Obstetric medical image recognition method and system based on image processing
By preprocessing, extracting features, and dynamically calibrating fetal images, the problem of low accuracy in fetal image recognition was solved, achieving higher recognition accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIANGYA HOSPITAL CENT SOUTH UNIV
- Filing Date
- 2023-06-14
- Publication Date
- 2026-05-29
AI Technical Summary
Current technologies for fetal image recognition have low accuracy, which can easily lead to medical misjudgments.
An image processing-based method is used to preprocess images of fetuses in the pregnant woman's womb, extract fetal feature maps, select selected feature maps by fetal masking coefficient and feature influence, perform segmentation and mask merging, determine fetal feature identification information, and compare it with a preset threshold to obtain the fetal identification image accuracy, and perform dynamic calibration to improve the recognition accuracy.
It significantly improves the accuracy of fetal image recognition and reduces the risk of medical misdiagnosis.
Smart Images
Figure CN116596917B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of obstetric medical image processing technology, and more specifically, to an obstetric medical image recognition method and system based on image processing. Background Technology
[0002] With the continuous advancement of medical imaging technology and the development of digital image processing methods, the processing of obstetric medical images can provide doctors with more accurate, rapid, and reliable diagnoses. Among these, fetal image processing has significant meaning and application value in prenatal diagnosis, fetal development monitoring, and congenital malformation detection. In existing technologies, fetal images can be created by sending ultrasound pulses and receiving their reflected signals. Ultrasound imaging generates images by measuring the propagation speed and reflection degree of sound waves in human tissues. During obstetric and gynecological examinations, doctors place an ultrasound probe on the pregnant woman's abdomen. The probe sends ultrasound waves and receives their reflected signals. After processing, these signals can generate fetal images. For the generated fetal images, experienced doctors need to manually identify the fetal organs, limbs, facial features, etc. However, the accuracy of manual identification is low, and misjudgments may even occur, leading to serious medical accidents. Summary of the Invention
[0003] This invention application provides a method and system for obstetric medical image recognition based on image processing, in order to solve the technical problem of low accuracy in fetal image recognition in the prior art.
[0004] To solve the above-mentioned technical problems, this application adopts the following technical solution:
[0005] In a first aspect, this application provides a method for obstetric medical image recognition based on image processing, comprising the following steps:
[0006] The images of the fetus in the pregnant woman's abdomen were preprocessed to obtain preprocessed fetal images;
[0007] The preprocessed fetal image is decomposed to obtain a preprocessed fetal image feature set, and different categories of fetal feature maps are extracted from the preprocessed fetal image feature set;
[0008] Fetal feature parameters are extracted from fetal feature images of each category. Fetal mask coefficients are determined based on the extracted fetal feature parameters. Fetal feature images of each category are selected based on the influence of fetal features. The fetal feature image with the highest feature influence in each category is selected as the selected feature image and used as the fetal template for that category. Fetal segmentation mask is determined based on the fetal mask coefficients and the fetal template. The fetal segmentation mask is obtained by weighted merging of the fetal template and the mask coefficients. Fetal feature identification information is then determined based on the fetal segmentation mask.
[0009] The fetal feature identification information is compared with a preset fetal information threshold to determine the fetal identification image difference, thereby obtaining the fetal identification image accuracy. The fetal correction feedback amount is obtained from the fetal identification image accuracy.
[0010] Based on the fetal correction feedback, the fetal feature map is dynamically calibrated to obtain the fetal body shape feature map.
[0011] In some embodiments, preprocessing the acquired images of the fetus in the pregnant woman's abdomen to obtain preprocessed fetal images may specifically include:
[0012] Determine the average grayscale value of the fetal image;
[0013] Determine the standard deviation of the grayscale values in the fetal image;
[0014] The fetal image is standardized based on its average grayscale value and the standard deviation of its grayscale values to obtain a preprocessed fetal image. The standardization of the fetal image is determined by the following formula:
[0015]
[0016] in, Standardize the grayscale values of the fetal image. The grayscale value of the fetal image. The average grayscale value of the fetal image. This represents the standard deviation of the grayscale values in the fetal image.
[0017] In some embodiments, the average grayscale value of the fetal image can be determined by the following formula:
[0018]
[0019] in, The average grayscale value of the fetal image. The number of grayscale values in the fetal image. Indicates the first Gray values of a fetal image.
[0020] In some embodiments, the standard deviation of the grayscale values of the fetal image can be determined by the following formula:
[0021]
[0022] in, The standard deviation of the grayscale values in the fetal image. The average grayscale value of the fetal image. The number of grayscale values in the fetal image. Indicates the first Gray values of a fetal image.
[0023] In some embodiments, decomposing the preprocessed fetal image to obtain a preprocessed fetal image feature set may specifically include:
[0024] The gray values in the preprocessed fetal image data are decomposed into fetal feature components with low temporal frequency and high spatial frequency according to their spatial location and time sequence in the image.
[0025] For each fetal characteristic component, a predetermined number of fetal characteristic functions are obtained;
[0026] Based on the fetal feature function obtained from each fetal feature component, feature information is extracted from the preprocessed fetal image to obtain the fetal feature mode;
[0027] The fetal feature modalities are combined to obtain the features of the original fetal image at various spatial and temporal scales, thereby obtaining the preprocessed fetal image feature set.
[0028] In some embodiments, the fetal feature components are features of the preprocessed fetal image at different temporal and spatial resolutions.
[0029] In some embodiments, determining the fetal identification image difference by comparing the fetal feature identification information with a preset fetal information threshold, and then obtaining the fetal identification image accuracy, may specifically include:
[0030] The difference between the fetal feature identification information and the preset fetal information threshold is calculated to obtain the fetal identification image difference.
[0031] Based on the difference between the fetal identification images, when the difference between the fetal identification images is less than 1, the fetal identification image is considered to be real and has the same number of pixels. Then, by calculating the proportion of the same pixels in the fetal identification images, the realism rate of the fetal identification images is obtained.
[0032] Secondly, this application provides an obstetric medical image recognition system based on image processing, which includes:
[0033] The fetal image preprocessing module is used to preprocess the acquired images of the fetus in the pregnant woman's abdomen to obtain preprocessed fetal images.
[0034] The fetal feature image extraction module decomposes the preprocessed fetal image to obtain a preprocessed fetal image feature set, and extracts different categories of fetal feature images from the preprocessed fetal image feature set;
[0035] The fetal feature identification information determination module is used to extract fetal feature parameters from fetal feature images of each category, determine the fetal mask coefficient of the fetal feature image of that category based on the extracted fetal feature parameters, perform a selection operation on the fetal feature images of that category based on the fetal feature influence, select the fetal feature image with the highest feature influence in that category as the selected feature image, use the selected feature image as the fetal template of that category, determine the fetal segmentation mask of that category based on the fetal mask coefficient and the fetal template, the fetal segmentation mask is obtained by weighted merging of the fetal template and the mask coefficient, and then determine the fetal feature identification information based on the fetal segmentation mask;
[0036] The fetal corrective feedback quantity determination module is used to compare the fetal feature identification information with a preset fetal information threshold, determine the fetal identification image difference, and then obtain the fetal identification image accuracy. The fetal corrective feedback quantity is obtained from the fetal identification image accuracy.
[0037] The fetal body shape feature map determination module is used to dynamically calibrate the fetal feature map based on the fetal correction feedback amount, thereby obtaining the fetal body shape feature map.
[0038] Thirdly, this application provides a computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described image processing-based obstetric medical image recognition method.
[0039] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described image processing-based obstetric medical image recognition method.
[0040] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:
[0041] The present application provides a method and system for obstetric medical image recognition based on image processing. First, preprocessing is performed on the acquired images of the fetus in the pregnant woman's abdomen to obtain preprocessed fetal images. The preprocessed fetal images are then decomposed to obtain a preprocessed fetal image feature set. Different categories of fetal feature images are extracted from the preprocessed fetal image feature set. Fetal feature parameters are extracted for each category of fetal feature images. Based on the extracted fetal feature parameters, a fetal mask coefficient for that category of fetal feature images is determined. A selection operation is performed on the fetal feature images of that category based on the influence of fetal features, selecting the fetal feature image with the highest feature influence in that category as the selected feature image. This selected feature image is then used as the reference for that category. A fetal template is used to determine a fetal segmentation mask for a given category based on the fetal masking coefficients and the fetal template itself. This fetal segmentation mask is obtained by weighted merging of the fetal template and the masking coefficients. Fetal feature identification information is then determined based on the fetal segmentation mask. The difference between the fetal feature identification information and a preset fetal information threshold is determined, thereby obtaining the fetal feature image accuracy. From this accuracy, a fetal correction feedback value is obtained. Based on the fetal correction feedback value, the fetal feature map is dynamically calibrated to obtain a fetal body shape feature map. This achieves intelligent obstetric medical image recognition, effectively improving the accuracy of fetal image recognition. Attached Figure Description
[0042] Figure 1 This is an exemplary flowchart of an image processing-based obstetric medical image recognition method according to some embodiments of this application;
[0043] Figure 2 These are schematic diagrams of exemplary hardware and / or software of an image processing-based obstetric medical image recognition system according to some embodiments of this application;
[0044] Figure 3 This is an exemplary structural diagram of a computer device that applies an image processing-based obstetric medical image recognition method, as shown in some embodiments of this application. Implementation
[0045] This application provides a method and system for obstetric medical image recognition based on image processing. The core of the method involves preprocessing acquired images of the fetus in the pregnant woman's abdomen to obtain preprocessed fetal images; decomposing the preprocessed fetal images to obtain a preprocessed fetal image feature set; extracting different categories of fetal feature images from the preprocessed fetal image feature set; extracting fetal feature parameters for each category of fetal feature images; determining the fetal mask coefficient for that category of fetal feature images based on the extracted fetal feature parameters; and selecting the fetal feature images of that category with the highest feature influence based on the fetal feature influence, thus selecting the selected fetal feature images. The feature map serves as a fetal template for this category. Based on the fetal mask coefficients and the fetal template, a fetal segmentation mask for this category is determined. The fetal segmentation mask is obtained by weighted merging of the fetal template and the mask coefficients. Fetal feature identification information is then determined based on the fetal segmentation mask. The difference between the fetal feature identification information and a preset fetal information threshold is determined, thereby obtaining the fetal identification image accuracy. The fetal correction feedback is obtained from the fetal identification image accuracy. Based on the fetal correction feedback, the fetal feature map is dynamically calibrated to obtain a fetal body shape feature map, which can effectively improve the recognition accuracy of fetal images.
[0046] To better understand the above technical solutions, a detailed description of the solutions will be provided below in conjunction with the accompanying drawings and specific implementation methods. (Reference) Figure 1 The figure is an exemplary flowchart of an image processing-based obstetric medical image recognition method according to some embodiments of this application. The image processing-based obstetric medical image recognition method 100 mainly includes the following steps:
[0047] In step S101, the collected images of the fetus in the pregnant woman's abdomen are preprocessed to obtain preprocessed fetal images.
[0048] It should be noted that image preprocessing of the acquired fetal images from the pregnant woman's abdomen is one of the important steps in obstetric medical image processing. Its purpose is to improve the quality of the fetal image, reduce noise in the fetal image, and enhance the details of the fetal image, so as to provide better input data for subsequent tasks such as feature extraction, segmentation, and recognition. In some embodiments of this application, image preprocessing of the acquired fetal images from the pregnant woman's abdomen is performed by calculating the average gray value of the fetal image and the standard deviation of the gray values of the fetal image, and then standardizing the fetal image according to the average gray value of the fetal image and the standard deviation of the gray values of the fetal image, thereby obtaining the preprocessed fetal image.
[0049] In practice, the average grayscale value of the fetal image can be determined by the following formula:
[0050]
[0051] in, The average grayscale value of the fetal image. The number of grayscale values in the fetal image. Indicates the first Grayscale values of individual fetal images;
[0052] In addition, the standard deviation of the grayscale values of the fetal image can be determined by the following formula:
[0053]
[0054] in, The standard deviation of the grayscale values in the fetal image. The average grayscale value of the fetal image. The number of grayscale values in the fetal image. Indicates the first Grayscale values of individual fetal images;
[0055] In addition, the standardized grayscale value of a fetal image can be determined by the following formula:
[0056]
[0057] in, Standardize the grayscale values of the fetal image. The grayscale value of the fetal image. The average grayscale value of the fetal image. This represents the standard deviation of the grayscale values in the fetal image.
[0058] It should be noted that acquired fetal images typically contain unavoidable noise, such as artifacts and speckles. Denoising algorithms, such as median filtering and Gaussian filtering, can be used to smooth the images and reduce the impact of noise. In addition, due to various factors during the acquisition process, fetal images may be distorted or rotated. Geometric correction methods, such as image rotation, translation, and scaling, can be used to adjust the image to the correct position and scale. If the acquired fetal images include images from multiple time points or different modalities (such as ultrasound and MRI images), image registration is required to ensure that they correspond to the same areas in space for effective comparison and analysis.
[0059] In step S102, the preprocessed fetal image is decomposed to obtain a preprocessed fetal image feature set, and different categories of fetal feature maps are obtained from the preprocessed fetal image feature set.
[0060] In some embodiments, the gray values in the preprocessed fetal image are decomposed into fetal feature components with low temporal frequency and high spatial frequency according to their spatial location and time sequence in the image. A predetermined number of fetal feature functions are obtained for each obtained fetal feature component. Feature information is extracted from the preprocessed fetal image based on the fetal feature functions obtained for each fetal feature component to obtain fetal feature modes. The fetal feature modes are combined to obtain features at various spatial and temporal scales of the original fetal image. The feature set of the preprocessed fetal image is then determined based on the features at various spatial and temporal scales of the original fetal image.
[0061] It should be noted that, in this application, fetal feature components refer to the features of the preprocessed fetal image at different temporal and spatial resolutions. Fetal feature functions are a set of functions representing the feature signals of the preprocessed fetal image, usually basis functions obtained by techniques such as Fourier transform or wavelet analysis. Fetal feature functions are used to describe the spatial structure and temporal variation of the fetal feature components, while fetal feature modes represent the features of the preprocessed fetal image at different spatial and temporal scales, used to capture the changes at different frequencies or scales in the preprocessed fetal image. Each fetal feature mode can be regarded as a component of the fetal feature mode at a specific scale or frequency. By combining the fetal feature modes, the features of the original fetal image at various spatial and temporal scales can be restored, and then the feature set of the preprocessed fetal image can be determined from the features of the original fetal image at various spatial and temporal scales. In addition, in some embodiments, fetal feature modes may correspond to important structures, morphologies and features in the fetal image.
[0062] In step S103, fetal feature parameters are extracted from the fetal feature images of each category. The fetal mask coefficient of the fetal feature image of that category is determined based on the extracted fetal feature parameters. A selection operation is performed on the fetal feature images of that category based on the influence of fetal features. The fetal feature image with the highest feature influence in that category is selected as the selected feature image. The selected feature image is used as the fetal template for that category. The fetal segmentation mask of that category is determined based on the fetal mask coefficient and the fetal template. The fetal segmentation mask is obtained by weighted merging of the fetal template and the mask coefficient. Then, the fetal feature identification information is determined based on the fetal segmentation mask.
[0063] In some embodiments, fetal feature parameters are extracted for each category of fetal feature image. These fetal feature parameters refer to feature parameters related to the fetal image in that category, such as size, shape, or texture features, which can be used to describe the fetal characteristics of that category. These fetal feature parameters are typically obtained through evaluation and analysis combining actual observation results and measurement data of the fetal feature images. Then, a fetal masking coefficient for the fetal feature image is determined based on these feature parameters. In some embodiments of this application, the fetal masking coefficient is determined by the following formula:
[0064]
[0065] in, For a category of fetal mask coefficients, This indicates the selected fetal characteristic parameters for this category. Indicates the first Fetal characteristic parameters of this category that were not selected. It is the base of the logarithm of natural numbers. The digital features of the fetal image for this category are input.
[0066] It should be noted that the selected fetal feature parameters for this category refer to the parameters of a specific region or feature of this category extracted and focused from the fetal image. The unselected fetal feature parameters for this category are parameters other than the selected fetal feature parameters. The fetal image digital features are numerical descriptions extracted from the fetal image, used to quantify and analyze the morphology, structure, and features of the fetus. The fetal mask coefficient is a coefficient used to represent the importance or weight of each pixel in the fetal image. It plays a role in weighing the contribution of pixels to fetal segmentation in the fetal image segmentation task.
[0067] In addition, the process of determining the fetal mask coefficient can also employ various machine learning or deep learning methods, such as Support Vector Machine (SVM). By training the model, patterns and rules related to fetal features in the fetal feature map can be learned. The fetal mask coefficient described in this application represents the probability or weight of each pixel belonging to the fetal feature.
[0068] In some embodiments, a selection operation is performed on the fetal feature images of the category based on the influence of fetal features. The fetal feature image with the highest feature influence in the category is selected as the selected feature image, and the selected feature image is used as the fetal template for the category. The fetal template is the template or model with the most prominent fetal image details in the fetal feature images of the category.
[0069] It should be noted that, in this application, the fetal feature influence refers to the degree of contribution or importance of different features to the final result during fetal image processing and analysis. It is used to measure the relative importance of different features in fetal image analysis in order to better understand and interpret the feature information of the image.
[0070] In some embodiments of this application, the influence of the fetal characteristics can be determined by the following formula:
[0071]
[0072] in, The degree of influence of fetal characteristics, This refers to the fetal shape variable in the fetal feature map. These are the fetal characteristic variables in the fetal characteristic map. This represents the mean value of the fetal shape variable in the fetal feature map. This represents the mean of fetal characteristic variables in the fetal characteristic map. Represents the square root. This represents the total number of variables in the fetal feature map, including fetal shape variables and fetal feature variables.
[0073] It should be noted that the aforementioned fetal shape variables may include head shape variables, trunk shape variables, limb shape variables, etc. In practice, fetal shape variables can be automatically extracted from fetal images through computer image processing and analysis algorithms, or they can be obtained through manual annotation and measurement by medical experts. In addition, fetal characteristic variables may include head circumference, abdominal circumference, spinal curvature angle, abdominal organ position, etc. In practice, fetal characteristic variables can be automatically extracted from fetal images through image processing and calculation algorithms or obtained through manual measurement by doctors, which will not be elaborated here.
[0074] It should be noted that the influence of fetal characteristics can also be determined by other methods, such as using feature selection algorithms or feature weighting evaluation methods. These methods can calculate the importance or weight of fetal characteristics based on indicators such as the correlation between the feature and the target variable, information gain, and analysis of variance.
[0075] In some embodiments of this application, a fetal segmentation mask for that category is determined based on the fetal mask coefficients and the fetal template. The fetal segmentation mask is the result of weighted merging of the fetal mask coefficients and the fetal template, which can accurately indicate the position and shape of the fetus in the image.
[0076] In some embodiments, the fetal image can be segmented using a fetal segmentation mask for each category to obtain feature regions or contours in the fetal image. Based on the location, shape, size, and other features of the feature regions, fetal feature identification information can be determined. For example, the fetal body shape characteristics can be described by calculating the area, perimeter, and center point coordinates of the feature regions. In addition, the pixel intensity, texture features, and shape features of the feature regions in the fetal image can be further analyzed to extract more fetal feature identification information, which will not be elaborated here.
[0077] In step S104, the fetal feature identification information is compared with a preset fetal information threshold to determine the fetal identification image difference, and then the fetal identification image accuracy is obtained. The fetal correction feedback amount is obtained from the fetal identification image accuracy.
[0078] A preset fetal information threshold is a pre-defined threshold for fetal information used to determine whether the identified image meets the expected threshold. This threshold can be set according to specific needs and actual circumstances, such as based on fetal shape, size, and position. In some embodiments, the fetal feature identification information is compared with the preset fetal information threshold. The accuracy of the fetal identification is assessed by calculating the difference between the two thresholds. This difference can be represented as the difference or error value of each pixel in the fetal identification image. Based on this difference, the fetal identification image accuracy rate can be calculated. Specifically, the difference between the fetal feature identification information and the preset fetal information threshold is calculated. When the difference is less than the preset fetal information threshold, the fetal identification image is considered real and has the same number of pixels. The accuracy rate is then obtained by calculating the proportion of identical pixels in the fetal identification image. Since the fetal identification image accuracy rate represents the degree of consistency between the fetal identification image and the preset fetal information, the closer it is to 100%, the more accurate the fetal identification image. In some embodiments, the fetal identification image accuracy rate can be determined by the following expression:
[0079] Fetal identification image accuracy = (1 - (difference / preset threshold)) * 100%
[0080] The difference is the difference or error value of each pixel in the fetal identification image, and the preset threshold is a pre-set threshold used to judge whether the identification image meets the expectations. It should be noted that the calculation method of the fetal identification image accuracy can be adjusted and improved according to specific circumstances to adapt to specific fetal identification tasks and algorithms.
[0081] In some embodiments, the fetal corrective feedback quantity is obtained from the accuracy of the fetal identification image, and the fetal corrective feedback quantity can be determined by the following expression:
[0082] Feedback amount = θ * TPR
[0083] Here, the feedback quantity represents the degree of calibration of the fetal feature map, θ is the adjustment factor, and TPR is the accuracy of the fetal identification map. According to this expression, when the accuracy of the fetal identification image is high, the feedback quantity will also be correspondingly high, so that the feature map can be effectively calibrated. The fetal correction feedback quantity can represent the quality, accuracy or other relevant indicators of the fetal image and is used to evaluate the reliability of the fetal image.
[0084] In step S105, the fetal feature map is dynamically calibrated based on the fetal correction feedback amount to obtain the fetal body shape feature map.
[0085] In some embodiments, the fetal feature map is dynamically calibrated based on the fetal correction feedback. For the fetal feature map that needs calibration, a high-speed processor and computer vision algorithm are used for matching and processing. By adjusting, rotating, cropping, or expanding the fetal feature map, and trying different combinations of transformation parameters, the fetal feature map is corrected to improve the success rate and quality of correction. By performing fine segmentation processing on the fetal feature map using different types of fetal segmentation masks, the calibration effect can be further improved, and the final fetal body shape feature map is obtained. The fetal body shape feature map can be the result of further processing or analysis of the fetal image to represent the fetal body shape features, such as fetal position features, fetal posture features, fetal size features, and fetal shape features, which are not specifically limited here.
[0086] In another aspect, in some embodiments, this application provides an image processing-based obstetric medical image recognition system, referencing... Figure 2 This figure is a schematic diagram of exemplary hardware and / or software of an image processing-based obstetric medical image recognition system according to some embodiments of this application. The image processing-based obstetric medical image recognition system 200 includes: a fetal image preprocessing module 201, a fetal feature map extraction module 202, a fetal feature identification information determination module 203, a fetal correction feedback amount determination module 204, and a fetal body shape feature map determination module 205, which are described below:
[0087] The fetal image preprocessing module 201 in this application is mainly used to preprocess the acquired images of the fetus in the pregnant woman's abdomen to obtain a preprocessed fetal image.
[0088] The fetal feature image extraction module 202 in this application is mainly used to decompose the preprocessed fetal image to obtain a preprocessed fetal image feature set, and extract different categories of fetal feature images from the preprocessed fetal image feature set;
[0089] The fetal feature identification information determination module 203 in this application is mainly used to extract fetal feature parameters from fetal feature images of each category, determine the fetal mask coefficient of the fetal feature image of that category based on the extracted fetal feature parameters, perform a selection operation on the fetal feature images of that category based on the fetal feature influence, select the fetal feature image with the highest feature influence in that category as the selected feature image, use the selected feature image as the fetal template of that category, determine the fetal segmentation mask of that category based on the fetal mask coefficient and the fetal template, the fetal segmentation mask is obtained by weighted merging of the fetal template and the mask coefficient, and then determine the fetal feature identification information based on the fetal segmentation mask;
[0090] The fetal correction feedback quantity determination module 204 in this application is mainly used to compare the fetal feature identification information with a preset fetal information threshold, determine the fetal identification image difference, and then obtain the fetal identification image accuracy. The fetal correction feedback quantity is obtained from the fetal identification image accuracy.
[0091] The fetal body shape feature map determination module 205 in this application is mainly used to dynamically calibrate the fetal feature map according to the fetal correction feedback amount, so as to obtain the fetal body shape feature map.
[0092] In addition, in some embodiments, this application also provides a computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described image processing-based obstetric medical image recognition method.
[0093] In some embodiments, reference Figure 3 The figure is a schematic diagram of the structure of a computer device applying the above-described image processing-based obstetric medical image recognition method according to some embodiments of this application. The image processing-based obstetric medical image recognition method in the above embodiments can... Figure 3 The computer device shown is used to implement this, and the computer device 300 includes at least one processor 301, a communication bus 302, a memory 303, and at least one communication interface 304.
[0094] The processor 301 may be a general-purpose central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more for controlling the execution of the image processing-based obstetric medical image recognition method in this application.
[0095] The communication bus 302 may include a path for transmitting information between the aforementioned components.
[0096] The memory 303 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 303 may exist independently and be connected to the processor 301 via the communication bus 302, or the memory 303 may be integrated with the processor 301.
[0097] The memory 303 stores program code for executing the scheme of this application, and its execution is controlled by the processor 301. The processor 301 executes the program code stored in the memory 303. The program code may include one or more software modules. In the above embodiments, fetal characteristic identification information can be implemented by the processor 301 and one or more software modules in the program code in the memory 303.
[0098] Communication interface 304 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.
[0099] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).
[0100] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.
[0101] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0102] This application also provides a computer-readable storage medium storing a computer program that, when run on a computer, enables the computer to execute the aforementioned image processing-based obstetric medical image recognition method.
[0103] In summary, the obstetric medical image recognition method and system based on image processing disclosed in this application first preprocesses the acquired images of the fetus in the pregnant woman's abdomen to obtain preprocessed fetal images; decomposes the preprocessed fetal images to obtain a preprocessed fetal image feature set, and extracts different categories of fetal feature images from the preprocessed fetal image feature set; extracts fetal feature parameters for each category of fetal feature images, determines the fetal mask coefficient of the category of fetal feature images based on the extracted fetal feature parameters, performs a selection operation on the category of fetal feature images based on the fetal feature influence, selects the fetal feature image with the highest feature influence in the category as the selected feature image, and uses the selected feature image as the fetal template for that category. Based on the fetal mask coefficients and the fetal template, a fetal segmentation mask for that category is determined. The fetal segmentation mask is obtained by weighted merging of the fetal template and the mask coefficients. Fetal feature identification information is then determined based on the fetal segmentation mask. The difference between the fetal feature identification information and a preset fetal information threshold is determined, thereby obtaining the fetal feature image accuracy. From the fetal feature image accuracy, a fetal correction feedback value is obtained. Based on the fetal correction feedback value, the fetal feature map is dynamically calibrated to obtain a fetal body shape feature map. This realizes an image processing-based obstetric medical image recognition method that can effectively improve the recognition accuracy of fetal images.
[0104] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.
[0105] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.
Claims
1. A method for obstetric medical image recognition based on image processing, characterized in that, Includes the following steps: The images of the fetus in the pregnant woman's abdomen were preprocessed to obtain preprocessed fetal images; The preprocessed fetal image is decomposed to obtain a preprocessed fetal image feature set, and different categories of fetal feature maps are extracted from the preprocessed fetal image feature set; Fetal feature parameters are extracted from fetal feature images of each category. Fetal mask coefficients are determined based on the extracted fetal feature parameters. Fetal feature images of each category are selected based on the influence of fetal features. The fetal feature image with the highest feature influence in each category is selected as the selected feature image and used as the fetal template for that category. Fetal segmentation mask is determined based on the fetal mask coefficients and the fetal template. The fetal segmentation mask is obtained by weighted merging of the fetal template and the mask coefficients. Fetal feature identification information is then determined based on the fetal segmentation mask. The fetal feature identification information is compared with a preset fetal information threshold to determine the fetal identification image difference, thereby obtaining the fetal identification image accuracy. The fetal correction feedback amount is obtained from the fetal identification image accuracy. Based on the fetal correction feedback, the fetal feature map is dynamically calibrated to obtain the fetal body shape feature map.
2. The method according to claim 1, characterized in that, The images of the fetus in the pregnant woman's womb are preprocessed to obtain preprocessed fetal images, specifically including: Determine the average grayscale value of the fetal image; Determine the standard deviation of the grayscale values in the fetal image; The fetal image is standardized based on its average grayscale value and the standard deviation of its grayscale values to obtain a preprocessed fetal image. The standardization of the fetal image is determined by the following formula: in, Standardize the grayscale values of the fetal image. The grayscale value of the fetal image. The average grayscale value of the fetal image. denoted as the standard deviation of the grayscale values in the fetal image.
3. The method according to claim 2, characterized in that, The average grayscale value of the fetal image is determined by the following formula: in, The average grayscale value of the fetal image. The number of grayscale values in the fetal image. Indicates the first Gray values of a fetal image.
4. The method according to claim 2, characterized in that, The standard deviation of the grayscale values of the fetal image is determined by the following formula: in, The standard deviation of the grayscale values in the fetal image. The average grayscale value of the fetal image. The number of grayscale values in the fetal image. Indicates the first Gray values of a fetal image.
5. The method according to claim 1, characterized in that, The decomposition of the preprocessed fetal image to obtain the preprocessed fetal image feature set specifically includes: The gray values in the preprocessed fetal image data are decomposed into fetal feature components with low temporal frequency and high spatial frequency according to their spatial location and time sequence in the image. For each fetal characteristic component, a predetermined number of fetal characteristic functions are obtained; Based on the fetal feature function obtained from each fetal feature component, feature information is extracted from the preprocessed fetal image to obtain the fetal feature mode; The fetal feature modalities are combined to obtain the features of the original fetal image at various spatial and temporal scales, thereby obtaining the preprocessed fetal image feature set.
6. The method according to claim 5, characterized in that, The fetal feature components are the features of the preprocessed fetal images at different temporal and spatial resolutions.
7. The method according to claim 1, characterized in that, Based on the comparison between the fetal feature identification information and a preset fetal information threshold, the difference between the fetal identification images is determined, and then the accuracy of the fetal identification images is obtained, specifically including: The difference between the fetal feature identification information and the preset fetal information threshold is calculated to obtain the fetal identification image difference. Based on the difference between the fetal identification images, when the difference between the fetal identification images is less than 1, the fetal identification image is considered to be real and has the same number of pixels. Then, by calculating the proportion of the same pixels in the fetal identification images, the realism rate of the fetal identification images is obtained.
8. An obstetric medical image recognition system based on image processing, characterized in that, include: The fetal image preprocessing module is used to preprocess the acquired images of the fetus in the pregnant woman's abdomen to obtain preprocessed fetal images. The fetal feature image extraction module decomposes the preprocessed fetal image to obtain a preprocessed fetal image feature set, and extracts different categories of fetal feature images from the preprocessed fetal image feature set; The fetal feature identification information determination module is used to extract fetal feature parameters from fetal feature images of each category, determine the fetal mask coefficient of the fetal feature image of that category based on the extracted fetal feature parameters, perform a selection operation on the fetal feature images of that category based on the fetal feature influence, select the fetal feature image with the highest feature influence in that category as the selected feature image, use the selected feature image as the fetal template of that category, determine the fetal segmentation mask of that category based on the fetal mask coefficient and the fetal template, the fetal segmentation mask is obtained by weighted merging of the fetal template and the mask coefficient, and then determine the fetal feature identification information based on the fetal segmentation mask; The fetal corrective feedback quantity determination module is used to compare the fetal feature identification information with a preset fetal information threshold, determine the fetal identification image difference, and then obtain the fetal identification image accuracy. The fetal corrective feedback quantity is obtained from the fetal identification image accuracy. The fetal body shape feature map determination module is used to dynamically calibrate the fetal feature map based on the fetal correction feedback amount, thereby obtaining the fetal body shape feature map.
9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory storing code, and the processor being configured to retrieve the code and execute the image processing-based obstetric medical image recognition method as described in any one of claims 1 to 7.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the obstetric medical image recognition method based on image processing as described in any one of claims 1 to 7.