Acetabular fracture typing prediction method and system based on deep learning

Abstract

The invention provides an acetabular fracture typing prediction method and system based on deep learning, and belongs to the technical field of deep learning, and the method comprises the steps: processing first data based on the basic information of a target person to obtain first target data, and processing second data to obtain second target data; wherein the first data is X-ray detection data of the acetabulum position of the target person, and the second data is CT three-dimensional reconstruction data of the acetabulum position of the target person; identifying the first target data and the second target data based on a deep learning model to obtain acetabular fracture typing features; and sending the acetabular fracture typing characteristics to a first device, wherein the first device is used for displaying the acetabular fracture typing characteristics. According to the acetabular fracture typing prediction method and system based on deep learning, the accuracy of acetabular fracture feature recognition can be improved, and auxiliary information is provided for doctors.

Description

Technical Field

[0001] The present disclosure belongs to the technical field of deep learning, and more specifically, relates to a method and system for predicting acetabular fracture classification based on deep learning. Background Art

[0002] Acetabular fracture is a serious hip joint injury, often caused by high-energy trauma such as traffic accidents, falls from heights, etc.

[0003] Traditional fracture classification methods, such as Letournel-Judet classification and AO classification, mainly rely on doctors to manually identify images such as X-rays and CTs. However, due to the complex anatomical structure of the acetabulum and many overlapping images, the traditional method of relying on doctors to manually interpret images for acetabular fracture classification is prone to inaccurate judgment of fracture lines, fracture positions, etc. in the images.

[0004] Therefore, a method that can assist doctors in judging the type of acetabular fracture is needed. Summary of the Invention

[0005] The purpose of the present disclosure is to provide a method and system for predicting acetabular fracture classification based on deep learning, so as to improve the accuracy of acetabular fracture feature recognition and provide auxiliary information for doctors.

[0006] In the first aspect of the embodiments of the present disclosure, a method for predicting acetabular fracture classification based on deep learning is provided, including: processing the first data based on the basic information of the target person to obtain first target data, and processing the second data to obtain second target data; wherein, the first data is the X-ray detection data of the acetabular position of the target person, and the second data is the CT three-dimensional reconstruction data of the acetabular position of the target person; Identifying the acetabular fracture classification features based on the deep learning model for the first target data and the second target data; Sending the acetabular fracture classification features to the first device, and the first device is used to display the acetabular fracture classification features.

[0007] In the second aspect of the embodiments of the present disclosure, a system for predicting acetabular fracture classification based on deep learning is provided, including: A preprocessing module, configured to process the first data based on the basic information of the target person to obtain first target data, and process the second data to obtain second target data; wherein, the first data is the X-ray detection data of the acetabular position of the target person, and the second data is the CT three-dimensional reconstruction data of the acetabular position of the target person; An identification module, configured to identify the acetabular fracture classification features based on the deep learning model for the first target data and the second target data; The central control module is used to send the acetabular fracture classification features to the first device, and the first device is used to display the acetabular fracture classification features.

[0008] In the third aspect of the embodiments of the present disclosure, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, the steps of the above-mentioned method for predicting acetabular fracture classification based on deep learning are implemented.

[0009] In the fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the above-mentioned method for predicting acetabular fracture classification based on deep learning are implemented.

[0010] The beneficial effects of the method and system for predicting acetabular fracture classification based on deep learning provided by the embodiments of the present disclosure are as follows: In the present disclosure, by preprocessing the X-ray detection data and CT three-dimensional reconstruction data based on the basic information of the person, and comprehensively analyzing the X-ray detection data and CT three-dimensional reconstruction data based on the deep learning model, the characteristics of acetabular fractures, such as the position, direction, and length of the fracture line, can be identified more accurately. Compared with the traditional manual interpretation method, the accuracy of judging the fracture line and fracture position in the patient's examination images is improved, thereby shortening the time spent by doctors in image observation and thus improving medical efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present disclosure. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0012] Figure 1 It is a schematic flowchart of a method for predicting acetabular fracture classification based on deep learning provided by an embodiment of the present disclosure; Figure 2 It is a schematic flowchart of an early fusion and a late fusion in a method for predicting acetabular fracture classification based on deep learning provided by an embodiment of the present disclosure; Figure 3 It is a structural block diagram of a system for predicting acetabular fracture classification based on deep learning provided by an embodiment of the present disclosure; Figure 4 It is a schematic block diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0013] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present disclosure. However, those skilled in the art should clearly understand that the present disclosure can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from hindering the description of the present disclosure.

[0014] As used in this disclosure, X-rays or X-rays refer to Roentgen rays, and CT refers to Computed Tomography, that is, computed tomography. Essentially, CT is a scan using X-rays. The difference is that CT irradiates the human body with X-rays in a sliced manner from multiple angles to obtain a large amount of cross-sectional image data, and the computer processes and reconstructs this data to finally form a three-dimensional stereoscopic image or a tomographic image of any plane. The principle will not be elaborated here.

[0015] To make the objectives, technical solutions, and advantages of the present disclosure clearer, the following will be described through specific embodiments in conjunction with the accompanying drawings.

[0016] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of a method for predicting acetabular fracture classification based on deep learning provided for an embodiment of the present disclosure. The method includes: S101: Process the first data based on the basic information of the target person to obtain the first target data, and process the second data to obtain the second target data; wherein, the first data is the X-ray detection data of the acetabular position of the target person, and the second data is the CT three-dimensional reconstruction data of the acetabular position of the target person.

[0017] In this embodiment, the target person refers to a patient undergoing acetabular fracture diagnosis, and the basic information may be the age, body mass index (BMI), gender, etc. of the target person.

[0018] The first target data is the X-ray detection data of the acetabular position of the target person after noise reduction processing, for example, it may be an X-ray image after noise reduction processing. The second target data is the CT three-dimensional reconstruction data of the acetabular position of the target person after noise reduction processing, for example, it may be a CT three-dimensional image after noise reduction processing.

[0019] Among them, the data source of the first data is an X-ray machine, and the data source of the second data is a CT machine.

[0020] In this embodiment, considering that the X-ray detection data and CT three-dimensional reconstruction data of people with different physical conditions are different, different processing methods should be adopted for people with different physical conditions.

[0021] For example, for infant or adolescent testers, in the current medical scenario, the corresponding radiation dose will be reduced accordingly, inevitably reducing a part of the image quality and resulting in some noise in the detection results. For doctors, they can make subjective judgments based on experience to exclude the influence of some noise and obtain the fracture classification results. However, for a computer, the collected data should be processed such as noise reduction before it can accurately identify whether there is a fracture in the data and the relevant features of the fracture classification.

[0022] S102: Identify the acetabular fracture classification features based on the deep learning model for the first target data and the second target data.

[0023] In this embodiment, after performing the aforementioned noise reduction processing, the first target data and the second target data can be input into the deep learning model. Among them, the deep learning model can be one or multiple. However, it should be noted that whether the deep learning model is one or multiple, it is trained through its corresponding training set.

[0024] For example, if there is only one deep learning model, when its input is two types of data, namely the first target data and the second target data, its training set should also include X-ray detection data and CT three-dimensional reconstruction data, and the output is the acetabular fracture classification features corresponding to the above two types of data.

[0025] In this embodiment, the acetabular fracture classification features refer to the data related to the doctor's judgment of the acetabular fracture classification results. For example, it can be the direction, length, width or position of the fracture line, or the size or shape of the fracture fragment, etc. The specific acetabular fracture classification features, their corresponding descriptions, and the role of assisting doctors in acetabular fracture classification are shown in Table 1.

[0026] Table 1 Acetabular fracture classification features and corresponding descriptions

[0027] S103: Send the acetabular fracture classification features to the first device, and the first device is used to display the acetabular fracture classification features.

[0028] In this embodiment, the first device is the terminal device that receives the acetabular fracture classification features. Its specific type can be various, such as the computers in the hospital imaging department, the mobile terminals of doctors, such as computers, smartphones, etc. The first device has a display function and can display the received acetabular fracture classification feature data.

[0029] In this embodiment, taking X-ray detection data as an example, the specific display form may be to mark the fracture line in the X-ray detection data (X-ray image) and label information such as length, width, or suspected fracture line. The acetabular fracture classification features referred to in this disclosure are extracted based on the X-ray data and CT three-dimensional reconstruction data of the target person, and are used to assist doctors in judging whether acetabular fracture occurs and providing relevant features for judging the acetabular fracture classification.

[0030] It can be concluded from the above that by preprocessing the X-ray detection data and CT three-dimensional reconstruction data based on the basic information of the person and comprehensively analyzing the X-ray detection data and CT three-dimensional reconstruction data based on the deep learning model, this disclosure can more accurately identify the features of acetabular fracture, such as the position, direction, and length of the fracture line. Compared with the traditional manual interpretation method, it improves the accuracy of judging the fracture line and fracture position in the patient's examination image, thereby shortening the time spent by doctors in image observation and improving medical efficiency.

[0031] In an embodiment of this disclosure, the basic information of the target person includes: age and body mass index; Processing the second data to obtain second target data, including: In response to the age of the target person being less than the preset age and the body mass index being less than or equal to the preset index, processing the second data based on the first noise reduction method to obtain second target data; In response to the body mass index of the target person being greater than the preset index and the age being greater than or equal to the preset age, processing the second data based on the second noise reduction method to obtain second target data; In response to the age of the target person being less than the preset age and the body mass index being greater than the preset index, processing the second data based on the third noise reduction method to obtain second target data; Among them, the noise reduction methods of the first noise reduction method, the second noise reduction method, and the third noise reduction method are different.

[0032] In this embodiment, different noise reduction methods are set for persons with different basic information. The preset age in this embodiment refers to the specified age limit, which is determined based on prior knowledge in the medical field and can generally be set to 12 - 14 years old, or 8 - 9 years old, etc. The preset index in this embodiment refers to the BMI index that will affect the X-ray detection data or CT three-dimensional reconstruction data and can be set to 28.

[0033] When the age of the target person is less than the preset age, when a doctor performs an acetabular fracture examination on him / her, the corresponding radiation dose will be reduced. Correspondingly, there will be certain impacts at the image level, such as noise. When the body mass index of the target person is greater than the preset index, it indicates that the target person is obese, and noise or artifacts may appear in the X-ray detection data and CT three-dimensional reconstruction data.

[0034] In this embodiment, the first noise reduction method can be adaptive median filtering, or noise reduction based on wavelet transform, etc. Taking noise reduction based on wavelet transform as an example, wavelet transform can decompose an image into different scales and directions. For the noise generated by low-dose scanning, this method can remove noise interference while retaining the high-frequency details of the image (such as bone texture), which is suitable for the situation where the bone features in this scenario are obvious but the image quality is affected by the radiation dose.

[0035] The second noise reduction method can be non-local means filtering. Non-local means filtering itself can utilize the self-similarity of the image to remove noise. During the filtering process, the pixel information that conforms to the bone features is retained, better removing noise and artifacts and retaining the true features of the bones.

[0036] The third noise reduction method can be a combination of the aforementioned two noise reduction methods, that is, the combination of adaptive median filtering and non-local means filtering, or the combination of noise reduction based on wavelet transform and non-local means filtering. The present disclosure does not limit the order relationship between the two sets of noise reduction methods.

[0037] In a preferred embodiment of the present disclosure, the third noise reduction method first performs adaptive median filtering or noise reduction based on wavelet transform in the time dimension, and then performs non-local means filtering. Taking noise reduction based on wavelet transform as an example, the noise reduction method based on wavelet transform can decompose the CT image into different scales and directions. By processing the wavelet coefficients, noise can be removed while retaining the high-frequency details of the image (such as acetabular bone texture). Performing this kind of noise reduction first in the present disclosure can effectively remove the noise generated by low-dose scanning and retain the important features of the image; when performing non-local means filtering then, the noise level of the image has been reduced and the bone features are well retained. Non-local means filtering can further utilize the self-similarity of the image to smooth the image, improving the overall quality of the image and not causing excessive damage to the previously retained high-frequency details.

[0038] If non - local means filtering is performed first in the present disclosure, global smoothing processing will be performed on the image, and some high - frequency details of the image will be blurred during this process. Then, when performing noise reduction based on wavelet transform, although the image can be further refined, due to the certain loss of high - frequency information caused by the previous non - local means filtering, the original high - frequency details of the CT image cannot be fully restored, resulting in the final image having a worse effect in retaining details such as bone texture than the case of performing wavelet transform - based noise reduction first.

[0039] In this embodiment, noise reduction processing is only performed on the second data, that is, the CT three - dimensional reconstruction data of the acetabular position of the target person; at the same time, as can be seen from the foregoing, CT essentially uses X - rays for scanning, so it should also be understood that the noise reduction methods and principles of the first data and the second data are the same, that is, processing the first data to obtain the first target data, including: Responding to the age of the target person being less than the preset age and the body mass index being less than or equal to the preset index, processing the first data based on the first noise reduction method to obtain the first target data; Responding to the body mass index of the target person being greater than the preset index and the age being greater than or equal to the preset age, processing the first data based on the second noise reduction method to obtain the first target data; Responding to the age of the target person being less than the preset age and the body mass index being greater than the preset index, processing the first data based on the third noise reduction method to obtain the first target data; Among them, the noise reduction methods of the first noise reduction method, the second noise reduction method, and the third noise reduction method are different.

[0040] When the age of the target person is greater than or equal to the preset age and the body mass index is less than or equal to the preset index, the radiation dose is within the normal range, and at the same time, it will not affect the X - ray detection data and the CT three - dimensional reconstruction data. Therefore, noise reduction processing does not need to be performed.

[0041] That is, responding to the age of the target person being greater than or equal to the preset age and the body mass index being less than or equal to the preset index, taking the first data as the first target data, and taking the second data as the target data.

[0042] In this embodiment, the process and principle of the noise reduction processing of the first data will not be elaborated.

[0043] It can be concluded from the above that the present disclosure selects an appropriate noise reduction method according to the age and body mass index of the target person, which helps to optimize the image quality, reduce noise and artifacts, and thus improve the recognition accuracy of the acetabular fracture characteristics by the subsequent deep learning model.

[0044] In an embodiment of the present disclosure, the first target data and the second target data are identified based on a deep learning model to obtain acetabular fracture classification features, including: Determine a target fusion period based on the similarity between the first target data and the second target data in multiple dimensions, where the target fusion period is the fusion period of the first target data and the second target data; Identify the first target data and the second target data based on the deep learning model corresponding to the target fusion period to obtain acetabular fracture classification features.

[0045] In this embodiment, the target fusion period includes early fusion and late fusion. The determination of the target fusion period is based on the similarity between the first target data and the second target data in multiple dimensions. The essence of the first target data and the second target data is an image, so the dimensions can include: aspects such as image quality, resolution, and data distribution. Data distribution refers to the distribution of other aforementioned acetabular fracture classification features for the fracture line on the image, which can be specifically reflected in the distribution of quantity or position.

[0046] In this embodiment, the target fusion period can be determined in the following manner: Determine the target fusion period based on the similarity between the first target data and the second target data in multiple dimensions, including: In response to the similarity between the first target data and the second target data in multiple dimensions being greater than the target similarity, use early fusion as the target fusion period; In response to the similarity of any one dimension between the first target data and the second target data being less than the target similarity, use late fusion as the target fusion period.

[0047] In this embodiment, the similarity calculation methods for each dimension can be different. For example, for the dimension of image quality, it can be obtained by calculating the mean square error of two images. For the resolution dimension, it can be determined by simply taking the ratio of the differences. For example, subtract the resolutions of two images and use one of the resolutions as the denominator. The larger the resulting value, the smaller the similarity, and the similarity can be obtained according to the mapping table relationship. The similarity of the data distribution dimension can be compared through histograms. For example, it can be obtained by calculating the Bhattacharyya distance or the chi-square distance, etc.

[0048] It should be noted that for the similarities of the above-mentioned each dimension, different target similarities should be set according to their calculation methods, or the similarities of the above-mentioned each dimension should be normalized, and at this time, there can be a corresponding target similarity.

[0049] If the X-ray and CT three-dimensional reconstruction data have high similarity in terms of image quality, resolution, data distribution, etc., it is more appropriate to perform early fusion on the first target data and the second target data. Because at this time, after directly fusing the two types of data, it is easier for the deep learning model to learn a unified feature representation.

[0050] For example, when the basic parameters such as the gray level range and pixel pitch of the X-ray image and the CT image are similar, early fusion can make full use of these similarities, extract features from both types of data simultaneously, and mine the complementary information between them.

[0051] On the contrary, if the two types of data are quite different in these aspects, late fusion has more advantages. Because late fusion can first perform preprocessing and feature extraction on different modality data respectively, using methods suitable for the characteristics of their respective data, better handle the data differences, and avoid the difficulties in training the deep learning model caused by data differences in early fusion.

[0052] From the above, it can be concluded that the present disclosure determines the target fusion period based on the similarity of the first target data (X-ray image) and the second target data (CT three-dimensional reconstruction data) in multiple dimensions, enabling the present disclosure to dynamically select the optimal fusion strategy, make full use of the complementary information between the two types of data, avoid the negative impacts brought by data differences, and thus improve the recognition performance of the deep learning model for the acetabular fracture classification features.

[0053] Figure 2 This is a schematic flowchart of early fusion and late fusion in a method for predicting acetabular fracture classification based on deep learning provided by an embodiment of the present disclosure. Refer to Figure 2 , in an embodiment of the present disclosure, the target fusion period includes early fusion; Based on the deep learning model corresponding to the target fusion period, the first target data and the second target data are recognized to obtain acetabular fracture classification features, including: In response to the target fusion period being early fusion, the first target data and the second target data are fused to obtain third target data; Based on the first deep learning model, the third target data is recognized to obtain acetabular fracture classification features.

[0054] In this embodiment, when it is determined that the target fusion period is early fusion, it indicates that the first target data and the second target data have high similarity. Directly fusing them into third target data allows the first deep learning model to comprehensively utilize the information of both types of data from the beginning. The advantage of doing this is that the model can learn features in a unified data space, mine the complementary information between the two types of data, avoid the information fragmentation problem that may be caused by processing the data separately, and thus more effectively learn the features related to acetabular fracture classification.

[0055] It should be noted that since the first target data is essentially two-dimensional image data and the second target data is essentially three-dimensional image data during early fusion, the three-dimensional image data can be converted into two-dimensional images for fusion, or the two-dimensional X-ray images can be converted into three-dimensional voxel representations. For example, an additional dimension can be added to the two-dimensional image and it can be regarded as three-dimensional voxel data with a thickness of 1, and then it can be fused with the CT three-dimensional data in the voxel space. During the fusion process, weighted summation or other fusion operations can be performed on the voxel values to integrate the information of the two types of data.

[0056] The training set of the first deep learning model consists of a data group composed of multiple X-ray detection data and their corresponding CT detection data, and multiple acetabular fracture classification features corresponding to this data group.

[0057] Reference Figure 2 , in another embodiment of the present disclosure, in addition to the aforementioned early fusion, there is also late fusion. The data processing method of the deep learning model corresponding to late fusion is different from that of the deep learning model corresponding to early fusion, that is, the target fusion period includes late fusion; Based on the deep learning model corresponding to the target fusion period, the first target data and the second target data are recognized to obtain acetabular fracture classification features, including: In response to the target fusion period being late fusion, the first target data is recognized based on the second deep learning model to obtain X-ray classification features; The second target data is recognized based on the third deep learning model to obtain CT classification features; The X-ray classification features and the CT classification features are weighted and fused to obtain comprehensive classification features; Based on the classifier, the comprehensive classification features are classified to obtain acetabular fracture classification features; Among them, the training sets of the second deep learning model and the third deep learning model are different.

[0058] In this embodiment, if the target fusion period is late fusion, it indicates that the first target data and the second target data are quite different. At this time, the second deep learning model and the third deep learning model are first used to process the first target data and the second target data respectively because different types of data may require different processing methods and feature extraction methods.

[0059] There are significant differences in the imaging principles, data structures, and information characteristics between X-ray images and CT three-dimensional reconstruction images. Different training sets are used to train different models, enabling each deep learning model to better adapt to the characteristics of the corresponding data and more accurately extract the classification features of acetabular fractures. The training set of the second deep learning model consists of multiple X-ray detection data and their corresponding acetabular fracture classification features, and the training set of the third deep learning model consists of multiple CT detection data and their corresponding acetabular fracture classification features.

[0060] After obtaining the X-ray classification features and CT classification features, they are weighted and fused to obtain comprehensive classification features. Different types of data contribute differently to the classification of acetabular fractures. By weighting, the two types of feature information can be more reasonably integrated. Finally, a classifier is used to classify the comprehensive classification features because the comprehensive classification features are just a feature vector containing various information and need to be output through the classifier. Since the features output by the deep learning model are an abstract representation of the acetabular fracture images, they only describe various attributes and characteristics of the images. For example, features such as the position of the fracture line, the length of the fracture line, and displacement mentioned above cannot directly determine which type of fracture it belongs to or give a clinical diagnosis result. Therefore, a classifier is needed to map them to the image results, that is, the role of the classifier is to transform the abstract feature data into visual data that can be understood by doctors.

[0061] In the aforementioned early fusion process, the final result output also requires the use of a classifier, and the classifier itself is a component of the deep learning model. Therefore, during the training process, the classifier can learn how to output the acetabular fracture classification features, which will not be elaborated in this disclosure. The deep learning model referred to in this disclosure can be a convolutional neural network or a recurrent neural network, etc.

[0062] It can be concluded from the above that in early fusion, by directly fusing the first target data (X-ray images) and the second target data (CT three-dimensional reconstruction images), this disclosure can comprehensively utilize the information of the two types of data at the beginning, which helps the model learn a more comprehensive feature representation, thereby improving the accuracy of acetabular fracture classification. The late fusion strategy is aimed at the situation where the data differences are large. By separately processing the X-ray images and CT three-dimensional reconstruction images and using different deep learning models for feature extraction, it ensures that the model can adapt to different types of data and enhances the generalization ability of this disclosure.

[0063] In an embodiment of this disclosure, the method for predicting the classification of acetabular fractures based on deep learning further includes: in response to the number of fracture line features in the X-ray classification features and / or CT classification features being greater than the first number, reducing the reference value of the X-ray classification weight based on the first step length to obtain the X-ray classification weight; In response to the number of fracture line features in both the X-ray classification features and the CT classification features being less than the second quantity, increase the reference value of the X-ray classification weight based on the second step size to obtain the X-ray classification weight. Among them, the first quantity is greater than the second quantity, and the X-ray classification weight is the weight corresponding to the X-ray classification features. The first quantity and the second quantity can be determined based on the actual situation or experience.

[0064] In this embodiment, considering that when acetabular fractures are relatively complex, a single X-ray image may be difficult to comprehensively and accurately display the details and spatial structure of the fractures. At this time, the CT three-dimensional reconstruction data can provide more detailed fracture information, including the size, shape, position of the fracture fragments, and their spatial relationships, etc. Therefore, for complex fractures, the weight of the CT classification features in the late fusion can be appropriately increased.

[0065] For relatively simple acetabular fractures, such as simple linear fractures or fractures with fewer fracture fragments and insignificant displacement, the X-ray image may be able to provide sufficient information for accurate classification. In this case, the reliability of the X-ray classification features is relatively high, and their weight in the late fusion can be increased accordingly.

[0066] Therefore, when the number of fracture line features in the X-ray classification features and / or the CT classification features is greater than the first quantity, this indicates that the fracture situation is relatively complex, while when the number of fracture line features in both the X-ray classification features and / or the CT classification features is less than the second quantity, it indicates that the fracture situation is relatively simple. Adjustment can be made according to the aforementioned logic.

[0067] The formula for the first step size can be The formula for the second step size can be , where represents the first step size, represents the second step size, represents the first proportionality coefficient, represents the second proportionality coefficient, represents the natural constant, represents the number of fracture line features in the X-ray classification features and the CT classification features, represents the first quantity, represents the second quantity. The first proportionality coefficient and the second proportionality coefficient can be preset according to experience.

[0068] The reference value of the X-ray classification weight is determined during the experiment process or defaulted to 0.5. The sum of the X-ray classification weight and the CT classification weight is 1. The CT classification weight is the weight corresponding to the CT classification features, that is, when adjusting the reference value of the X-ray classification weight, the reference value of the CT classification weight should also be adjusted in the opposite direction with the same step size.

[0069] In response to the number of fracture line features in both the X-ray classification features and the CT classification features being less than or equal to the first number and greater than or equal to the second number, the reference value of the X-ray classification weight is used as the X-ray classification weight, and the reference value of the CT classification weight is used as the CT classification weight. At this time, the number of fracture lines is in a moderate range, and the reference values of the above two weights can be not adjusted.

[0070] As can be seen from the above, the present disclosure adjusts the weights of the two classification features according to the number of fracture line features, so as to ensure more accurate classification prediction results in both complex and simple fracture cases. When the fracture situation is complex, the present disclosure increases the weight of the CT classification features to obtain more detailed fracture information; when the fracture situation is simple, the present disclosure increases the weight of the X-ray classification features to make full use of its high reliability, which helps to improve the accuracy and reliability of the extraction of acetabular fracture analysis features.

[0071] Corresponding to the above-mentioned embodiment of the deep learning-based acetabular fracture classification prediction method, Figure 3 is a structural block diagram of a deep learning-based acetabular fracture classification prediction system provided by an embodiment of the present disclosure. For the convenience of description, only the parts related to the embodiments of the present disclosure are shown. Refer to Figure 3 The deep learning-based acetabular fracture classification prediction system 20 includes: a preprocessing module 21, an identification module 22, and a central control module 23.

[0072] Among them, the preprocessing module 21 is configured to process the first data based on the basic information of the target person to obtain the first target data, and process the second data to obtain the second target data; wherein, the first data is the X-ray detection data of the acetabular position of the target person, and the second data is the CT three-dimensional reconstruction data of the acetabular position of the target person; The identification module 22 is configured to identify the first target data and the second target data based on a deep learning model to obtain acetabular fracture classification features; The central control module 23 is configured to send the acetabular fracture classification features to a first device, and the first device is configured to display the acetabular fracture classification features.

[0073] In an embodiment of the present disclosure, the basic information includes: age and body mass index; The preprocessing module 21 is specifically configured to, in response to the age of the target person being less than a preset age and the body mass index being less than or equal to a preset index, process the second data based on a first noise reduction method to obtain the second target data; In response to the body mass index of the target person being greater than the preset index and the age being greater than or equal to the preset age, process the second data based on a second noise reduction method to obtain the second target data; In response to the age of the target person being less than the preset age and the body mass index being greater than the preset index, the second data is processed based on the third noise reduction method to obtain the second target data; Among them, the noise reduction methods of the first noise reduction method, the second noise reduction method, and the third noise reduction method are different.

[0074] In an embodiment of the present disclosure, the recognition module 22 is specifically configured to determine the target fusion period based on the similarity between the first target data and the second target data in multiple dimensions, and the target fusion period is the fusion period of the first target data and the second target data; Based on the deep learning model corresponding to the target fusion period, the first target data and the second target data are recognized to obtain the acetabular fracture classification features.

[0075] In an embodiment of the present disclosure, the recognition module 22 is further specifically configured to, in response to the similarity between the first target data and the second target data in multiple dimensions being greater than the target similarity, use early fusion as the target fusion period; In response to the similarity of any dimension among the multiple dimensions of the first target data and the second target data being less than the target similarity, use late fusion as the target fusion period.

[0076] In an embodiment of the present disclosure, the target fusion period includes early fusion; The recognition module 22 is further specifically configured to, in response to the target fusion period being early fusion, fuse the first target data and the second target data to obtain the third target data; Based on the first deep learning model, the third target data is recognized to obtain the acetabular fracture classification features.

[0077] In an embodiment of the present disclosure, the target fusion period includes late fusion; The recognition module 22 is further specifically configured to, in response to the target fusion period being late fusion, recognize the first target data based on the second deep learning model to obtain the X-ray classification features; Based on the third deep learning model, the second target data is recognized to obtain the CT classification features; The X-ray classification features and the CT classification features are weighted and fused to obtain the comprehensive classification features; Based on the classifier, the comprehensive classification features are classified to obtain the acetabular fracture classification features; Among them, the training set of the second deep learning model is different from the training set of the third deep learning model.

[0078] In one embodiment of the present disclosure, the acetabular fracture classification prediction system 20 based on deep learning further includes: a weight adjustment module, configured to, in response to the number of fracture line features in the X-ray classification features and / or the CT classification features being greater than a first number, reduce the reference value of the X-ray classification weight based on a first step length to obtain the X-ray classification weight; in response to the number of fracture line features in both the X-ray classification features and the CT classification features being less than a second number, increase the reference value of the X-ray classification weight based on a second step length to obtain the X-ray classification weight; wherein the first number is greater than the second number, and the X-ray classification weight is the weight corresponding to the X-ray classification features.

[0079] See Figure 4 , Figure 4 is a schematic block diagram of an electronic device provided by an embodiment of the present disclosure. As Figure 4 shown, the electronic device 300 in this embodiment may include: one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The above-mentioned processors 301, input devices 302, output devices 303, and memories 304 communicate with each other through a communication bus 305. The memory 304 is used to store a computer program, and the computer program includes program instructions. The processor 301 is configured to execute the program instructions stored in the memory 304. Among them, the processor 301 is configured to call the program instructions to execute the functions of each module / unit in the above-mentioned device embodiments, such as Figure 3 the functions of the preprocessing module 21, the recognition module 22, and the central control module 23 shown.

[0080] It should be understood that in the embodiments of the present disclosure, the so-called processor 301 may be a central processing unit (CPU), and this processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0081] The input device 302 may include a touchpad, a fingerprint acquisition sensor (for acquiring the fingerprint information and the direction information of the fingerprint of the user), a microphone, etc., and the output device 303 may include a display (such as an LCD), a speaker, etc.

[0082] The memory 304 may include a read-only memory and a random access memory, and provide instructions and data to the processor 301. A part of the memory 304 may further include a non-volatile random access memory. For example, the memory 304 may also store information about the device type.

[0083] In a specific implementation, the processor 301, the input device 302, and the output device 303 described in the embodiments of the present disclosure may implement the implementation manners described in the first and second embodiments of the acetabular fracture classification prediction method based on deep learning provided in the embodiments of the present disclosure, and may also implement the implementation manner of the electronic device described in the embodiments of the present disclosure, which will not be elaborated herein.

[0084] In another embodiment of the present disclosure, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, all or part of the processes in the methods of the above embodiments are implemented. It may also be completed by instructing relevant hardware through the computer program. The computer program may be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above method embodiments may be implemented. Among them, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0085] The computer-readable storage medium may be an internal storage unit of the electronic device in any of the foregoing embodiments, such as the hard disk or memory of the electronic device. The computer-readable storage medium may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device. Further, the computer-readable storage medium may include both an internal storage unit and an external storage device of the electronic device. The computer-readable storage medium is used to store the computer program and other programs and data required by the electronic device. The computer-readable storage medium may also be used to temporarily store data that has been output or will be output.

[0086] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this disclosure.

[0087] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described electronic devices and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0088] In several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed couplings or direct couplings or communication connections to each other can be indirect couplings or communication connections through some interfaces or units, or can also be in electrical, mechanical, or other forms of connection.

[0089] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of this disclosure.

[0090] In addition, the functional units in various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0091] The above is only the specific implementation manner of this disclosure, but the protection scope of this disclosure is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or substitutions within the technical scope disclosed by this disclosure, and these modifications or substitutions should all be covered within the protection scope of this disclosure. Therefore, the protection scope of this disclosure should be subject to the protection scope of the claims.

Claims

1. A deep learning-based prediction method for acetabular fracture classification, characterized in that: include: The first data is processed based on the basic information of the target person to obtain the first target data, and the second data is processed to obtain the second target data; wherein the first data is the X-ray detection data of the acetabulum position of the target person, and the second data is the CT three-dimensional reconstruction data of the acetabulum position of the target person; The first target data and the second target data are identified based on the deep learning model to obtain the classification characteristics of acetabular fractures; The acetabular fracture classification feature is sent to a first device, and the first device is used to display the acetabular fracture classification feature.

2. The acetabular fracture classification prediction method based on deep learning according to claim 1, characterized in that: The basic information includes: age and body mass index; The processing of the second data to obtain the second target data includes: In response to the target person's age being less than a preset age and having a body mass index less than or equal to a preset index, processing the second data based on a first noise reduction method to obtain second target data; In response to the body mass index of the target person being greater than a preset index and the age being greater than or equal to the preset age, processing the second data based on a second noise reduction method to obtain second target data; In response to the target person's age being less than a preset age and body mass index being greater than a preset index, processing the second data based on a third noise reduction method to obtain second target data; The first noise reduction method, the second noise reduction method and the third noise reduction method are different noise reduction methods.

3. The acetabular fracture classification prediction method based on deep learning according to claim 1, characterized in that: The first target data and the second target data are identified based on the deep learning model to obtain the acetabular fracture classification features, including: Determining a target fusion period based on similarities between the first target data and the second target data in multiple dimensions, the target fusion period being a fusion period of the first target data and the second target data; The first target data and the second target data are identified based on the deep learning model corresponding to the target fusion period to obtain acetabular fracture classification characteristics.

4. The acetabular fracture classification prediction method based on deep learning as claimed in claim 3, characterized in that: The determining the target fusion period based on the similarity between the first target data and the second target data in multiple dimensions includes: In response to similarities between the first target data and the second target data in multiple dimensions being greater than target similarities, using early fusion as the target fusion period; In response to the similarity of any dimension of the plurality of dimensions between the first target data and the second target data being less than the target similarity, late fusion is used as the target fusion period.

5. The acetabular fracture classification prediction method based on deep learning as claimed in claim 3, characterized in that: The target fusion period includes early fusion; The first target data and the second target data are identified based on the deep learning model corresponding to the target fusion period to obtain acetabular fracture classification features, including: In response to the target fusion period being early fusion, fusing the first target data and the second target data to obtain third target data; The third target data is identified based on the first deep learning model to obtain acetabular fracture classification characteristics.

6. The acetabular fracture classification prediction method based on deep learning as claimed in claim 3, characterized in that: The target fusion period includes late fusion; The first target data and the second target data are identified based on the deep learning model corresponding to the target fusion period to obtain acetabular fracture classification features, including: In response to the target fusion period being late fusion, identifying the first target data based on a second deep learning model to obtain an X-ray typing feature; Identify the second target data based on the third deep learning model to obtain CT typing features; Performing weighted fusion on the X-ray typing feature and the CT typing feature to obtain a comprehensive typing feature; Classifying the comprehensive classification features based on the classifier to obtain acetabular fracture classification features; Among them, the training set of the second deep learning model is different from the training set of the third deep learning model.

7. The acetabular fracture classification prediction method based on deep learning according to claim 6, characterized in that: Also includes: In response to the number of fracture line features in the X-ray typing feature and / or the CT typing feature being greater than a first number, reducing a reference value of the X-ray typing weight based on the first step length to obtain an X-ray typing weight; In response to the number of fracture line features in the X-ray typing feature and the CT typing feature being both less than a second number, increasing a reference value of the X-ray typing weight based on a second step length to obtain an X-ray typing weight; The first number is greater than the second number, and the X-ray typing weight is the weight corresponding to the X-ray typing feature.

8. A deep learning-based acetabular fracture classification prediction system, characterized in that: include: A preprocessing module, used to process the first data based on the basic information of the target person to obtain the first target data, and to process the second data to obtain the second target data; wherein the first data is the X-ray detection data of the acetabulum position of the target person, and the second data is the CT three-dimensional reconstruction data of the acetabulum position of the target person; An identification module, used to identify the first target data and the second target data based on a deep learning model to obtain acetabular fracture classification characteristics; The central control module is used to send the acetabular fracture classification characteristics to the first device, and the first device is used to display the acetabular fracture classification characteristics.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images