Physiological age prediction method, device, equipment and medium

Abstract

This application provides a physiological age prediction method, apparatus, device, and medium. By acquiring fundus images, the method predicts age groups based on ocular features in the images, and then determines an age offset based on lesion features in the images. Based on the age group prediction results and the age offset, the user's physiological age can be obtained. Furthermore, this application can adjust the resolution of fundus images, using low-resolution images for age group prediction and high-resolution images for lesion detection. This allows physiological age to be predicted using fundus images, saving prediction costs and shortening prediction time.

Description

Technical Field

[0001] The present application relates to the field of digital medicine, and in particular to a method, apparatus, device and medium for predicting physiological age. Background Art

[0002] Everyone ages at different rates. Compared to actual age, biological age more accurately reflects the degree of aging. Currently, there are many methods for assessing biological age, such as measuring telomere length, assessing the DNA methylation clock, or using magnetic resonance imaging (MRI) for prediction. However, these methods are all expensive and require a long testing cycle.

[0003] Therefore, predicting physiological age is difficult, costly, and time-consuming, and is an urgent problem that needs to be solved. Summary of the Invention

[0004] The present application provides a physiological age prediction method, apparatus, device and medium. By acquiring a fundus image, the eye features in the fundus image are used to predict the age group, and then the age offset is determined based on the lesion features in the fundus image; the user's physiological age can be obtained based on the age group prediction result and the age offset.

[0005] The object and other objects are achieved by the features of the independent claims. Further implementations are revealed in the dependent claims, the description and the drawings.

[0006] In a first aspect, the present application provides a physiological age prediction method, characterized in that a fundus image of a user is obtained; based on the fundus image, a first result and a second result are determined, the first result is determined based on the probability that the fundus image belongs to each age group, and the second result is determined based on the number and type of lesions in the fundus image; based on the first result and the second result, the physiological age of the user is determined.

[0007] In the second aspect, the present application provides a physiological age prediction device, characterized in that it includes: an acquisition unit and a determination unit, the acquisition unit is used to acquire the fundus image of the user; the determination unit is used to determine a first result and a second result based on the fundus image, the first result is determined based on the probability that the fundus image belongs to each age group, and the second result is determined based on the number and type of lesions in the fundus image; the determination unit is used to determine the user's physiological age based on the first result and the second result.

[0008] In a third aspect, the present application provides a computer device, characterized in that it includes: a processor and a memory, the memory storing a computer program, and the processor executing the computer program in the memory to implement the method described in the first aspect.

[0009] In a fourth aspect, the present application provides a computer-readable storage medium, which stores a computer program, and is characterized in that when the above-mentioned computer program is run on a computer, the above-mentioned computer executes the method described in the first aspect.

[0010] In summary, the physiological age prediction method provided in the embodiments of the present application obtains fundus images, predicts age groups based on the ocular features in the fundus images, and then determines an age offset based on the lesion features in the fundus images. Based on the age group prediction results and the age offset, the user's physiological age can be obtained. Furthermore, the present application can adjust the resolution of the fundus images, using low-resolution images for age group prediction and high-resolution images for lesion detection. In this way, physiological age can be predicted using fundus images, saving prediction costs and shortening prediction time. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art.

[0012] Figure 1 A schematic diagram of the architecture of an AI system provided in an embodiment of the present application;

[0013] Figure 2 A schematic diagram of a flow chart of a physiological age prediction method provided in an embodiment of the present application;

[0014] Figure 3 A schematic diagram of a flow chart of a physiological age prediction method in an application scenario provided in an embodiment of the present application;

[0015] Figure 4 A schematic diagram of the structure of a physiological age prediction device provided in an embodiment of the present application;

[0016] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] The terms used in the following examples of the present application are for the purpose of describing specific embodiments only and are not intended to limit the present application. As used in the specification and appended claims of the present application, the singular expressions "a," "an," "said," "above," "the," and "this" are intended to include plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in the present application refers to and encompasses any or all possible combinations of one or more of the listed items.

[0018] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to achieve optimal results.

[0019] Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0020] An AI model is a set of mathematical methods used to implement AI. A large amount of sample data can be used to train the target model, resulting in a target model with predictive capabilities. The target model can then be fed the data to be predicted to generate predictions.

[0021] The following is an explanation of the structure of the AI ​​system. Figure 1 As shown, Figure 1 This is an architecture diagram of an AI system. The system 100 is a commonly used system architecture in the AI ​​field. The system 100 includes: a database 110, a training device 120 and an execution device 130.

[0022] The database 110 is used to store sample sets, wherein the samples in the sample set may be graphics, images, voices, texts, etc. The database 110 is also used to send the samples to the training device 120 for model training. In medical application scenarios, the sample data may be medical images, and the type of objects contained in the sample data is a lesion, i.e., a part of the body where a lesion occurs. Medical images refer to images of internal tissues obtained in a non-invasive manner for medical treatment or medical research, such as images of the fundus, stomach, abdomen, heart, knee, and brain, such as computed tomography (CT), magnetic resonance imaging (MRI), ultrasound (US), X-ray images, electroencephalograms, and optical photographic images generated by medical instruments.

[0023] The training device 120 is used to train the model to be trained using samples. Specifically, it includes: using a batch of samples to perform an iterative training on the model to be trained, comparing the output results of the model to be trained with the labels of the samples, and adjusting the model parameters of the model to be trained based on the comparison results, and then performing the next iterative training until the model training meets the preset termination conditions, then terminating the training and obtaining the target model. The preset termination conditions can be that the training reaches the number of training iterations, or the value of the loss function (loss function) or the objective function (objective function) is less than a preset value, wherein the loss function and the objective function are used to measure the difference between the model output results and the sample data labels, that is, the difference between the predicted value and the target value.

[0024] The execution device 130 is used to implement various functions based on the target model trained by the training device 120. Specifically, the execution device 130 obtains the data to be predicted from the user, and then inputs the user data into the target model to obtain the prediction result.

[0025] In summary, the implementation of various applications in the AI ​​field depends on AI models, which implement different functions such as classification, recognition, detection, etc., and AI models need to be pre-trained using sample sets before they can be deployed to the execution device 130 for use.

[0026] Currently, biological age can be predicted by measuring telomere length, assessing the DNA methylation clock, or using MRI. Compared to actual age, biological age more accurately reflects the degree of aging. However, these methods require high testing costs and long testing cycles, making them unsuitable for large-scale use.

[0027] In order to solve the problem that predicting physiological age is difficult, costly, and time-consuming, the present solution provides a physiological age prediction method 200 , which can be applied to the AI ​​system 100 .

[0028] S210 : The execution device 130 obtains a fundus image of the user.

[0029] The user's fundus image acquired by execution device 130 can be a standard color fundus photograph or an ultra-wide-angle color fundus photograph. It should be understood that ultra-wide-angle photography has a wider field of view, and an ultra-wide-angle fundus image can observe more fundus information than a standard color fundus photograph. Therefore, if the fundus image is not an ultra-wide-angle color fundus photograph, it is necessary to obtain multiple standard color fundus photographs and stitch them together to obtain the fundus image.

[0030] In some embodiments, the fundus image is a clear and complete image of the user's eye, which can be obtained through a cleaning and screening operation. For example, fundus images with blur, closed eyes, and malposition are screened out. For fundus images that do not meet the requirements, it is necessary to re-acquire a clear and complete fundus image.

[0031] In some embodiments, the fundus image is a standardized image. Specifically, the process includes: selecting a fundus image from a database as a standard image, wherein the standard image is a fundus image with clear imaging. The fundus image is standardized according to the standard image. The specific process is referred to the following formula (1):

[0032]

[0033] Wherein, sta is the pixel value matrix of the fundus image after standardization, x is the pixel value matrix of the original fundus image, μ is the mean of the pixel values ​​of the standard image, and σ is the standard deviation of the pixel values ​​of the standard image.

[0034] In some embodiments, fundus image normalization can be performed by separating the pixel values ​​of the three RGB channels of the fundus image. Alternatively, since the fundus image is acquired using red and green lasers, there is little valid data in the blue channel. Therefore, fundus image normalization can also be performed by separating the pixel values ​​of the R and G channels of the fundus image.

[0035] In some embodiments, the fundus image is processed with Contrast Limited Adaptive Histogram Equalization (CLAHE). CLAHE can change the image contrast by calculating the local histogram of the image and then redistributing the brightness, thereby obtaining more image details in subsequent feature extraction.

[0036] In some embodiments, the fundus image is a normalized image. Specifically, the normalized fundus image is obtained by representing each pixel value in the fundus image with a value in the interval [0, 1] according to the maximum and minimum values ​​of the pixel values ​​in the fundus image. The specific process can be referred to the following formula (2):

[0037]

[0038] Wherein, nor is the pixel value of the fundus image after normalization, y is the pixel value in the fundus image, min is the minimum value among the pixel values ​​in the fundus image, and max is the maximum value among the pixel values ​​in the fundus image. It should be understood that the normalization processing of the fundus image can be performed separately on the pixel values ​​of the three RGB channels of the fundus image.

[0039] In some embodiments, the fundus image is subjected to standardization processing, CLAHE processing, and normalization processing, or a combination of any two of the above processing. The fundus image obtained by the training device 120 from the database 110 may also be subjected to the above processing.

[0040] In some embodiments, the execution device 130 may also obtain user information, such as the user's gender, actual age, height, and user's living habits, including whether the user smokes, whether the user often stays up late, and whether the user often drinks.

[0041] In some embodiments, the user information is normalized data. For example, for discrete data such as whether to smoke, drink, or stay up late, the results are represented by values ​​of 1 and 0 to represent "yes" and "no".

[0042] It should be understood that to ensure the accuracy of the prediction results, in some embodiments, it is also necessary to ensure that the acquisition time of the user's fundus image and user information does not exceed a preset time period. For example, the acquisition time of the user's fundus image and user information does not exceed one year.

[0043] S220 : The execution device 130 performs age group prediction based on the fundus image to obtain a first result.

[0044] The execution device 130 first uses the fundus image to predict the age range, obtains the score of the fundus image belonging to each age interval, and then multiplies the score of each age interval with the representative value of the age interval and adds them together to obtain a preliminary predicted age value, that is, the first result.

[0045] Specifically, the execution device 130 first extracts the fundus features of the fundus image, such as the optic disc size, the optic ring size, the diameter of the retinal artery and vein, etc., and then compresses it in the pooling layer to obtain a feature vector, and then passes it through the fully connected layer and softmax to obtain the score of the fundus image corresponding to each age range, and then multiplies the score of each age range with the representative value of the age range and adds them together to obtain the first result.

[0046] The score for each age range can be the similarity between the fundus features and those in each age range. That is, the database stores fundus images for each age range, and the similarity between the user's fundus image and the fundus images in each age range is calculated. The score for each age range can also be determined based on the conditional probability of the fundus feature appearing in the fundus image. That is, the conditional probability of the fundus feature appearing in each age group is stored in the database. For example, if the diameter of the user's retinal artery and vein is k, the database stores the probability of the retinal artery and vein having the same diameter for each age group. This probability then becomes the score for the fundus image in each age range.

[0047] The representative value of an age interval can be any value in the age interval. For example, the minimum value of the age interval is used as the representative value of the age interval. The first result can be expressed by the following formula (3):

[0048]

[0049] Among them, age is the first result, Min(range i ) represents the minimum value of each age range, score i Represents the scores of each age range, N is the number of intervals, i and N are positive integers.

[0050] For example, the age range can be divided into 11 intervals with a size of 10 years, from 1 to 111 years old, namely [1,11), [11,21), ..., [101,111).

[0051] It should be understood that the prediction model for age group prediction can use a classification model, such as a residual network (resnet50), Inceptionv3, etc., and this application does not impose any specific restrictions on the type of prediction model. The training device 120 can use a mean squared error (MSE) loss function during the training process. For details, please refer to the following formula (4):

[0052]

[0053] Among them, MSE L is the MSE loss function value, M is the number of samples used in one iterative training; is the label value; z i is the predicted value.

[0054] In some embodiments, before performing age group prediction using the fundus image, the execution device 130 further includes adjusting the resolution of the fundus image to a low resolution to obtain a first image, and performing age group prediction based on the first image to obtain a first result. For example, the resolution of the fundus image can be adjusted to 256*256 to obtain the first image.

[0055] It should be understood that the first image may also be subjected to standardization processing, CLAHE processing, normalization processing, or any combination of the above processing. When the first image is standardized, the resolution of the standard image also needs to be adjusted to the same low resolution as the first image to obtain a first standard image, and the first image is standardized using the first standard image.

[0056] S230 : The execution device 130 performs lesion detection according to the fundus image to obtain a second result.

[0057] The second result represents the age offset based on the fundus lesions in the fundus image. This is used to correct the first result for a more accurate age prediction. Fundus lesions can reflect the user's physical condition. The more severe the fundus lesion, the older the user's physiological age corresponding to the fundus image.

[0058] The calculation process of the second result specifically includes: the execution device 130 obtains lesion features from the database, and each lesion corresponds to a different weight; the lesions in the fundus image are detected using the lesion detection model to obtain the similarity between the fundus image and the features of each lesion; then the similarity of each lesion is multiplied by the corresponding lesion weight, and then the multiplication results of all lesions are added to obtain the second result.

[0059] For example, the database includes lesion types such as microaneurysms, hard exudates, pellucid membrane oil, hemorrhage, and geographic atrophy, with the weights of these lesions increasing in order. This means that geographic atrophy has a higher weight. If the fundus image detects a higher probability of geographic atrophy in the user, i.e., a higher confidence score, then the second result is higher, indicating that the user's physiological age is higher.

[0060] It should be understood that the lesion detection model can be YOLO (You Only Look Once) or a regional convolutional neural network (Regions with CNN features, RCNN). This application does not specifically limit the type of lesion detection model. Before using the lesion detection model for lesion detection, the lesion detection model needs to be trained. The loss function in the training process can be a Smooth L1 Loss function. For details, refer to the following formula (5):

[0061]

[0062] Among them, a is the difference between the predicted value and the label value.

[0063] In some embodiments, before performing lesion detection using a fundus image, the execution device 130 further includes adjusting the resolution of the fundus image to a high resolution to obtain a second image, and performing lesion prediction based on the second image to obtain a second result. The resolution of the second image is greater than the resolution of the fundus image used for age group prediction. When the fundus image used for age group prediction is the first image, the resolution of the second image is greater than the resolution of the first image. For example, when the resolution of the first image is 256*256, the resolution of the fundus image can be adjusted to 512*512 to obtain the second image.

[0064] It should be understood that the second image may also be subjected to standardization, CLAHE, normalization, or any combination of the above processes. When the second image is standardized, the resolution of the standard image needs to be adjusted to the same high resolution as the second image to obtain a second standard image, which is then used to perform standardization on the second image.

[0065] In some embodiments, the execution device 130 may first perform step S230 to detect lesions based on the fundus image to obtain the second result, and then perform step S220 to predict age based on the fundus image to obtain the first result. Alternatively, steps S220 and 230 may be performed simultaneously.

[0066] S240. Obtain physiological age based on the first result and the second result.

[0067] Execution device 130 adds the first and second results together to determine the user's physiological age. The first result is the predicted age based on the prediction model, while the second result is the age offset based on the lesion detection model. In other words, the final physiological age is the sum of the fundus features and the lesion condition.

[0068] In some embodiments, the physiological age may be obtained by weighting the first result and the second result and then adding them together.

[0069] In summary, the physiological age prediction method 200 provided in this application obtains fundus images, predicts age groups based on ocular features in the fundus images, and then determines an age offset based on lesion features in the fundus images. Based on the age group prediction results and the age offset, the user's physiological age can be obtained. Furthermore, this application can adjust the resolution of the fundus images, using low-resolution images for age group prediction and high-resolution images for lesion detection. In this way, physiological age can be predicted using fundus images, saving prediction costs and shortening prediction time.

[0070] The following is an example of the physiological age prediction method 200 of the present application. Figure 3 As shown, Figure 3 The following example shows the process of obtaining the fundus image of user A and predicting the physiological age of user A based on the fundus image of user A. Figure 3 The steps of this method are introduced in detail.

[0071] Step 1: Obtain the user's fundus image.

[0072] After capturing a fundus image 300 of user A using fundus imaging technology, the execution device acquires the fundus image 300. It should be understood that the execution device also screens the fundus image to ensure that the fundus image 300 is clear and complete. If the fundus image is blurred, the eye is closed, or the image is misaligned, a clear and complete fundus image needs to be acquired again.

[0073] After acquiring the fundus image 300 , the execution device 130 adjusts the resolution of the fundus image 300 to obtain a first image and a second image, wherein the resolution of the first image is smaller than the resolution of the second image.

[0074] The execution device 130 further performs normalization processing on the first image and the second image. When normalizing the first image, the resolution of the standard image needs to be adjusted to the same low resolution as the first image to obtain a first standard image, which is then used to perform normalization processing on the first image. When normalizing the second image, the resolution of the standard image needs to be adjusted to the same high resolution as the second image to obtain a second standard image, which is then used to perform normalization processing on the second image.

[0075] The execution device 130 further performs one or more of standardization processing, CLAHE processing, and normalization processing on the first image and the second image. The specific process can be referred to the description of the aforementioned step S210, which will not be repeated here.

[0076] Step 2: Perform age group prediction based on the first image to obtain a first result.

[0077] The execution device 130 first extracts fundus features of the first image. In this example, the execution device extracts the diameter of the fundus artery, where the diameter of the artery of user A is d, and d is a positive number.

[0078] Then, the execution device 130 obtains the probability of the artery diameter d belonging to each age group from the database, wherein the probability of the artery diameter d belonging to the age group of 1-10 years old is P1, the probability of the artery diameter d belonging to the age group of 11-20 years old is P2, ..., the probability of the artery diameter d belonging to the age group of 91-100 years old is P 10 .

[0079] The execution device 130 takes the maximum value of each age group as the representative value of the age group, multiplies the probability of each age group by the representative value of each age group, and then adds them together to obtain the first result X. The specific process can be referred to the description of the above step S220 and will not be repeated here.

[0080] Step 3: Perform lesion detection based on the second image to obtain a second result.

[0081] The execution device 130 obtains existing lesion samples from the database. The lesion samples include images of the lesions, lesion types, and lesion weights. Figure 3 , an image of lesion A and its corresponding weight a, and an image of lesion B and its corresponding weight b are obtained.

[0082] The execution device 130 calculates the similarity between the second image and the lesion samples one by one, and finally obtains the similarity between the fundus image of user A and the lesion A as K a , and the similarity with lesion B is K b .

[0083] Finally, the execution device 130 multiplies the similarity of each lesion by the weight corresponding to the lesion, and then adds the results of all lesions to obtain the second result Y. The specific process can be referred to the description of the aforementioned step S230 and will not be repeated here.

[0084] It should be understood that the execution device 130 may first perform step 3 to detect lesions based on the second image to obtain the second result, and then perform step 2 to predict age based on the first image to obtain the first result. Alternatively, steps 2 and 3 may be performed simultaneously.

[0085] Step 4: Obtain physiological age based on the first result and the second result.

[0086] Execution device 130 adds the first and second results, resulting in the physiological age of user A. It should be understood that physiological age includes both the age range predicted based on fundus features and the age offset based on fundus lesion characteristics. For details, please refer to the description of step S240 above and will not be repeated here.

[0087] To summarize, the physiological age prediction method provided in this application obtains fundus images, predicts age groups based on the eye features in the fundus images, and then determines the age offset based on the lesion features in the fundus images; based on the age group prediction results and the age offset, the user's physiological age can be obtained.

[0088] In order to solve the problem that predicting physiological age is difficult, costly and time-consuming, the present application provides a physiological age prediction device, such as Figure 4 As shown, the physiological age prediction device 400 includes: an acquisition unit 410 and a determination unit 420.

[0089] The acquisition unit 410 is used to acquire a fundus image of the user.

[0090] The determination unit 420 is used to determine a first result and a second result based on the fundus image, the first result being determined based on the probability that the fundus image belongs to each age group, and the second result being determined based on the number and type of lesions in the fundus image; the determination unit 420 is also used to determine the user's physiological age based on the first result and the second result.

[0091] In some embodiments, the determination unit 420 is also used to determine the probability that the fundus image belongs to each age group based on the similarity between the fundus features in the fundus image and the fundus features of each age group, where the fundus features include one or more of the optic disc size, arteriovenous diameter, and blood color.

[0092] In some embodiments, the acquisition unit 410 is further configured to acquire the probability of the fundus image belonging to each age group from the database according to the fundus features, wherein the database stores the probability of each fundus feature appearing in each age group.

[0093] In some embodiments, the determination unit 420 is also used to obtain a first result based on the probability that the fundus image belongs to each age group multiplied by the representative value corresponding to each age group, where the representative value corresponding to the age group is a numerical value within the age group range, such as the maximum value, minimum value, average value or median within the age range.

[0094] In some embodiments, the determination unit 420 is also used to adjust the resolution of the fundus image to obtain a first image and a second image, wherein the resolution of the first image is smaller than the resolution of the second image; determine a first result based on the first image; and determine a second result based on the second image.

[0095] In some embodiments, the acquisition unit 410 is further configured to acquire multiple lesion samples from the database, where the lesion samples include lesion images, lesion types, and lesion weights. The determination unit 420 is further configured to determine, based on the second image and the lesion image of each lesion sample, a similarity between the second image and each lesion sample; and determine a second result based on the similarity between the second image and each lesion sample and the lesion weight corresponding to each lesion sample.

[0096] In some embodiments, the determining unit 420 is further configured to add the first result and the second result to obtain the physiological age, or the determining unit 420 is further configured to weight the first result and the second result respectively and then add them to obtain the physiological age.

[0097] In some embodiments, the determination unit 420 is further used to perform normalization processing on the first image to obtain a normalized first image, wherein each pixel value in the normalized first image is represented by a numerical value between 0 and 1; and to determine a first result based on the normalized first image; the determination unit 420 is further used to perform normalization processing on the second image to obtain a normalized second image; and to determine a second result based on the normalized second image.

[0098] It should be understood that the first image and the second image may also have undergone standardization processing, CLAHE processing, normalization processing, or any combination of the above processing. Specifically, when standardizing the first image, it is necessary to adjust the resolution of the standard image to the same low resolution as the first image to obtain a first standard image, and then use the first standard image to perform standardization on the first image. When standardizing the second image, it is necessary to adjust the resolution of the standard image to the same high resolution as the second image to obtain a second standard image, and then use the second standard image to perform standardization on the second image.

[0099] In summary, the physiological age prediction device 400 provided in this application acquires fundus images, predicts age groups based on the ocular features in the fundus images, and then determines an age offset based on the lesion features in the fundus images. Based on the age group prediction results and the age offset, the user's physiological age can be obtained. Furthermore, this application can adjust the resolution of the fundus images, using low-resolution images for age group prediction and high-resolution images for lesion detection. This allows physiological age to be predicted using fundus images, saving prediction costs and shortening prediction time.

[0100] See also Figure 5 , Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device 500 may be the physiological age prediction device 400 mentioned above. Figure 5 As shown, the electronic device 500 includes: a processor 510 , a communication interface 520 and a memory 530 . The processor 510 , the communication interface 520 and the memory 530 are interconnected via an internal bus 540 .

[0101] The processor 510, the communication interface 520, and the memory 530 can be connected via a bus, or can communicate via other means such as wireless transmission. The embodiment of the present application takes the connection via bus 540 as an example, wherein the bus 540 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus 540 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 5 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0102] The processor 510 may be composed of one or more general-purpose processors, such as a central processing unit (CPU), or a combination of a CPU and a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof. The processor 510 executes various types of digitally stored instructions, such as software or firmware programs stored in the memory 530, which enables the electronic device 500 to provide a wide variety of services.

[0103] Specifically, the processor 510 may be composed of at least one general-purpose processor, such as a central processing unit (CPU), or a combination of a CPU and a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The PLD may be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL), or any combination thereof. The processor 510 executes various types of digitally stored instructions, such as software or firmware programs stored in the memory 530, which enables the electronic device 500 to provide a wide variety of services.

[0104] The memory 530 may include a volatile memory (Volatile Memory), such as a random access memory (Random Access Memory, RAM); the memory 530 may also include a non-volatile memory (Non-Volatile Memory), such as a read-only memory (Read-Only Memory, ROM), a flash memory (Flash Memory), a hard disk drive (Hard Disk Drive, HDD) or a solid-state drive (SSD); the memory 530 may also include a combination of the above types. Among them, the memory 530 may store application code and program data. The program code can obtain fundus images, predict the age of the eye features in the fundus images, and then determine the age offset based on the lesion features in the fundus images; based on the age prediction results and the age offset, the user's physiological age can be obtained, etc. It can also be used to execute Figure 2 The other steps described in the embodiment will not be described here in detail. The code of the memory 530 may include the code for implementing the functions of the acquisition unit and the determination unit. The functions of the acquisition unit include Figure 4 The functions of the acquisition unit 410 in the embodiment, such as acquiring the fundus image of the user, can be used to execute step S210 and its optional steps of the aforementioned method, which will not be described in detail here. The functions of the generation unit include Figure 4The functions of the determination unit 420 in the method include, for example, determining a first result and a second result based on the fundus image, where the first result is determined based on the probability that the fundus image belongs to each age group, and the second result is determined based on the number of lesions in the fundus image and the type of each lesion; and determining the user's physiological age based on the first and second results. Specifically, the method may be used to execute steps S220 to S240 of the aforementioned method and their optional steps, which will not be described in detail here.

[0105] The communication interface 520 can be a wired interface (such as an Ethernet interface), an internal interface (such as a high-speed serial computer expansion bus (Peripheral Component Interconnect express, PCIe) bus interface), a wired interface (such as an Ethernet interface) or a wireless interface (such as a cellular network interface or a wireless local area network interface) for communicating with other devices or modules.

[0106] Need to explain, Figure 5 This is only one possible implementation of the embodiment of the present application. In actual applications, the electronic device may also include more or fewer components, which is not limited here. Figure 2 The relevant explanations in the embodiments are not repeated here. Figure 5 The electronic device shown may also be a computer cluster composed of multiple computing nodes, which is not specifically limited in this application.

[0107] The embodiment of the present application further provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a processor, Figure 2 The method flow shown is realized.

[0108] The present application also provides a computer program product. When the computer program product is run on a processor, Figure 4 The method flow shown is realized.

[0109] The above embodiments may be implemented in whole or in part through software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product comprises one or more computer instructions. When the computer program instructions are loaded or executed on a computer, the processes or functions described in accordance with the embodiments of the present invention are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes a collection of one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a high-density digital video disc (DVD), or a semiconductor medium. The semiconductor medium may be an SSD.

[0110] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A method for predicting physiological age, characterized in that: include: Acquire a fundus image of the user, wherein the fundus image includes an ultra-wide-angle color fundus photograph; Determine a first result and a second result based on the fundus image, wherein the first result is determined based on the probability that the fundus image belongs to each age group, and the second result is determined based on the number and type of lesions in the fundus image. Determining the first result and the second result based on the fundus image includes: adjusting the resolution of the fundus image to obtain a first image and a second image, wherein the resolution of the first image is smaller than the resolution of the second image; determining the first result based on the first image; obtaining a plurality of lesion samples from a database, wherein the lesion samples include lesion images, lesion types, and lesion weights, and determining a similarity between the second image and each lesion sample based on the lesion image of each lesion sample; and determining the second result based on the similarity between the second image and each lesion sample and the lesion weight corresponding to each lesion sample, wherein the second result is used to represent an age offset obtained based on the fundus lesion condition of the fundus image, and is used to correct the first result; The physiological age of the user is determined according to the first result and the second result.

2. The method according to claim 1, characterized in that The probability that the fundus image belongs to each age group is determined based on the similarity between the fundus features in the fundus image and the fundus features of each age group, where the fundus features include one or more of optic disc size, arteriovenous diameter, and blood color.

3. The method according to claim 1, characterized in that The probability that the fundus image belongs to each age group is obtained from a database based on fundus features, wherein the database stores the probability of each fundus feature appearing in each age group.

4. The method according to claim 3, characterized in that The physiological age is determined based on a weighted sum of the first result and the second result.

5. The method according to claim 4, characterized in that Determining the first result according to the first image includes: performing normalization processing on the first image to obtain a normalized first image, wherein each pixel value in the normalized first image is represented by a numerical value between 0 and 1; Determining the first result according to the normalized first image; Determining the second result according to the second image includes: performing normalization processing on the second image to obtain a normalized second image; The second result is determined according to the normalized second image.

6. A physiological age prediction device, the device being used to execute the method according to any one of claims 1 to 5, characterized in that: include: Get unit, determine unit, The acquisition unit is used to acquire a fundus image of the user; The determining unit is configured to determine a first result and a second result based on the fundus image, wherein the first result is determined based on the probability that the fundus image belongs to each age group, and the second result is determined based on the number and type of lesions in the fundus image; The determining unit is configured to determine the physiological age of the user according to the first result and the second result.

7. A computer device, characterized in that: include: A processor and a memory, wherein the memory stores a computer program, and the processor executes the computer program in the memory to implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is run on a computer, the computer is caused to perform the method according to any one of claims 1 to 5.

Images