An image data processing method and apparatus

Abstract

This invention discloses an image data processing method and apparatus. The method involves acquiring an initial lesion image and at least one preoperative lesion image; determining a first mask image corresponding to the initial lesion image and a second mask image corresponding to the at least one preoperative lesion image; determining a first subtraction image and a second subtraction image based on the initial lesion image and the at least one preoperative lesion image; determining a first topological feature and a second topological feature based on the first and second subtraction images; determining target features corresponding to the lesion site based on the first subtraction image, the second subtraction image, the first topological feature, and the second topological feature; and inputting the target features into a pre-trained classification model to determine the judgment result corresponding to the lesion site. This method solves the technical problems of low efficiency and low accuracy when evaluating the disease information of patients based on image data, and improves the efficiency and accuracy of disease assessment.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an image data processing method and apparatus. Background Technology

[0002] In medical big data, images contain rich semantic information and play a crucial role in assessing the progression of a patient's condition. Image data is unstructured, characterized by its complexity and redundancy; therefore, extracting useful structured data from image data to assist physicians in clinical decision-making is of paramount importance.

[0003] Currently, there are two methods to assess the progression of a patient's condition based on imaging data. The first method involves physicians analyzing the patient's imaging data based on their experience to make clinical decisions. The second method uses machine learning algorithms to predict disease progression. Machine learning methods standardize unstructured imaging data from the disease information, then use the standardized data for training and classification to ultimately obtain a predicted disease progression result.

[0004] However, the first method has the problem of manual processing, which is time-consuming, labor-intensive, and relatively inefficient. The second method has the problem of using machine learning algorithms to make predictions based on single image data. The information extracted from the image data is not directly effective, and there are limitations in accurately assessing the patient's condition based on image data. This results in technical problems such as low prediction efficiency and low accuracy. Summary of the Invention

[0005] This invention provides an image data processing method and apparatus that enables effective evaluation of the patient's condition information based on image data from different preoperative stages, thereby improving the efficiency and accuracy of condition assessment.

[0006] In a first aspect, the present invention provides an image data processing method, the method comprising:

[0007] Obtain initial lesion images and at least one preoperative lesion image; wherein the initial lesion image and at least one preoperative lesion image include the same lesion site;

[0008] A first mask image corresponding to the initial lesion image and a second mask image corresponding to at least one preoperative lesion image are determined respectively; wherein the number of second mask images is consistent with the number of at least one preoperative lesion image;

[0009] Based on the initial lesion image and at least one preoperative lesion image, a first subtraction image and a second subtraction image are determined; wherein the number of second subtraction images is consistent with the number of at least one preoperative lesion image;

[0010] Based on the first subtraction image and the second subtraction image, a first topological feature and a second topological feature are determined; wherein, the first topological feature corresponds to the first subtraction image, and the second topological feature corresponds to the second subtraction image;

[0011] Based on the first subtraction image, the second subtraction image, the first topological feature, and the second topological feature, the target features corresponding to the lesion site are determined;

[0012] The target features are input into a pre-trained classification model to determine the judgment result corresponding to the lesion location.

[0013] In a second aspect, the present invention provides an image data processing apparatus, the apparatus comprising:

[0014] The image acquisition module is used to acquire an initial lesion image and at least one preoperative lesion image; wherein the initial lesion image and at least one preoperative lesion image include the same lesion site;

[0015] A mask image determination module is used to determine a first mask image corresponding to the initial lesion image and a second mask image corresponding to at least one preoperative lesion image, wherein the number of second mask images is consistent with the number of at least one preoperative lesion image;

[0016] The subtraction image determination module is used to determine a first subtraction image and a second subtraction image based on an initial lesion image and at least one preoperative lesion image; wherein the number of second subtraction images is consistent with the number of at least one preoperative lesion image;

[0017] The topological feature determination module is used to determine a first topological feature and a second topological feature based on a first subtraction image and a second subtraction image; wherein the first topological feature corresponds to the first subtraction image and the second topological feature corresponds to the second subtraction image;

[0018] The target feature determination module is used to determine the target features corresponding to the lesion site based on the first subtraction image, the second subtraction image, the first topological feature, and the second topological feature.

[0019] The judgment result determination module is used to input the target features into the pre-trained classification model and determine the judgment result corresponding to the lesion site.

[0020] Thirdly, the present invention provides a data processing electronic device, comprising:

[0021] At least one processor; and

[0022] A memory that is communicatively connected to at least one processor; wherein,

[0023] The memory stores a computer program that can be executed by at least one processor, such that the at least one processor is able to perform the image data processing method of any embodiment of the present invention.

[0024] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the image data processing method of any embodiment of the present invention.

[0025] Fifthly, the present invention provides a computer program product, which includes a computer program that, when executed by a processor, implements the image data processing method of any embodiment of the present invention.

[0026] The technical solution provided by this invention involves acquiring an initial lesion image and at least one preoperative lesion image, determining a first mask image corresponding to the initial lesion image and a second mask image corresponding to the at least one preoperative lesion image, respectively. Subsequently, based on the initial lesion image and at least one preoperative lesion image, a first subtraction image and a second subtraction image are determined. Further, based on the first and second subtraction images, a first topological feature and a second topological feature are determined. Then, based on the first and second subtraction images, the first and second topological features, target features corresponding to the lesion site are determined. Finally, the target features are input into a pre-trained classification model to determine the judgment result corresponding to the lesion site. This invention solves the technical problems of low efficiency and low accuracy in evaluating the patient's disease information based on image data, realizing effective evaluation of the patient's disease information based on image data from different preoperative stages, thereby improving the efficiency and accuracy of disease evaluation.

[0027] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a flowchart of an image data processing method provided in Embodiment 1 of the present invention;

[0030] Figure 2This is a flowchart of an image data processing method provided in Embodiment 2 of the present invention;

[0031] Figure 3 This is a flowchart of an image data processing method provided in Embodiment 3 of the present invention;

[0032] Figure 4 This is a schematic diagram of an image data processing device provided in Embodiment 4 of the present invention;

[0033] Figure 5 This is a schematic diagram of the structure of an electronic device provided in Embodiment 5 of the present invention. Detailed Implementation

[0034] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0035] It should be noted that the terms "first preset condition," "second preset condition," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0036] Before introducing this technical solution, the application scenario can be described first. The technical solution provided by the embodiments of this invention can be applied to scenarios where the development of any disease is predicted. If a patient has a disease and subsequently undergoes treatment over a period of time, the effectiveness of the treatment can be evaluated. Based on the evaluation results, the subsequent treatment method is determined, which may include at least one surgical and / or non-surgical treatment.

[0037] Example 1

[0038] Figure 1This is a flowchart of an image data processing method provided in Embodiment 1 of the present invention. This embodiment is applicable to situations where image data from different preoperative stages is used to effectively assess the patient's medical condition. The method can be executed by an image data processing device, which can be implemented in hardware and / or software. This device can be configured on a computer device, such as a laptop, desktop computer, or tablet. Figure 1 As shown, the method includes:

[0039] S110. Obtain initial lesion images and at least one preoperative lesion image.

[0040] The initial lesion image refers to the image taken during the patient's first visit, or an image taken at a time interval from the preoperative lesion image, but this time interval cannot be too long. The lesion image includes the size and shape of the initial lesion. For example, the initial lesion image can be a magnetic resonance imaging (MRI) image. The preoperative lesion image is an image taken after a period of treatment. There can be one or multiple preoperative lesion images; the specific number is not limited here.

[0041] Specifically, initial lesion images and preoperative lesion images can be uploaded to the data processing platform via mobile terminal devices. When it is necessary to predict the symptom development of patients based on their imaging data, the data uploaded by the users can be retrieved directly.

[0042] Based on the above embodiments, after acquiring the initial lesion image and at least one preoperative lesion image, the method further includes: using the initial lesion image as a reference image, and performing registration processing on at least one preoperative lesion image based on the reference image to obtain at least one registered preoperative lesion image.

[0043] The reference image can be understood as using the initial lesion image as a reference standard for registration with the preoperative lesion image. Registration involves spatially aligning the two lesion images to ensure the organ location is aligned between the initial and preoperative lesion images. Because patients' postures and equipment vary during each lesion image capture, the lesion location in the images may deviate slightly; therefore, registration of the two lesion images is necessary.

[0044] In this embodiment, the initial lesion image can be used as a reference standard to perform registration processing on at least one preoperative lesion image. The specific processing process is as follows: using the three-dimensional matrix corresponding to the initial lesion image as a reference space, the preoperative lesion image is registered to the reference space to ensure that the organ parts of the initial lesion image and the preoperative lesion image are aligned.

[0045] S120, determine a first mask image of the initial lesion image and a second mask image of at least one preoperative lesion image.

[0046] The mask image can be understood as a black-and-white image corresponding to the lesion site. For example, the pixel value corresponding to the lesion site is 1, and the pixel value corresponding to the non-lesion site is 0. The pixel color corresponding to the pixel value of 1 on the mask image is black, and the pixel color corresponding to the pixel value of 0 on the mask image is white. The mask image contains the specific location information of the lesion for the patient. The first mask image is a mask image for marking the lesion in the initial lesion image. The second mask image is a mask image for marking the lesion in the preoperative lesion image.

[0047] Specifically, since the lesion image is actually a matrix composed of multiple pixel values, we can first determine the matrix corresponding to the initial lesion image and the preoperative lesion image, perform masking processing on the matrix corresponding to the initial lesion image to obtain the first mask image, and perform masking processing on the matrix corresponding to the preoperative lesion image to obtain the second mask image.

[0048] Based on the above embodiments, the method for determining the first mask image and the second mask image may be: determining a first matrix of the initial lesion image and a second matrix corresponding to at least one preoperative lesion image respectively; processing the first matrix and the second matrix based on the pre-trained segmentation framework to obtain the first mask image and the second mask image.

[0049] The first matrix is ​​the 3D matrix corresponding to the initial lesion image, and the second matrix is ​​the 3D matrix corresponding to the preoperative lesion image. Each element in the matrix is ​​the pixel value of each pixel. The segmentation framework is pre-trained. Optionally, the segmentation framework can be the 3DUNet automated segmentation framework. During training, it can be trained based on a large amount of lesion image data. The lesion image data is input into the 3DUNet automated segmentation framework, which can read the 3D matrix corresponding to the lesion image data. Furthermore, it multiplies each layer of matrix data with a pre-set region of interest mask matrix to obtain the mask image of each layer, thereby obtaining the mask image corresponding to the lesion image and realizing the segmentation processing of the 3D matrix.

[0050] In practical applications, inputting the first matrix of the initial lesion image into the segmentation frame yields the first mask image corresponding to the initial lesion image, and inputting the second matrix of the preoperative lesion image into the segmentation frame yields the second mask image corresponding to the preoperative lesion image.

[0051] For example, firstly, the first matrix X1 of the initial lesion image and the second matrix X2 of the preoperative lesion image are determined. Then, X1 and X2 are input into the pre-trained 3DUNet automated segmentation framework to obtain the first mask image M1 of the region of interest of lesion X1 and the first mask image M2 of the region of interest of lesion X2.

[0052] S130. Based on the initial lesion image and at least one preoperative lesion image, determine the first subtraction image and the second subtraction image.

[0053] The number of the second subtraction images is consistent with the number of at least one preoperative lesion image. The subtraction image is determined by the difference in pixel values ​​at corresponding locations between two lesion images.

[0054] In this embodiment, two subtraction images can be constructed using a first matrix of initial lesion images and a second matrix of at least one preoperative lesion image. These constructed subtraction images are a first subtraction image S1 and a second subtraction image S2. S1 is the subtraction image determined by subtracting the second matrix from the first matrix, and can be called a positive subtraction image; S2 is the subtraction image determined by subtracting the first matrix from the second matrix, and can be called a negative subtraction image. S1 and S2 can be represented as:

[0055] S1=f(X1-X2) (1)

[0056] S2=f(X2-X1) (2)

[0057] In the formula, f is the activation function, X1 is the first matrix of the initial lesion image, and X2 is the second matrix of the preoperative lesion image.

[0058] It should be noted that the characteristic of the activation function is that for each element in the matrix, if the value of the element is greater than zero, the value of the element is retained; if the value of the element is less than zero, the element is set to zero.

[0059] S140. Based on the first subtraction image and the second subtraction image, determine the first topological feature and the second topological feature.

[0060] Wherein, the first topological feature is a feature vector obtained by feature processing the first subtraction image, and the second topological feature is a feature vector obtained by feature processing the second subtraction image.

[0061] Specifically, the feature processing performed on the first subtraction image can include binarization segmentation, morphological erosion, and dilation. Binarization segmentation can be understood as pre-setting pixel thresholds. For the subtraction image matrix, if a pixel value is greater than the threshold, its value is set to 1; if it is less than the threshold, its value is set to 0. Erosion can be understood as eliminating isolated pixel values ​​in the segmented image. Its purpose is to eliminate boundary points in the segmented image, causing the boundaries to shrink inward, thereby separating two lesion edges with thin connections, and also removing burrs, small protrusions, etc. Dilation can be understood as enlarging the lesion edges in the segmented image to obtain clearer lesion edges. After feature processing of the first subtraction image, the first topological feature corresponding to the first subtraction image is obtained. The second topological feature can be obtained using the same feature processing method.

[0062] Based on the above embodiments, determining the first topological feature and the second topological feature based on the first subtraction image and the second subtraction image can be achieved by: performing binarization segmentation on the first subtraction image and the second subtraction image based on a pre-set pixel threshold to obtain a first segmented image and a second segmented image; performing erosion and dilation processing on the first segmented image and the second segmented image respectively to obtain the number of first genus cells in the first segmented image and the number of second genus cells in the second segmented image; and determining the first topological feature and the second topological feature based on the number of first genus cells and the number of second genus cells.

[0063] Here, the genus number is the number of pixels in the segmented image that have been zeroed out. For example, if the two-dimensional matrix corresponding to the first layer of the first segmented image is: The number of genus cells in the first layer of the first segmented image is 4. By calculating the number of genus cells in each layer of the first segmented image and summing them, the number of genus cells in the first segmented image can be obtained.

[0064] In this embodiment, the first subtraction image is first binarized and segmented based on a pixel threshold to obtain a first segmented image; the second subtraction image is then binarized and segmented based on the same pixel threshold to obtain a second segmented image. Further, the first segmented image is subjected to erosion and dilation processing to obtain clear lesion edges. The first segmented data after erosion and dilation processing is statistically analyzed to obtain the number of first genus cells corresponding to the first segmented image. The value representing the number of first genus cells is then the first topological feature. The method for determining the second topological feature is the same as that for determining the first topological feature, and will not be described again here.

[0065] S150. Based on the first subtraction image, the second subtraction image, the first topological feature, and the second topological feature, determine the target feature corresponding to the lesion site.

[0066] The target feature is a feature vector selected from multiple feature vectors for predicting disease progression.

[0067] In this embodiment, radiomics feature extraction methods are used to extract subtraction features from the first and second subtraction images. The extracted subtraction features may include shape features, grayscale features, texture features, and filtering features. In addition, there are the first and second topological features determined in the above steps, resulting in a large number of feature vectors. However, some of these feature vectors have relatively low correlation with the prediction of disease development at the lesion site. If disease development prediction at the lesion site is based on a large number of useless feature vectors, it will not only waste server resources but also result in unsatisfactory prediction accuracy. Therefore, feature vectors can be filtered to obtain features with relatively high correlation with the prediction of disease development at the lesion site as target features.

[0068] S160. Input the target features into the pre-trained classification model to determine the judgment result corresponding to the lesion site.

[0069] The classification model can be pre-trained and used to determine the corresponding judgment result based on the target features and the lesion site.

[0070] The determination result indicates whether surgery is necessary. For example, the result can be "Y" or "N". If the result is "Y", it means that after a period of treatment, the patient's condition has not been effectively controlled; the lesions have not shrunk or enlarged after treatment. In this case, surgery can be used to treat the condition. If the result is "N", it means that after a period of treatment, the patient's condition has been effectively controlled; the lesions have shrunk, and treatment can continue without surgery.

[0071] Optionally, the target features can be input into a pre-trained classification model corresponding to the lesion site to obtain a judgment result.

[0072] It should be noted that the lesion site corresponds to any lesion site of the target object.

[0073] The lesion site refers to the area where a lesion is present. The target audience is patients who need to predict the progression of their condition. The lesion site refers to the body part where a lesion is present, such as the breast, liver, or head.

[0074] In this embodiment, multiple classification models corresponding to lesion sites can be pre-trained. For example, classification models corresponding to breast lesions and liver lesions can be pre-trained. In practical applications, a classification model corresponding to the lesion site is determined based on the corresponding lesion site. Furthermore, the target features are input into the classification model corresponding to the lesion site to obtain a determination result corresponding to the lesion site.

[0075] For example, the target features corresponding to a user seeking medical treatment for breast tumors can be input into a pre-trained classification model corresponding to the location of the breast lesion to obtain a judgment result corresponding to the location of the breast lesion. If the judgment result is "yes," the condition can be treated surgically; if the judgment result is "no," the condition does not require surgical treatment.

[0076] The above technical solution involves acquiring an initial lesion image and at least one preoperative lesion image, determining a first mask image corresponding to the initial lesion image and a second mask image corresponding to the at least one preoperative lesion image, respectively. Then, based on the initial lesion image and at least one preoperative lesion image, a first subtraction image and a second subtraction image are determined. Further, based on the first and second subtraction images, a first topological feature and a second topological feature are determined. And based on the first and second subtraction images, the first and second topological features, target features corresponding to the lesion site are determined. Finally, the target features are input into a pre-trained classification model to determine the judgment result corresponding to the lesion site. This invention solves the technical problems of low efficiency and low accuracy in evaluating the patient's disease information based on image data, realizing effective evaluation of the patient's disease information based on image data from different preoperative stages, thereby improving the efficiency and accuracy of disease evaluation.

[0077] Example 2

[0078] Figure 2 This is a flowchart of a data processing method provided in Embodiment 2 of the present invention. Based on the above embodiments, this embodiment further refines the contents corresponding to S130 and S150 of the aforementioned embodiments. This embodiment can be combined with various optional solutions in one or more of the above embodiments. For example... Figure 2 As shown, the method includes:

[0079] S210. Obtain initial lesion images and at least one preoperative lesion image.

[0080] S220, determine a first mask image of the initial lesion image and a second mask image of at least one preoperative lesion image.

[0081] S230. Determine at least one set of images to be processed.

[0082] At least one set of images to be processed includes lesion images with adjacent timestamps.

[0083] For example, the images to be processed may include: image 1, image 2, image 3, and image 4. Image 1 has a capture timestamp of January 22nd, image 2 has a capture timestamp of March 25th, image 3 has a capture timestamp of June 10th, and image 4 has a capture timestamp of August 30th. A set of images to be processed may include image 1 and image 2, image 2 and image 3, or image 3 and image 4. A set of images to be processed represents a set where there is no lesion image captured between the timestamp differences.

[0084] S240. For each group of images to be processed, the adjacent lesion images are processed based on the activation function to obtain the subtraction image corresponding to each lesion image.

[0085] In this embodiment, not only can disease progression be predicted within a timeframe determined by the timestamps of the initial lesion image and a preoperative lesion image, but lesion progression can also be predicted for lesion images at any treatment stage based on two adjacent lesion images. Therefore, for each group of images to be processed, an activation function can be used to process two lesion images within the group to obtain subtraction images corresponding to these two lesion images respectively.

[0086] For example, based on S230, if the determined set of images to be processed includes image 3 and image 4, then the matrix corresponding to image 3 is X3, the matrix corresponding to image 4 is X4, the subtraction image corresponding to image 3 is S3 = f(X3-X4), and the subtraction image corresponding to image 4 is S4 = f(X4-X3), where f is the activation function.

[0087] S250. Based on the first subtraction image and the second subtraction image, determine the first topological feature and the second topological feature.

[0088] S260. Based on the first subtraction image and the first mask image, determine the first subtraction feature; and based on the second subtraction image and the corresponding second mask image, determine the second subtraction feature, so as to determine the target feature based on the first subtraction feature, the second subtraction feature, the first topological feature and the second topological feature.

[0089] In this embodiment, since the first mask image includes information for marking the initial lesion image, and the second mask image includes information for marking the preoperative lesion image, the feature vector of the first subtraction image can be extracted from the first subtraction image based on the first mask image as the first subtraction feature. Simultaneously, the feature vector of the second subtraction image can be extracted from the second subtraction image based on the second mask image as the second subtraction feature. Both the first and second subtraction features can include features such as shape, grayscale, texture, and filtering; therefore, the final number of determined first and second subtraction features is relatively large. Subsequently, the large number of first, second, first, and second topological features can be filtered to obtain a limited number of features with a relatively high correlation to the disease development prediction of the lesion site as target features.

[0090] Optionally, the first subtraction feature, the second subtraction feature, the first topological feature, and the second topological feature are processed based on the feature filtering module to obtain the target feature.

[0091] The feature selection module can be pre-trained and used to determine target features based on the first subtraction feature, the second subtraction feature, the first topological feature, and the second topological feature. For example, the feature selection module can include strategies such as T-test, Pearson correlation, and Lasso regression.

[0092] In practical applications, multiple feature vectors determined by the first subtraction feature, the second subtraction feature, the first topological feature, and the second topological feature are input into the feature filtering module. The feature filtering module can filter the feature vectors to obtain the features with a relatively high correlation with the disease development prediction of the lesion site as the target features.

[0093] S270. Input the target features into the pre-trained classification model to determine the judgment result corresponding to the lesion site.

[0094] The above technical solution involves acquiring an initial lesion image and at least one preoperative lesion image, determining a first mask image corresponding to the initial lesion image and a second mask image corresponding to at least one preoperative lesion image, then determining at least one set of images to be processed. For each set of images to be processed, adjacent lesion images are processed based on an activation function to obtain a subtraction image corresponding to each lesion image. Further, based on the first and second subtraction images, a first and a second topological feature are determined, and based on the first and first mask images, a first subtraction feature is determined; and based on the second subtraction image and the corresponding second mask image, a second subtraction feature is determined. Based on the first, second, first, and second topological features, a target feature is determined. Finally, the target feature is input into a pre-trained classification model to determine the judgment result corresponding to the lesion location. This invention can accurately predict the symptom development of patients by using a classification model based on the target features corresponding to two adjacent lesion images, regardless of the treatment stage, thus improving the accuracy and efficiency of symptom development prediction.

[0095] Example 3

[0096] In this embodiment of the invention, an image data processing method is described using a specific implementation method. Figure 3 This is a schematic diagram of an image data processing method provided in Embodiment 3 of the present invention. The method includes the following steps:

[0097] 1. Using the initial lesion image as a reference space, register the preoperative breast cancer lesion image to the baseline space to ensure that the organ location of the initial lesion image and the preoperative lesion image are aligned.

[0098] 2. Using the 3DUNet automated segmentation framework, the regions of interest (ROIs) of breast cancer lesions on the registered initial lesion image and the preoperative lesion image were detected using the first mask image M1 and the second mask image M2.

[0099] 3. Construct two subtraction images S1 and S2 using images from two different periods, satisfying:

[0100] S1=f(X1-X2) (1)

[0101] S2=f(X2-X1) (2)

[0102] Where S1 represents the first subtraction image, S2 represents the second subtraction image, and f represents the ReLU activation function.

[0103] 4. Based on the first mask image, extract the feature vector of the first subtraction image on the first subtraction image as the first subtraction feature. Based on the second mask image, extract the feature vector of the second subtraction image on the second subtraction image as the second subtraction feature. Use radiomics feature extraction methods to extract subtraction features, specifically including shape, grayscale, texture, and filtering features.

[0104] 5. Perform binarization segmentation and morphological erosion and dilation processing on the first and second subtraction images respectively, and count the number of genus cells in the processed morphological binary images, which are the topological features of the subtraction images.

[0105] 6. Use the feature filtering module to filter the extracted subtraction features and topological features to obtain the target features. The feature filtering module includes strategies such as T-test, Pearson correlation, and Lasso regression.

[0106] 7. Input the target features into the pre-trained classification model to determine the judgment result corresponding to the lesion location. The classification model can be an SVM judgment model.

[0107] The technical solution provided in this invention firstly aligns the locations of the initial lesion image and the preoperative lesion image through nonlinear registration, thereby obtaining a subtraction image. Then, an automatic segmentation algorithm is used to segment the regions of interest in the images at different stages. Based on the segmentation results of the two processed images and the subtraction image, subtraction feature extraction and topological feature extraction are performed, and effective feature selection is conducted using methods such as Lasso regression. Finally, a classification model such as SVM is used to construct a model for predicting the development of breast cancer symptoms. This classification model provides accurate predictions of symptom development for patients, improving the accuracy and efficiency of symptom development prediction.

[0108] Example 4

[0109] Figure 4 This is a schematic diagram of an image data processing device according to Embodiment 3 of the present invention. The device can execute the image data processing method provided in this embodiment of the invention. The device includes: an image acquisition module 410, a mask image determination module 420, a subtraction image determination module 430, a topological feature determination module 440, a target feature determination module 450, and a determination result determination module 460.

[0110] The image acquisition module 410 is used to acquire an initial lesion image and at least one preoperative lesion image; wherein the initial lesion image and at least one preoperative lesion image include the same lesion site;

[0111] The mask image determination module 420 is used to determine a first mask image corresponding to the initial lesion image and a second mask image corresponding to at least one preoperative lesion image, wherein the number of second mask images is consistent with the number of at least one preoperative lesion image;

[0112] The subtraction image determination module 430 is used to determine a first subtraction image and a second subtraction image based on an initial lesion image and at least one preoperative lesion image; wherein the number of second subtraction images is consistent with the number of at least one preoperative lesion image;

[0113] The topological feature determination module 440 is used to determine a first topological feature and a second topological feature based on a first subtraction image and a second subtraction image; wherein the first topological feature corresponds to the first subtraction image and the second topological feature corresponds to the second subtraction image;

[0114] The target feature determination module 450 is used to determine the target features corresponding to the lesion site based on the first subtraction image, the second subtraction image, the first topological feature, and the second topological feature.

[0115] The determination result module 460 is used to input the target features into the pre-trained classification model to determine the determination result corresponding to the lesion site.

[0116] Based on the above technical solutions, the image acquisition module 410 is also used to use the initial lesion image as a reference image, and to perform registration processing on at least one preoperative lesion image based on the reference image to obtain at least one registered preoperative lesion image.

[0117] Based on the above technical solutions, the mask image determination module 420 includes: an image matrix determination unit and a mask image determination unit.

[0118] The image matrix determination unit is used to determine a first matrix of initial lesion images and a second matrix corresponding to at least one preoperative lesion image.

[0119] The mask image determination unit is used to process the first matrix and the second matrix based on the pre-trained segmentation framework to obtain the first mask image and the second mask image.

[0120] Based on the above technical solutions, the subtraction image determination module 430 includes: a to-be-processed image determination unit and a subtraction image acquisition unit.

[0121] The image to be processed determination unit is used to determine at least one set of images to be processed; wherein, the at least one set of images to be processed includes lesion images with adjacent capture timestamps;

[0122] The subtraction image acquisition unit is used to process adjacent lesion images based on an activation function for each group of images to be processed, and obtain subtraction images corresponding to each lesion image.

[0123] Based on the above technical solutions, the topology feature determination module 440 is further configured to determine a first subtraction feature based on the first subtraction image and the first mask image; and to determine a second subtraction feature based on the second subtraction image and the corresponding second mask image, so as to determine the target feature based on the first subtraction feature, the second subtraction feature, the first topology feature and the second topology feature.

[0124] Based on the above technical solutions, the target feature determination module 450 includes: a binarization segmentation unit, a genus number determination unit, and a topological feature determination module.

[0125] The binarization segmentation unit is used to perform binarization segmentation on the first subtraction image and the second subtraction image based on a preset pixel threshold, so as to obtain the first segmented image and the second segmented image.

[0126] The genus number determination unit is used to perform erosion and dilation processing on the first segmented image and the second segmented image respectively, so as to obtain the first genus number of the binary image in the first segmented image and the second genus number of the binary image in the second segmented image.

[0127] The topology feature determination module is used to determine the first topology feature and the second topology feature based on the first genus number and the second genus number.

[0128] Based on the above technical solutions, the target feature determination module 450 is also used to: process the first subtraction feature, the second subtraction feature, the first topological feature and the second topological feature based on the feature filtering module to obtain the target feature.

[0129] Based on the above technical solutions, the determination result module 460 is also used to input the target features into a pre-trained classification model corresponding to the lesion site to obtain the determination result; wherein, the determination result is used to characterize whether surgery should be performed.

[0130] Based on the above technical solutions, the lesion site corresponds to any lesion site of the target object.

[0131] The technical solution provided by this invention involves acquiring an initial lesion image and at least one preoperative lesion image, determining a first mask image corresponding to the initial lesion image and a second mask image corresponding to the at least one preoperative lesion image, respectively. Subsequently, based on the initial lesion image and at least one preoperative lesion image, a first subtraction image and a second subtraction image are determined. Further, based on the first and second subtraction images, a first topological feature and a second topological feature are determined. Then, based on the first and second subtraction images, the first and second topological features, target features corresponding to the lesion site are determined. Finally, the target features are input into a pre-trained classification model to determine the judgment result corresponding to the lesion site. This invention solves the technical problems of low efficiency and low accuracy in evaluating the patient's disease information based on image data, realizing effective evaluation of the patient's disease information based on image data from different preoperative stages, thereby improving the efficiency and accuracy of disease evaluation.

[0132] The data processing apparatus provided in this disclosure can execute the video determination method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects of executing the method.

[0133] It is worth noting that the various units and modules included in the above-mentioned device are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be realized; in addition, the specific names of each functional unit are only for easy differentiation and are not used to limit the protection scope of the embodiments of this disclosure.

[0134] Example 5

[0135] Figure 5 This is a schematic diagram of an electronic device provided in Embodiment 5 of the present invention. The electronic device 10 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0136] like Figure 5As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0137] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0138] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as road surface recognition methods.

[0139] In some embodiments, the road surface recognition method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the road surface recognition method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the road surface recognition method by any other suitable means (e.g., by means of firmware).

[0140] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0141] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0142] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0143] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0144] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0145] A computing system may include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is established by computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within the cloud computing service system, addressing the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability. It should be understood that various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired results of the technical solution of this invention are achieved, and this is not limited herein. The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. An image data processing method, characterized in that, include: Obtain initial lesion images and at least one preoperative lesion image; wherein the initial lesion image and at least one preoperative lesion image include the same lesion site; A first mask image of the initial lesion image and a second mask image of at least one preoperative lesion image are determined respectively; wherein the number of second mask images is consistent with the number of at least one preoperative lesion image; Based on the initial lesion image and at least one preoperative lesion image, a first subtraction image and a second subtraction image are determined; wherein, the first subtraction image is a positive subtraction image determined by subtracting the second matrix corresponding to the preoperative lesion image from the first matrix corresponding to the initial lesion image, and the second subtraction image is a negative subtraction image determined by subtracting the first matrix from the second matrix, and the number of second subtraction images is consistent with the number of at least one preoperative lesion image; Based on the first subtraction image and the second subtraction image, a first topological feature and a second topological feature are determined; wherein, the first topological feature corresponds to the first subtraction image, and the second topological feature corresponds to the second subtraction image; Based on the first subtraction image, the second subtraction image, the first topological feature, and the second topological feature, the target features corresponding to the lesion site are determined; The target features are input into a pre-trained classification model to determine the judgment result corresponding to the lesion location; The determination of the first topological feature and the second topological feature based on the first subtraction image and the second subtraction image includes: The first subtraction image and the second subtraction image are binarized based on a pre-set pixel threshold to obtain the first segmented image and the second segmented image. Erosion and dilation are performed on the first segmented image and the second segmented image respectively to obtain the number of first-genus binary images in the first segmented image and the number of second-genus binary images in the second segmented image; The first topological feature and the second topological feature are determined based on the number of the first genus and the number of the second genus.

2. The method according to claim 1, characterized in that, After obtaining the initial lesion image and at least one preoperative lesion image, the procedure also includes: The initial lesion image is used as the reference image, and at least one preoperative lesion image is registered based on the reference image to obtain at least one registered preoperative lesion image.

3. The method according to claim 1, characterized in that, Determine a first mask image relative to the initial lesion image and a second mask image of at least one preoperative lesion image, including: A first matrix is ​​determined to identify the initial lesion image, and a second matrix is ​​determined to correspond to at least one preoperative lesion image. The first and second matrices are processed based on the pre-trained segmentation framework to obtain the first mask image and the second mask image.

4. The method according to claim 1, characterized in that, Based on the initial lesion image and at least one preoperative lesion image, determine the first subtraction image and the second subtraction image, including: Identify at least one set of images to be processed; wherein, at least one set of images to be processed includes lesion images with adjacent capture timestamps; For each group of images to be processed, adjacent lesion images are processed based on the activation function to obtain a subtraction image corresponding to each lesion image.

5. The method according to claim 1, characterized in that, The target features are input into a pre-trained classification model to determine the judgment result corresponding to the lesion location, including: The target features are input into a pre-trained classification model corresponding to the lesion site to obtain the judgment result; The determination result is used to indicate whether surgery should be performed.

6. The method according to any one of claims 1-5, characterized in that, The lesion site corresponds to any lesion site in the target object.

7. An image data processing apparatus, characterized in that, include: The image acquisition module is used to acquire an initial lesion image and at least one preoperative lesion image; wherein the initial lesion image and at least one preoperative lesion image include the same lesion site; The mask image determination module is used to determine a first mask image of the initial lesion image and a second mask image of at least one preoperative lesion image, wherein the number of second mask images is consistent with the number of at least one preoperative lesion image; The subtraction image determination module is used to determine a first subtraction image and a second subtraction image based on an initial lesion image and at least one preoperative lesion image; wherein, the first subtraction image is a positive subtraction image determined by subtracting a second matrix corresponding to the preoperative lesion image from a first matrix corresponding to the initial lesion image, and the second subtraction image is a negative subtraction image determined by subtracting the first matrix from the second matrix, and the number of second subtraction images is consistent with the number of at least one preoperative lesion image; The topological feature determination module is used to determine a first topological feature and a second topological feature based on a first subtraction image and a second subtraction image; wherein the first topological feature corresponds to the first subtraction image and the second topological feature corresponds to the second subtraction image; The target feature determination module is used to determine the target features corresponding to the lesion site based on the first subtraction image, the second subtraction image, the first topological feature, and the second topological feature. The judgment result determination module is used to input the target features into the pre-trained classification model and determine the judgment result corresponding to the lesion site; Specifically, the topological feature determination module is used to perform binarization segmentation on the first subtraction image and the second subtraction image based on a pre-set pixel threshold to obtain a first segmented image and a second segmented image; to perform erosion and dilation processing on the first segmented image and the second segmented image respectively to obtain the number of first genus cells in the first segmented image and the number of second genus cells in the second segmented image; and to determine the first topological feature and the second topological feature based on the number of first genus cells and the number of second genus cells.

Images