown. The execution subject of the method can include a terminal device and / or a server, which is not specifically limited by the embodiments of the present application. The terminal device can also be referred to as a terminal. It can be a mobile terminal, a fixed terminal, or a portable terminal, such as a mobile phone, a station, a unit, a device, a multimedia computer, a multimedia tablet, an Internet node, a communicator, a desktop computer, a laptop computer, a notebook computer, a netbook computer, a tablet computer, a personal communication system device, a personal navigation device, a personal digital assistant, an audio / video player, a digital camera / camcorder, a positioning device, a television receiver, a radio broadcast receiver, an electronic book device, a game device, or any combination thereof, including accessories, peripherals, or any combination thereof. The server can be one or more servers. The server can also be an entity server or a virtual server. The server can be a standalone physical server, a server cluster composed of multiple physical servers, or a distributed system. It can also be a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, and basic cloud computing services such as big data and artificial intelligence platforms.

[0146] As shown in Figure 2 , the method can include:

[0147] 201, obtaining a kidney structure image, a kidney blood perfusion image, and a T1 relaxation time quantitative image of a user.

[0148] As shown in Figure 3 , the PCASL sequence is processed to obtain an RBF image, and the T1Mapping sequence is processed to obtain a T1Map.

[0149] 202, image registration is performed on the kidney structure image, the kidney blood perfusion image, and the T1 relaxation time quantitative image to obtain a target kidney structure image, a target kidney blood perfusion image, and a target T1 relaxation time quantitative image.

[0150] As shown in Figure 3 , the RBF image, the T1Map, and the structure image are registered to obtain a target kidney structure image, a target kidney blood perfusion image, and a target T1 relaxation time quantitative image.

[0151] 203. The target kidney structure map is segmented using the trained kidney segmentation model to obtain the first segmentation mask of the renal parenchyma, the second segmentation mask of the renal cortex, and the third segmentation mask of the renal medulla.

[0152] like Figure 3 As shown, a segmentation model is constructed, and the target kidney structure diagram is segmented by the constructed segmentation model to obtain the first segmentation mask of the renal parenchyma, the second segmentation mask of the renal cortex, and the third segmentation mask of the renal medulla.

[0153] 204. Based on the first segmentation mask, the second segmentation mask, and the third segmentation mask, determine the morphology of the renal parenchyma from the renal structure diagram, determine the blood perfusion values ​​of the renal cortex and the renal medulla from the target renal blood perfusion diagram, and determine the T1 relaxation time of the renal cortex and the renal medulla from the target T1 relaxation time quantification diagram.

[0154] like Figure 3 As shown, the morphology of the renal parenchyma includes its surface area and volume. By obtaining the morphology of the renal parenchyma, the blood perfusion values ​​of the renal cortex and the renal medulla, and the T1 relaxation times of the renal cortex and the renal medulla, the condition of the kidneys can be assessed.

[0155] The specific implementation of steps 201 to 204 above can be found in the corresponding contents of the above embodiments, and will not be repeated here.

[0156] Alternatively, the above method may also include:

[0157] 205. The user's kidney status is assessed based on the morphology of the renal parenchyma, the blood perfusion values ​​of the renal cortex and the renal medulla, and the T1 relaxation time of the renal cortex and the renal medulla.

[0158] It should be noted that any steps in the method provided in this application that are not described in detail can be found in the corresponding content of the above embodiments, and will not be repeated here. Furthermore, the method provided in this application may include other parts or all of the steps in the above embodiments in addition to the steps described above; for details, please refer to the corresponding content of the above embodiments, and will not be repeated here.

[0159] Figure 4 A schematic diagram of the structure of an electronic device according to an embodiment of this application is shown. Figure 4As shown, the electronic device includes a memory 1101 and a processor 1102. The memory 1101 can be configured to store various data to support operations on the electronic device. Examples of these data include instructions for any application or method operating on the electronic device. The memory 1101 can be implemented by any type of volatile or nonvolatile memory or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read only memory (EEPROM), erasable programmable read only memory (EPROM), programmable read only memory (PROM), read only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0160] The memory 1101 is configured to store programs.

[0161] The processor 1102 is coupled to the memory 1101 and is configured to execute the programs stored in the memory 1101 to implement the kidney data processing method provided by any of the above method embodiments.

[0162] Further, as shown, Figure 4 The electronic device further includes a communication component 1103, a display 1104, a power supply component 1105, an audio component 1106, and other components. Figure 4 Only some components are shown in the figure, and it does not mean that the electronic device only includes Figure 4 the components shown in the figure.

[0163] Correspondingly, the embodiments of the present application also provide a computer readable storage medium storing a computer program, which can implement the steps or functions of the kidney data processing method provided by any of the above method embodiments when the computer program is executed by a computer.

[0164] The embodiments of the present application also provide a computer program product, which includes a computer program, and the computer program can implement the steps or functions of the method provided by any of the above method embodiments when the computer program is executed by a processor.

[0165] The apparatus embodiments described above are merely illustrative, wherein the units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement it without creative labor.

[0166] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM (Read Only Memory), a RAM (Random Access Memory), a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0167] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A device for assessing a state of a kidney, characterized by, The method comprises the following steps: An acquisition module is configured to acquire a renal structure map, a renal blood perfusion map, and a T1 relaxation time quantitative map of a user based on data collected by a computed tomography device and / or a magnetic resonance imaging device; A registration module is configured to perform image registration on the renal structure map, the renal blood perfusion map, and the T1 relaxation time quantitative map to obtain a target renal structure map, a target renal blood perfusion map, and a target T1 relaxation time quantitative map; A segmentation module is configured to segment the target renal structure map by using a trained renal segmentation model to obtain a first segmentation mask of renal parenchyma, a second segmentation mask of renal cortex, and a third segmentation mask of renal medulla; A processing module is configured to determine a morphology of the renal parenchyma from the target renal structure map based on the first segmentation mask, determine a blood perfusion value of the renal cortex from the target renal blood perfusion map based on the second segmentation mask, determine a T1 relaxation time of the renal cortex from the target T1 relaxation time quantitative map based on the second segmentation mask, determine a blood perfusion value of the renal medulla from the target renal blood perfusion map based on the third segmentation mask, and determine a T1 relaxation time of the renal medulla from the target T1 relaxation time quantitative map based on the third segmentation mask; An evaluation module is configured to compare the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla, and the T1 relaxation times of the renal cortex and the renal medulla with prior information of a renal disease state to evaluate the renal state of the user.

2. The device of claim 1, wherein The registration module is configured to convert the renal blood perfusion map and the T1 relaxation time quantitative map to a coordinate system of the renal structure map based on a feature point matching non-rigid registration algorithm to obtain the target renal blood perfusion map and the target T1 relaxation time quantitative map, and the target renal structure map is the renal structure map.

3. The apparatus of claim 2, wherein, The registration module is configured to: determine a plurality of first anatomical feature points in a target map, the target map being one of the renal blood perfusion map and the T1 relaxation time quantitative map; determine a plurality of second anatomical feature points in the renal structure map; pair the plurality of first anatomical feature points and the plurality of second anatomical feature points by using a point-by-point mutual information method to obtain a plurality of feature point pairs; determine a transformation matrix based on the plurality of feature point pairs; convert the target map to the coordinate system of the renal structure map according to the transformation matrix.

4. The device of claim 1, wherein The acquisition module is configured to acquire a renal structure map of a user, an ASL image sequence of a kidney of the user, and a T1 Mapping sequence of the kidney of the user, determine a renal blood perfusion map of the user based on the ASL image sequence, and determine a T1 relaxation time quantitative map of the user based on the T1 Mapping sequence.

5. The apparatus of claim 1, wherein, The training process of the renal segmentation model comprises: acquiring a training sample, the training sample comprising a sample renal structure map and an expected segmentation result; inputting the sample kidney structure image into the kidney segmentation model to obtain an actual segmentation result of the kidney segmentation model; performing parameter optimization on the kidney segmentation model according to the expected segmentation result and the actual segmentation result by using a cross-entropy loss function and a Dice loss function.

6. A kidney data processing method characterized by, The method comprises the following steps: acquiring a kidney structure image, a kidney perfusion image and a T1 relaxation time quantitative image of a user based on data acquired by a computed tomography device and / or a magnetic resonance imaging device; performing image registration on the kidney structure image, the kidney perfusion image and the T1 relaxation time quantitative image to obtain a target kidney structure image, a target kidney perfusion image and a target T1 relaxation time quantitative image; segmenting the target kidney structure image by using the trained kidney segmentation model to obtain a first segmentation mask of kidney parenchyma, a second segmentation mask of kidney cortex and a third segmentation mask of kidney medulla; determining a morphology of the kidney parenchyma from the target kidney structure image based on the first segmentation mask, determining a blood perfusion value of the kidney cortex from the target kidney perfusion image based on the second segmentation mask, determining a T1 relaxation time of the kidney cortex from the target T1 relaxation time quantitative image based on the second segmentation mask, determining a blood perfusion value of the kidney medulla from the target kidney perfusion image based on the third segmentation mask, and determining a T1 relaxation time of the kidney medulla from the target T1 relaxation time quantitative image based on the third segmentation mask; comparing and analyzing the morphology of the kidney parenchyma, the blood perfusion values of the kidney cortex and the kidney medulla respectively and the T1 relaxation times of the kidney cortex and the kidney medulla respectively with prior information of a kidney disease state to evaluate the kidney state of the user.

7. An electronic device, comprising: The method comprises the following steps: a memory and a processor, wherein the memory is configured to store a program; the processor is coupled to the memory and is configured to execute the program stored in the memory to implement the following steps: acquiring a kidney structure image, a kidney perfusion image and a T1 relaxation time quantitative image of a user based on data acquired by a computed tomography device and / or a magnetic resonance imaging device; performing image registration on the kidney structure image, the kidney perfusion image and the T1 relaxation time quantitative image to obtain a target kidney structure image, a target kidney perfusion image and a target T1 relaxation time quantitative image; segmenting the target kidney structure image by using the trained kidney segmentation model to obtain a first segmentation mask of kidney parenchyma, a second segmentation mask of kidney cortex and a third segmentation mask of kidney medulla; determining a morphology of the kidney parenchyma from the target kidney structure image based on the first segmentation mask, determining a blood perfusion value of the kidney cortex from the target kidney perfusion image based on the second segmentation mask, determining a T1 relaxation time of the kidney cortex from the target T1 relaxation time quantitative image based on the second segmentation mask, determining a blood perfusion value of the kidney medulla from the target kidney perfusion image based on the third segmentation mask, and determining a T1 relaxation time of the kidney medulla from the target T1 relaxation time quantitative image based on the third segmentation mask; determine the morphology of the renal parenchyma from the target kidney structure image based on the first segmentation mask, determine the blood perfusion value of the renal cortex from the target kidney blood perfusion image based on the second segmentation mask, determine the T1 relaxation time of the renal cortex from the target T1 relaxation time quantitative image based on the second segmentation mask, determine the blood perfusion value of the renal medulla from the target kidney blood perfusion image based on the third segmentation mask, and determine the T1 relaxation time of the renal medulla from the target T1 relaxation time quantitative image based on the third segmentation mask; compare and analyze the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively with prior information of kidney disease states, so as to evaluate the kidney state of the user.

8. A computer readable storage medium storing a computer program, characterized in that, The computer program, when executed by a computer, can implement: based on the data collected by the computed tomography device and / or the magnetic resonance imaging device, obtaining the kidney structure image, the kidney blood perfusion image, and the T1 relaxation time quantitative image of the user; image registration is performed on the kidney structure image, the kidney blood perfusion image, and the T1 relaxation time quantitative image, to obtain a target kidney structure image, a target kidney blood perfusion image, and a target T1 relaxation time quantitative image; segmenting the target kidney structure image by using the trained kidney segmentation model to obtain a first segmentation mask of the renal parenchyma, a second segmentation mask of the renal cortex, and a third segmentation mask of the renal medulla; determine the morphology of the renal parenchyma from the target kidney structure image based on the first segmentation mask, determine the blood perfusion value of the renal cortex from the target kidney blood perfusion image based on the second segmentation mask, determine the T1 relaxation time of the renal cortex from the target T1 relaxation time quantitative image based on the second segmentation mask, determine the blood perfusion value of the renal medulla from the target kidney blood perfusion image based on the third segmentation mask, and determine the T1 relaxation time of the renal medulla from the target T1 relaxation time quantitative image based on the third segmentation mask; compare and analyze the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively with prior information of kidney disease states, so as to evaluate the kidney state of the user.

9. A computer program product comprising a computer program, characterized in that, The computer program, when executed by a computer, can implement: based on the data collected by the computed tomography device and / or the magnetic resonance imaging device, obtaining the kidney structure image, the kidney blood perfusion image, and the T1 relaxation time quantitative image of the user; image registration is performed on the kidney structure image, the kidney blood perfusion image, and the T1 relaxation time quantitative image, to obtain a target kidney structure image, a target kidney blood perfusion image, and a target T1 relaxation time quantitative image; segmenting the target kidney structure image by using the trained kidney segmentation model to obtain a first segmentation mask of the renal parenchyma, a second segmentation mask of the renal cortex, and a third segmentation mask of the renal medulla; determine the morphology of the renal parenchyma from the target kidney structure image based on the first segmentation mask, determine the blood perfusion value of the renal cortex from the target kidney blood perfusion image based on the second segmentation mask, determine the T1 relaxation time of the renal cortex from the target T1 relaxation time quantitative image based on the second segmentation mask, determine the blood perfusion value of the renal medulla from the target kidney blood perfusion image based on the third segmentation mask, and determine the T1 relaxation time of the renal medulla from the target T1 relaxation time quantitative image based on the third segmentation mask; compare and analyze the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively with prior information of kidney disease states, so as to evaluate the kidney state of the user. determine the morphology of the renal parenchyma from the target kidney structure image based on the first segmentation mask, determine the blood perfusion value of the renal cortex from the target kidney blood perfusion image based on the second segmentation mask, determine the T1 relaxation time of the renal cortex from the target T1 relaxation time quantitative image based on the second segmentation mask, determine the blood perfusion value of the renal medulla from the target kidney blood perfusion image based on the third segmentation mask, and determine the T1 relaxation time of the renal medulla from the target T1 relaxation time quantitative image based on the third segmentation mask; compare and analyze the morphology of the renal parenchyma, the blood perfusion values of the renal cortex and the renal medulla respectively, and the T1 relaxation times of the renal cortex and the renal medulla respectively with the kidney disease state prior information, and realize the evaluation of the kidney state of the user.

Images