Brain age prediction system, apparatus, and storage medium

Abstract

This invention discloses a brain age prediction system, device, and storage medium. The system includes a processor configured to execute a brain age prediction method comprising: extracting brain region volume features of at least two target brain regions from a subject's structural magnetic resonance imaging (MRI) image to obtain a target brain region volume feature set; and inputting the target brain region volume feature set into a trained width learning system model to obtain a brain age prediction result for the subject. This invention solves the problem of low accuracy in brain age prediction results in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of medical image processing technology, and in particular to a brain age prediction system, device, and storage medium. Background Technology

[0002] Brain structural changes associated with brain development and aging can easily trigger neurodegenerative diseases such as depression, schizophrenia, and Alzheimer's disease. Constructing effective biomarkers to assess the degree of brain structural abnormalities has become a hot research topic in neuroscience.

[0003] Existing technologies typically use machine learning methods to analyze the volume characteristics of all brain regions to determine the predicted brain age of a subject. However, this approach fails to fully utilize the linear and nonlinear relationships between the volume characteristics of each brain region and brain age, resulting in low accuracy in brain age prediction. Summary of the Invention

[0004] This invention provides a brain age prediction system to address the problem of low accuracy in existing brain age prediction methods.

[0005] According to one aspect of the present invention, a brain age prediction system is provided, comprising a processor configured to perform a brain age prediction method, the method comprising:

[0006] Brain region volume features of at least two target brain regions are extracted from the structural magnetic resonance images of the subject to obtain a set of target brain region volume features;

[0007] The target brain region volume feature set is input into a trained width learning system model to obtain the subject's brain age prediction results.

[0008] According to another aspect of the present invention, a brain age prediction device is provided, comprising:

[0009] The feature extraction module is used to extract brain region volume features of at least two target brain regions from the subject's structural magnetic resonance images to obtain a set of target brain region volume features;

[0010] The prediction module is used to input the set of volumetric features of the target brain region into a trained width learning system model to obtain the brain age prediction result of the subject.

[0011] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the following brain age prediction method, including:

[0012] Brain region volume features of at least two target brain regions are extracted from the structural magnetic resonance images of the subject to obtain a set of target brain region volume features;

[0013] The target brain region volume feature set is input into a trained width learning system model to obtain the subject's brain age prediction results.

[0014] The technical solution of the brain age prediction system provided in this invention uses a trained width learning system model to quickly extract linear and nonlinear features related to brain age from the target brain region volume feature set. Compared with a single linear or nonlinear feature, brain age prediction based on both linear and nonlinear features can obtain more accurate brain age prediction results.

[0015] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a structural block diagram of the brain age prediction system provided in an embodiment of the present invention;

[0018] Figure 2A This is a flowchart of the brain age prediction method provided in an embodiment of the present invention;

[0019] Figure 2B This is a schematic diagram of a width learning system model provided in an embodiment of the present invention;

[0020] Figure 3A This is a flowchart of yet another brain age prediction method provided according to an embodiment of the present invention;

[0021] Figure 3B This is a schematic diagram illustrating the importance of redundant brain regions according to an embodiment of the present invention;

[0022] Figure 4 This is a flowchart of yet another brain age prediction method provided according to an embodiment of the present invention;

[0023] Figure 5 This is a flowchart of yet another brain age prediction method provided according to an embodiment of the present invention;

[0024] Figure 6 This is a flowchart of a width learning system model training method provided in an embodiment of the present invention;

[0025] Figure 7 This is a structural block diagram of a brain age prediction device provided according to an embodiment of the present invention. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification, claims and accompanying drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such processes, methods, products or devices.

[0028] As disclosed in this article, the term "brain age" refers to the age of the brain. During human growth, development, and aging, brain age may not match a person's chronological age. Clinically, it is sometimes necessary to estimate the discrepancy between brain age and chronological age.

[0029] As disclosed in this article, the term "structural magnetic resonance imaging" is a type of magnetic resonance imaging specifically used to visualize static tissue structures within the brain.

[0030] Figure 1 The diagram illustrates the structure of a brain age prediction system 10 provided in this embodiment of the invention. The system includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the processor 11. The memory stores computer programs executable by the processor 11. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0031] Multiple components in the brain age prediction system 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, optical disk, etc.; and a communication unit 19, such as a network card, modem, wireless transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0032] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described below, such as brain age prediction methods.

[0033] Figure 2A This is a flowchart of a brain age prediction method provided in an embodiment of the present invention. This embodiment is applicable to situations where the brain age prediction result of a subject is determined based on structural magnetic resonance images. This method can be executed by a brain age prediction device, which can be implemented in hardware and / or software and can be configured in the processor of a brain age prediction system. Figure 2A As shown, the method includes:

[0034] S110. Extract brain region volume features of at least two target brain regions from the subject's structural magnetic resonance images to obtain a set of target brain region volume features.

[0035] Among them, the number of brain regions corresponding to the brain region volume feature set of at least two target brain regions is less than the total number of brain regions of the subject.

[0036] The location and extent of each brain region in the structural magnetic resonance imaging (SMRI) image are determined using internationally accepted brain region segmentation methods, and the volumetric features of each brain region are extracted. In one embodiment, 101 brain regions included in the subject's SMRI image are identified, and the volumetric features of each brain region are extracted. This embodiment does not impose specific limitations on the method of brain region feature extraction, as long as the volumetric features of each target brain region can be extracted.

[0037] S120. Input the set of target brain region volume features into the trained width learning system model to obtain the subject's brain age prediction results.

[0038] In one embodiment, such as Figure 2BAs shown, the width learning system model includes a feature layer, an enhancement layer, and an output layer. The feature layer includes at least two feature nodes, each used to extract linear features of the brain volume characteristics of the at least two target brain regions. The enhancement layer includes at least two enhancement feature nodes, each used to perform a nonlinear mapping on the corresponding linear features to obtain nonlinear features. The output layer is used to perform a weighted sum of the linear features extracted by each feature node and the nonlinear features extracted by each enhancement feature node to obtain the brain age prediction result. Specifically, after obtaining the nonlinear and linear features, the brain age prediction result is determined by the following formula. ,in, Here, A represents the model weights, A represents the linear and nonlinear feature sets, and Y represents the brain age prediction result. These are the regularization parameters for optimizing the model. In this embodiment, linear features refer to features that are linearly related to brain age, while nonlinear features refer to features that are nonlinearly related to brain age. The input layer consists of other feature extraction algorithms or feature extraction models used in conjunction with these algorithms.

[0039] The technical solution of the brain age prediction method provided in this invention uses a trained width learning system model to quickly extract linear and nonlinear features related to brain age from the target brain region volume feature set. Compared with a single linear or nonlinear feature, brain age prediction based on both linear and nonlinear features can obtain more accurate brain age prediction results.

[0040] Figure 3A A flowchart illustrating another brain age prediction method provided in this embodiment of the invention, which belongs to the same inventive concept as the aforementioned embodiments. The method includes:

[0041] S2101. Extract brain region volume features of all brain regions from the subject's structural magnetic resonance images.

[0042] In one embodiment, brain volume features of all brain regions are extracted from the subject's structural magnetic resonance images, i.e., brain volume features of 101 brain regions are extracted.

[0043] S2102. Delete at least one brain region volume feature that is not related to brain age from all brain region volume features to obtain the target brain region volume feature set.

[0044] Among all brain regions in the human brain, some regions show a linear relationship with brain age changes, some show a non-linear relationship, and some may be unrelated or have a low correlation. In this embodiment, the volume of brain regions that are unrelated to or have a low correlation with brain age changes are designated as redundant brain regions. Here, "unrelated" means that the correlation between brain region volume characteristics and brain age is lower than a set correlation threshold.

[0045] In one embodiment, such as Figure 3BAs shown, brain regions with an importance score below 0.2 are considered redundant brain regions. These redundant brain regions include the globus pallidus of the right hemisphere, the periphery of the calcarine fissure of the left hemisphere, the lateral occipital region of the right hemisphere, the inferior temporal gyrus of the right hemisphere, and the posterior slope of the superior temporal sulcus of the right hemisphere.

[0046] This embodiment removes all redundant brain region volume features from all brain region volume features to obtain brain region volume features of at least two target brain regions related to brain age, i.e., a set of target brain region volume features. It is understood that removing redundant brain region volume features helps improve the speed of brain age prediction by reducing data computation, and improves the accuracy of brain age prediction by reducing the impact of irrelevant or low-relevance features on the prediction process.

[0047] In one embodiment, all brain region volume features are input into a trained random forest model. The trained random forest model can evaluate the correlation between brain region volume features and brain age by fusing the results of multiple decision trees, and evaluate the importance of each brain region volume feature in the form of weighted ranking. Then, by deleting brain region volume features with weights below a set threshold and retaining brain region volume features with weights greater than or equal to the weights, redundant brain regions are deleted.

[0048] S220. Input the set of target brain region volume features into the trained width learning system model to obtain the subject's brain age prediction results.

[0049] In this embodiment of the invention, redundant brain region volume features are removed from all brain region volume features to obtain brain region volume features of at least two target brain regions related to brain age, i.e., a set of target brain region volume features. The brain age of the subject is predicted using the set of target brain region volume features, which achieves the goal of predicting brain age using only target brain region volume features related to brain age, and helps to improve the speed and accuracy of determining brain age prediction results.

[0050] Figure 4 A flowchart illustrating another brain age prediction method provided in this embodiment of the invention, which belongs to the same inventive concept as the aforementioned embodiments. The method includes:

[0051] S3101. Extract brain region volume features of all brain regions from the subject's structural magnetic resonance images.

[0052] S3202. Determine the type of structural magnetic resonance image.

[0053] In one embodiment, the type of structural magnetic resonance image refers to the magnetic field strength of the magnetic resonance device that generates the structural magnetic resonance image. For example, the type of structural magnetic resonance image generated by a magnetic resonance device with a magnetic field strength of 1.5T is 1.5T, and the type of structural magnetic resonance image generated by a magnetic resonance device with a magnetic field strength of 3.0T is 3.0T.

[0054] S3203. Based on the type of structural magnetic resonance image and the correspondence between the type of pre-created structural magnetic resonance image and the set of redundant brain regions, determine the set of redundant brain regions corresponding to the structural magnetic resonance image.

[0055] The redundant brain region set includes at least two redundant brain region identifiers, and the brain regions corresponding to these at least two redundant brain region identifiers are either unrelated to brain age or have a low correlation with it.

[0056] A configuration file is created beforehand, which records the correspondence between the types of structural magnetic resonance images (SMRs) and the sets of redundant brain regions. Once the type of SMR is determined, the set of redundant brain regions corresponding to that SMR can be determined based on the SMR type and the correspondence.

[0057] S3204. Remove the redundant brain region volume features from the set of all brain region volume features to obtain the target brain region volume feature set.

[0058] After the redundant brain region set is determined, the brain region volume features corresponding to each brain region identifier in the redundant brain region set are deleted from all brain region volume features, thereby obtaining brain region volume features of at least two target brain regions related to brain age, and the brain region volume features of the at least two target brain regions are used as the target brain region volume feature set.

[0059] S320. Input the set of target brain region volume features into the trained width learning system model to obtain the subject's brain age prediction results.

[0060] The technical solution of the brain age prediction method provided in this invention improves the speed of deleting redundant brain region volume features from all brain region volume features by pre-creating a correspondence between the types of magnetic resonance images of each structure and the set of redundant brain regions, thereby improving the speed of determining the set of target brain region volume features and the speed of determining the brain age prediction result.

[0061] Figure 5 A flowchart illustrating another brain age prediction method provided in this embodiment of the invention, which belongs to the same inventive concept as the aforementioned embodiments. The method includes:

[0062] S4101. Extract brain region volume features of all brain regions from the subject's structural magnetic resonance images.

[0063] S4102. Determine the type of structural magnetic resonance image.

[0064] S4103. Based on the type of structural magnetic resonance image and the correspondence between the type of pre-created structural magnetic resonance image and the target brain region set, determine the target brain region set corresponding to the structural magnetic resonance image.

[0065] The target brain region set includes at least two target brain region identifiers, and the brain regions corresponding to these at least two target brain region identifiers are all associated with brain age.

[0066] A configuration file is created beforehand, which records the correspondence between the types of structural magnetic resonance images (SMRs) and the sets of target brain regions. Once the type of SMR is determined, the set of target brain regions corresponding to that SMR can be determined based on the SMR type and the correspondence.

[0067] S4104. Extract the brain region volume features of each brain region from the target brain region volume feature set to obtain brain region volume features of at least two target brain regions related to brain age, and use the brain region volume features of the at least two target brain regions as the target brain region volume feature set.

[0068] Once the target brain region set is determined, the brain region volume features corresponding to each brain region in the target brain region set are extracted from all brain region volume features, thereby obtaining the brain region volume features of at least two target brain regions related to brain age.

[0069] S420. Input the set of target brain region volume features into the trained width learning system model to obtain the subject's brain age prediction results.

[0070] The technical solution of the brain age prediction method provided in this invention improves the speed of selecting brain region volume features of at least two target brain regions related to brain age from the target brain region volume feature set by pre-creating a correspondence between the type of magnetic resonance images of each structure and the target brain region set, thereby improving the speed of determining the target brain region volume feature set and the speed of determining the brain age prediction result.

[0071] Figure 6 A flowchart illustrating a method for training a width learning system model according to an embodiment of the present invention. The method includes:

[0072] S510. Obtain a training sample set, which includes at least two training samples. The training samples include a set of target brain region volume features and the corresponding subject's brain age label.

[0073] In one embodiment, the subjects corresponding to the training samples are all healthy individuals, at least those with healthy brains, meaning their brain development or aging rate is consistent with their physiological age. Therefore, the brain age label of the training samples is the age label of the corresponding subjects.

[0074] S520. Based on the leave-one-out method, determine the test samples and training samples included in the training sample set, and input the training samples into the width learning system model for model training.

[0075] Leave-one-out is a method for evaluating learners in machine learning, and it is a special case of cross-validation. Specifically, it involves dividing a large dataset into k smaller datasets, using k-1 of these smaller datasets as the training set, and the remaining smaller dataset as the test set.

[0076] S530. When it is detected that the width learning system model has completed one model training, determine the optimal regularization parameter corresponding to the test sample.

[0077] The optimal regularization parameters for the test samples are determined through the following steps:

[0078] Step a1: Obtain the set of regularization parameters. For any regularization parameter in the set, determine the sum of squared predicted residuals of the test samples corresponding to the model weights under the regularization parameter constraint, and the mean square error corresponding to the sum of squared predicted residuals.

[0079] Since the linear features of the feature layer, the nonlinear features of the enhancement layer, the model weights, and the brain age prediction results satisfy the formula: ,in, Here, A represents the model weights, A represents the linear and nonlinear feature sets, and Y represents the brain age prediction result. The regularization parameter is used to optimize the model. Therefore, given the linear and nonlinear characteristics, the brain age prediction results of the test samples corresponding to the model weights under the regularization parameter constraint are calculated, thereby calculating the sum of squared prediction residuals corresponding to the test samples.

[0080] In one embodiment, the set of regularization parameters is limited by boundary regularization parameters and step size, for example, Step size is .

[0081] Obtain the range of regularization parameters, calculate the sum of squared predicted residuals of the test samples corresponding to the current regularization parameter using the hat matrix, estimate the mean squared error of the test samples corresponding to the corresponding regularization parameter using this sum of squared predicted residuals, store the mean squared error, and determine whether the current regularization parameter is the last regularization parameter. If not, take the next regularization parameter as the current regularization parameter and return to the step of calculating the sum of squared predicted residuals of the test samples corresponding to the current regularization parameter using the hat matrix. If yes, determine that the mean squared errors corresponding to all regularization parameters have been stored.

[0082] Here, the hat matrix refers to a type of projection matrix used to predict the sum of squared residuals. For a linear model Y=Xβ+e, E(e)=0, cov(e)=σ2I, the matrix H... ...X(XTX)⁻¹XT is the projection matrix that projects the observation vector Y onto the subspace generated by the column vectors of X. Y^=HY, and H is conventionally called the hat matrix.

[0083] Step a2: Take the regularization parameter corresponding to the minimum mean square error as the optimal regularization parameter.

[0084] It is understandable that the better the regularization parameter, the smaller the mean square error of the corresponding model. Therefore, the regularization parameter corresponding to the minimum mean square error is taken as the optimal regularization parameter.

[0085] S540. Determine the model weights of the width learning system model based on the optimal regularization parameters to generate the trained width learning system model.

[0086] The minimum mean square error is determined, and then the model weights are calculated using a fast pseudo-inverse algorithm, thereby completing the training of the width learning system model.

[0087] The training method for the width learning system model provided in this embodiment of the invention introduces the leave-one-out method into the training of the width learning system model, so as to evaluate the mean square error of the model under different regularization parameters based on the squared predicted residuals and brain age labels, which helps to reduce computational complexity, improve model training speed and model stability.

[0088] Figure 7 This is a schematic diagram of the brain age prediction device provided in an embodiment of the present invention. This embodiment belongs to the same inventive concept as the aforementioned embodiments. Figure 7 As shown, the device includes:

[0089] The feature extraction module 710 is used to extract brain region volume features of at least two target brain regions from the structural magnetic resonance images of the subject to obtain a set of target brain region volume features;

[0090] The prediction module 720 is used to input the set of volume features of the target brain region into a trained width learning system model to obtain the brain age prediction result of the subject.

[0091] In one embodiment, the feature extraction module 710 includes:

[0092] The feature extraction unit is used to extract brain region volume features of all brain regions from the subject's structural magnetic resonance images;

[0093] The feature selection unit is used to remove at least one target brain region volume feature that is not related to brain age from all target brain region volume features to obtain a set of target brain region volume features.

[0094] In one embodiment, the feature selection unit is used to input all brain region volume features into a trained random forest model to obtain brain region volume features of at least two target brain regions related to brain age, and to use the brain region volume features of the at least two target brain regions as a set of target brain region volume features.

[0095] In one embodiment, the feature selection unit is used for:

[0096] Determine the type of the structural magnetic resonance image;

[0097] Based on the type of the structural magnetic resonance image and the correspondence between the type of the pre-created structural magnetic resonance image and the set of redundant brain regions, the set of redundant brain regions corresponding to the structural magnetic resonance image is determined.

[0098] The brain region volume features of each brain region in the redundant brain region set are deleted from all brain region volume features to obtain the target brain region volume feature set.

[0099] In one embodiment, a redundant brain region set determination module is further included, which is used to:

[0100] For any type of structural magnetic resonance image, the set of brain region features corresponding to the structural magnetic resonance image is input into a trained random forest model to obtain brain region volume features of at least two target brain regions related to brain age.

[0101] Remove the brain region volume features of at least two target brain regions from all brain region volume features to obtain at least one redundant brain region volume feature;

[0102] The brain regions corresponding to the volume features of the at least one redundant brain region are defined as the set of redundant brain regions.

[0103] In one embodiment, the structural magnetic resonance image type is the magnetic field strength of the magnetic resonance device that generates the structural magnetic resonance image.

[0104] In one embodiment, the width learning model is a width learning system model.

[0105] In one embodiment, a training module is further included, the training module comprising:

[0106] A sample unit is used to acquire a training sample set, the training sample set including at least two training samples, the training samples including the target brain region volume feature set and the corresponding subject's brain age label;

[0107] The training unit is used to determine the test samples and training samples included in the training sample set based on the leave-one-out method, and input the training samples into the width learning system model for model training.

[0108] The optimal regularization parameter determination unit is used to determine the optimal regularization parameter corresponding to the test sample when the wide learning system model has completed one model training.

[0109] The weight determination unit is used to determine the model weights of the width learning system model according to the optimal regularization parameters, so as to generate a trained width learning system model.

[0110] In one embodiment, the optimal regularization parameter determination unit is used for:

[0111] Obtain the set of regularization parameters. For any regularization parameter in the set, determine the sum of squared predicted residuals of the test samples corresponding to the model weights under the regularization parameter constraint, and the mean squared error corresponding to the sum of squared predicted residuals.

[0112] The regularization parameter corresponding to the minimum mean square error is taken as the optimal regularization parameter.

[0113] The technical solution of the brain age prediction device provided in this invention uses a trained width learning system model to quickly extract linear and nonlinear features related to brain age from the target brain region volume feature set. Compared with a single linear or nonlinear feature, brain age prediction based on both linear and nonlinear features can obtain more accurate brain age prediction results.

[0114] The brain age prediction device provided in the embodiments of the present invention can execute the brain age prediction method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0115] In some embodiments, the brain age prediction method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as... Figure 1 The storage unit 18 is used in the memory. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the brain age prediction method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the brain age prediction method by any other suitable means (e.g., by means of firmware).

[0116] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0117] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0118] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0119] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0120] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0121] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0122] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0123] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A brain age prediction system, characterized in that, Includes a processor configured to perform a brain age prediction method, the method comprising: Extracting brain region volume features of at least two target brain regions from the subject's structural magnetic resonance images to obtain a set of target brain region volume features includes: extracting brain region volume features of all brain regions from the subject's structural magnetic resonance images; deleting at least one brain region volume feature that is not related to brain age from all brain region volume features to obtain a set of target brain region volume features. The target brain region volume feature set is input into a trained width learning system model to obtain the subject's brain age prediction result. The width learning system model includes a feature layer, an enhancement layer, and an output layer. The feature layer includes at least two feature nodes, each used to extract linear features of the brain region volume features of the at least two target brain regions. The enhancement layer includes at least two enhancement feature nodes, each used to perform nonlinear mapping on the corresponding linear features to obtain nonlinear features. The output layer is used to perform a weighted sum of the linear features extracted by each feature node and the nonlinear features extracted by each enhancement feature node to obtain the brain age prediction result. The step of deleting at least one brain region volume feature that is not related to brain age from all brain region volume features to obtain a target brain region volume feature set includes: The type of the structural magnetic resonance image is determined, wherein the structural magnetic resonance image type is the magnetic field strength of the magnetic resonance device that generates the structural magnetic resonance image; Based on the type of the structural magnetic resonance image and the correspondence between the type of the pre-created structural magnetic resonance image and the set of redundant brain regions, the set of redundant brain regions corresponding to the structural magnetic resonance image is determined. The brain region volume features of each brain region in the redundant brain region set are deleted from all brain region volume features to obtain the target brain region volume feature set. The process of determining the set of redundant brain regions corresponding to the type of each structural magnetic resonance image includes the following steps: For any type of structural magnetic resonance image, the set of brain region features corresponding to the structural magnetic resonance image is input into a trained random forest model to obtain brain region volume features of at least two target brain regions related to brain age. Remove the brain region volume features of at least two target brain regions from all brain region volume features to obtain at least one redundant brain region volume feature; The brain regions corresponding to the volume features of the at least one redundant brain region are defined as the set of redundant brain regions.

2. The system according to claim 1, characterized in that, The training of the width learning system model is completed through the following steps: Obtain a training sample set, which includes at least two training samples, each of which includes the target brain region volume feature set and the corresponding subject's brain age label. The test samples and training samples included in the training sample set are determined based on the leave-one-out method, and the training samples are input into the width learning system model for model training. If the width learning system model has completed one training iteration, determine the optimal regularization parameter corresponding to the test sample. The model weights of the width learning system model are determined based on the optimal regularization parameters to generate a trained width learning system model.

3. The system according to claim 2, characterized in that, Determining the optimal regularization parameter corresponding to the test sample includes: Obtain a set of regularization parameters. For any regularization parameter in the set of regularization parameters, determine the sum of squared predicted residuals of the test samples corresponding to the model weights under the constraints of the regularization parameter, and the mean squared error corresponding to the sum of squared predicted residuals. The regularization parameter corresponding to the minimum mean square error is taken as the optimal regularization parameter.

4. A brain age prediction device, characterized in that, include: The feature extraction module is used to extract brain region volume features of at least two target brain regions from the subject's structural magnetic resonance images to obtain a set of target brain region volume features, including: extracting brain region volume features of all brain regions from the subject's structural magnetic resonance images; deleting at least one brain region volume feature that is not related to brain age from all brain region volume features to obtain a set of target brain region volume features; A prediction module is used to input the target brain region volume feature set into a trained width learning system model to obtain the subject's brain age prediction result. The width learning system model includes a feature layer, an enhancement layer, and an output layer. The feature layer includes at least two feature nodes, each used to extract linear features of the brain region volume features of the at least two target brain regions. The enhancement layer includes at least two enhancement feature nodes, each used to perform nonlinear mapping on the corresponding linear features to obtain nonlinear features. The output layer is used to perform a weighted summation of the linear features extracted by each feature node and the nonlinear features extracted by each enhancement feature node to obtain the brain age prediction result. The feature extraction module determines the target brain region volume feature set through a feature selection unit, wherein the feature selection unit is used for: The type of the structural magnetic resonance image is determined, wherein the structural magnetic resonance image type is the magnetic field strength of the magnetic resonance device that generates the structural magnetic resonance image; Based on the type of the structural magnetic resonance image and the correspondence between the type of the pre-created structural magnetic resonance image and the set of redundant brain regions, the set of redundant brain regions corresponding to the structural magnetic resonance image is determined. The brain region volume features of each brain region in the redundant brain region set are deleted from all brain region volume features to obtain the target brain region volume feature set. The redundant brain region set determination module determines the redundant brain region set corresponding to the type of each structural magnetic resonance image through the redundant brain region set determination module, which is used for: For any type of structural magnetic resonance image, the set of brain region features corresponding to the structural magnetic resonance image is input into a trained random forest model to obtain brain region volume features of at least two target brain regions related to brain age. Remove the brain region volume features of at least two target brain regions from all brain region volume features to obtain at least one redundant brain region volume feature; The brain regions corresponding to the volume features of the at least one redundant brain region are defined as the set of redundant brain regions.

5. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the following brain age prediction method, including: Extracting brain region volume features of at least two target brain regions from the subject's structural magnetic resonance images to obtain a set of target brain region volume features includes: extracting brain region volume features of all brain regions from the subject's structural magnetic resonance images; deleting at least one brain region volume feature that is not related to brain age from all brain region volume features to obtain a set of target brain region volume features. The target brain region volume feature set is input into a trained width learning system model to obtain the subject's brain age prediction result. The width learning system model includes a feature layer, an enhancement layer, and an output layer. The feature layer includes at least two feature nodes, each used to extract linear features of the brain region volume features of the at least two target brain regions. The enhancement layer includes at least two enhancement feature nodes, each used to perform nonlinear mapping on the corresponding linear features to obtain nonlinear features. The output layer is used to perform a weighted sum of the linear features extracted by each feature node and the nonlinear features extracted by each enhancement feature node to obtain the brain age prediction result. The step of deleting at least one brain region volume feature that is not related to brain age from all brain region volume features to obtain a target brain region volume feature set includes: The type of the structural magnetic resonance image is determined, wherein the structural magnetic resonance image type is the magnetic field strength of the magnetic resonance device that generates the structural magnetic resonance image; Based on the type of the structural magnetic resonance image and the correspondence between the type of the pre-created structural magnetic resonance image and the set of redundant brain regions, the set of redundant brain regions corresponding to the structural magnetic resonance image is determined. The brain region volume features of each brain region in the redundant brain region set are deleted from all brain region volume features to obtain the target brain region volume feature set. The process of determining the set of redundant brain regions corresponding to the type of each structural magnetic resonance image includes the following steps: For any type of structural magnetic resonance image, the set of brain region features corresponding to the structural magnetic resonance image is input into a trained random forest model to obtain brain region volume features of at least two target brain regions related to brain age. Remove the brain region volume features of at least two target brain regions from all brain region volume features to obtain at least one redundant brain region volume feature; The brain regions corresponding to the volume features of the at least one redundant brain region are defined as the set of redundant brain regions.

Images