Artifact correction method and device, computer device and storage medium

Abstract

The application relates to an artifact correction method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring a medical image to be corrected; the medical image is an image with crosstalk artifacts; a preset correction model is used to correct the medical image, so as to obtain a corrected medical image; wherein the correction model is used to remove the crosstalk artifacts in the medical image; and the corrected medical image is an image after the crosstalk artifacts are removed. The method can improve the accuracy of the obtained corrected medical image.

Description

Technical Field

[0001] This application relates to the field of medical imaging technology, and in particular to an artifact correction method, apparatus, computer device, and storage medium. Background Technology

[0002] Typically, because the distance between two adjacent pixels in an array detector is very small, signal diffusion from one pixel to another can easily occur, resulting in crosstalk. This is seen in crosstalk phenomena, such as those observed when scanning a subject using computed tomography (CT). Crosstalk can occur in multiple directions around the detector, leading to artifacts in the reconstructed image, such as ring artifacts. Therefore, it is necessary to correct for crosstalk artifacts in the detector.

[0003] In traditional techniques, smoothing algorithms are typically used to fit image data with crosstalk to obtain image data that approximates that without crosstalk, and then the image data with crosstalk is corrected based on the fitting coefficients.

[0004] However, current methods for correcting crosstalk artifacts suffer from low accuracy. Summary of the Invention

[0005] Therefore, it is necessary to provide an artifact correction method, apparatus, computer equipment, and storage medium that can improve the accuracy of crosstalk artifact correction in CT images, addressing the aforementioned technical problems.

[0006] An artifact correction method, the method comprising:

[0007] Acquire the medical image to be corrected; the medical image is an image with crosstalk artifacts;

[0008] The medical image is corrected according to a preset correction model to obtain a corrected medical image; wherein the correction model is used to remove crosstalk artifacts in the medical image; the corrected medical image is the image after removing the crosstalk artifacts.

[0009] In one embodiment, the step of correcting the medical image according to a preset correction model to obtain a corrected medical image includes:

[0010] The medical image is input into the correction model, and the corrected medical image is obtained through the correction model.

[0011] In one embodiment, the step of correcting the medical image according to a preset correction model to obtain a corrected medical image includes:

[0012] A preset matrix is ​​input into the correction model, and a correction matrix corresponding to the medical image is obtained through the correction model; wherein, the size of the preset matrix is ​​the same as the size of the medical image, and all elements in the preset matrix are the same;

[0013] The medical image is corrected according to the correction matrix to obtain the corrected medical image.

[0014] In one embodiment, the correction model is trained in the following manner:

[0015] Obtain the sample medical image to be calibrated and the corresponding gold standard image; the sample medical image is an image including sample crosstalk artifacts; the corresponding gold standard image is an image after removing the sample crosstalk artifacts;

[0016] The sample medical image is input into a preset initial correction model to obtain a sample corrected image;

[0017] The initial calibration model is trained based on the sample calibration image and the corresponding gold standard image to obtain the calibration model.

[0018] In one embodiment, acquiring the sample medical image to be calibrated and the corresponding gold standard image of the sample medical image includes:

[0019] Acquire medical images free of crosstalk artifacts;

[0020] According to the preset construction method, the crosstalk-free medical image is constructed as the sample medical image to be corrected;

[0021] The medical image without crosstalk artifacts is used as the gold standard image corresponding to the sample medical image.

[0022] In one embodiment, constructing the crosstalk-free medical image into the sample medical image to be corrected according to a preset construction method includes:

[0023] Based on the feature information of the crosstalk-free medical image, the projection data corresponding to the crosstalk-free medical image is obtained; the feature information includes the attenuation correction coefficient and physical coefficient of the crosstalk-free medical image.

[0024] Multiply the projection data by a preset crosstalk artifact coefficient to obtain projection data with crosstalk artifacts.

[0025] The projection data with crosstalk artifacts is reconstructed to obtain the sample medical image to be corrected.

[0026] In one embodiment, acquiring the sample medical image to be calibrated and the corresponding gold standard image of the sample medical image includes:

[0027] Acquire a medical image including the sample crosstalk artifact, and use the medical image including the sample crosstalk artifact as the sample medical image;

[0028] Using a preset crosstalk correction method, the crosstalk artifacts of the sample are removed to obtain the gold standard image corresponding to the medical image of the sample.

[0029] An artifact correction device, the device comprising:

[0030] The first acquisition module is used to acquire the medical image to be corrected; the medical image is an image with crosstalk artifacts;

[0031] The correction module is used to correct the medical image according to a preset correction model to obtain a corrected medical image; wherein the correction model is used to remove crosstalk artifacts in the medical image; and the corrected medical image is the image after removing the crosstalk artifacts.

[0032] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program performing the following steps:

[0033] Acquire the medical image to be corrected; the medical image is an image with crosstalk artifacts;

[0034] The medical image is corrected according to a preset correction model to obtain a corrected medical image; wherein the correction model is used to remove crosstalk artifacts in the medical image; the corrected medical image is the image after removing the crosstalk artifacts.

[0035] A computer-readable storage medium having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0036] Acquire the medical image to be corrected; the medical image is an image with crosstalk artifacts;

[0037] The medical image is corrected according to a preset correction model to obtain a corrected medical image; wherein the correction model is used to remove crosstalk artifacts in the medical image; the corrected medical image is the image after removing the crosstalk artifacts.

[0038] The aforementioned artifact correction method, apparatus, computer equipment, and storage medium first acquire a medical image to be corrected containing crosstalk artifacts. Then, based on a preset correction model, the acquired medical image to be corrected with crosstalk artifacts is corrected, removing the crosstalk artifacts from the medical image to be corrected, thus obtaining a corrected medical image. Since the correction model is a pre-trained model for removing crosstalk artifacts from the medical image to be corrected, and the accuracy of the correction model trained with a large number of sample images is also high, the crosstalk artifacts in the medical image to be corrected can be accurately removed through the correction model, thereby improving the accuracy of the obtained corrected medical image. Attached Figure Description

[0039] Figure 1 A schematic diagram of the internal structure of a computer device provided for one embodiment;

[0040] Figure 2 This is a flowchart illustrating an artifact correction method in one embodiment;

[0041] Figure 2a This is a schematic diagram of the structure of the correction model in one embodiment;

[0042] Figure 3 This is a flowchart illustrating the artifact correction method in another embodiment;

[0043] Figure 3a This is a schematic diagram of the correction matrix in one embodiment;

[0044] Figure 4 This is a flowchart illustrating the artifact correction method in another embodiment;

[0045] Figure 5 This is a flowchart illustrating the artifact correction method in another embodiment;

[0046] Figure 5a A schematic diagram illustrating a method for determining crosstalk artifact coefficients as provided in one embodiment;

[0047] Figure 5b A schematic diagram of the detector configuration provided in one embodiment;

[0048] Figure 5c A schematic diagram of the detector configuration provided in one embodiment;

[0049] Figure 5d A schematic diagram illustrating a method for determining crosstalk artifact coefficients as provided in one embodiment;

[0050] Figure 6 This is a flowchart illustrating the artifact correction method in another embodiment;

[0051] Figure 7This is a structural block diagram of an artifact correction device in one embodiment. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0053] The artifact correction method provided in this application embodiment can be applied to, for example, Figure 1 The computer device shown includes a processor and a memory connected via a system bus. The memory stores a computer program, and the processor executes the computer program to perform the steps described in the method embodiments below. Optionally, the computer device may further include a network interface, a display screen, and an input device. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with external terminals via a network connection. Optionally, the computer device may be a server, a personal computer, a personal digital assistant, or other terminal devices, such as tablet computers, mobile phones, etc., or it may be a cloud or remote server. This application embodiment does not limit the specific form of the computer device.

[0054] In one embodiment, such as Figure 2 As shown, an artifact correction method is provided, which is applied to... Figure 1 Taking a computer device as an example, the explanation includes the following steps:

[0055] S201, Obtain the medical image to be corrected; the medical image is an image with crosstalk artifacts.

[0056] The medical image to be corrected is a CT (Computed Tomography) image. The crosstalk artifact in the medical image to be corrected occurs during the acquisition of the medical image data. Due to the small distance between two adjacent pixels of the array detector, the signal spreads from one pixel to another, causing crosstalk, which in turn leads to crosstalk artifacts in the reconstructed CT image.

[0057] Specifically, the computer device first acquires the medical image to be corrected, which contains crosstalk artifacts. Optionally, the computer device can acquire the medical image to be corrected in real time from a CT scanner, or it can acquire it from a PACS (Picture Archiving and Communication Systems) server. Optionally, the crosstalk artifacts in the medical image to be corrected can be ring-shaped or arc-shaped; this embodiment does not limit the shape of the crosstalk artifacts. It should be noted that the shape of the crosstalk artifacts in the medical image to be corrected is related to the location of crosstalk occurring on the detector; different crosstalk locations will produce different crosstalk artifacts.

[0058] S202, The medical image is corrected according to the preset correction model to obtain the corrected medical image; wherein, the correction model is used to remove crosstalk artifacts in the medical image; the corrected medical image is the image after removing crosstalk artifacts.

[0059] Specifically, the computer device corrects the acquired medical image to be corrected according to a preset correction model, resulting in a corrected medical image. The preset correction model is used to remove crosstalk artifacts from the medical image to be corrected; the resulting corrected medical image is the image after removing crosstalk artifacts. Optionally, the network structure of the preset correction model can be any neural network structure. It should be noted that the correction model described in this embodiment can be applied to correct crosstalk artifacts generated by any CT image acquired by a normal detector system capable of complete scanning.

[0060] For example, Figure 2a A schematic diagram of the structure of a correction model for removing crosstalk artifacts in medical images, as provided in one embodiment, is shown below. Figure 2aAs shown, the correction model can include: an input layer, convolutional layers, batch normalization layers, pooling layers, fully connected layers, loss layers, and an output layer. The input layer is used for data input. During training, the input data consists of several training samples, which can be several medical images to be corrected and their corresponding gold standard images. The medical images to be corrected include images with sample crosstalk artifacts, and the corresponding gold standard images are images after removing the sample crosstalk artifacts. The convolutional layers are used for feature extraction and feature mapping. Low-level convolutional layers may only extract low-level features such as edges, lines, and corners; more layers can iteratively extract more complex features from low-level features. The batch normalization layer is used to force the input distribution, which gradually approaches the saturation region of the value range after mapping to the nonlinear function, back to a relatively standard normal distribution with a mean of 0 and a variance of 1. This ensures that the input value of the nonlinear transformation function falls into a region that is more sensitive to the input, thus avoiding the gradient vanishing problem. Pooling layers are used to downsample the data, learn and classify multi-scale data features, improve the model's classification accuracy, and provide non-linearity, reducing the number of model parameters and mitigating overfitting. Fully connected layers are used for refitting at the end of the model, reducing feature loss. The loss layer accepts two inputs: the neural network's predicted value and the ground truth label (i.e., the gold standard image corresponding to the medical image to be corrected). The loss layer performs a series of operations on these two inputs to obtain the current network's loss function. The goal of deep learning is to find the weights in the weight space that minimize the loss function. The loss function is obtained during forward propagation and is also the starting point for backpropagation. The loss function is generally composed of the true value and the predicted value. A correct loss function can keep the predicted value close to the true value; the loss is minimized when the predicted value equals the true value. In this embodiment, the preferred loss function is the normalized exponential function, cross-entropy loss function, or squared error loss function. The output layer outputs the corrected medical image; in this example, the corrected medical image is the one after removing crosstalk artifacts. It should be noted that the convolutional layer, batch normalization layer, and pooling layer in the structure of the correction model described above can be one or more. This embodiment does not limit the number of convolutional layers, batch normalization layers, and pooling layers. In actual use, the number of convolutional layers, batch normalization layers, and pooling layers can be adjusted according to actual needs.

[0061] In the aforementioned artifact correction method, the computer device first acquires the medical image to be corrected, which contains crosstalk artifacts. Then, based on a preset correction model, it can correct the acquired medical image to be corrected, removing the crosstalk artifacts and obtaining the corrected medical image. Since the correction model is a pre-trained model for removing crosstalk artifacts from the medical image to be corrected, and the accuracy of the correction model trained with a large number of sample images is also high, the crosstalk artifacts in the medical image to be corrected can be accurately removed through the correction model, thereby improving the accuracy of the obtained corrected medical image.

[0062] In the scenario described above, where a medical image to be corrected is corrected according to a preset correction model to obtain a corrected medical image, the computer device can directly use the correction model to correct the medical image, or it can use the correction model to obtain correction parameters and then use these parameters to correct the medical image to obtain the corrected medical image. The following details the specific implementation methods for removing crosstalk artifacts from the medical image to be corrected, resulting in a medical image with crosstalk artifacts removed.

[0063] In one embodiment, if a computer device corrects a medical image to be corrected according to a preset correction model to obtain a corrected medical image, the above-mentioned S202 includes: inputting the medical image into the correction model, and obtaining the corrected medical image through the correction model.

[0064] Specifically, the computer device inputs the medical image to be corrected into the preset correction model. The preset correction model removes crosstalk artifacts from the medical image to obtain the corrected medical image. Optionally, the computer device can input the entire medical image to be corrected into the preset correction model to remove crosstalk artifacts, or it can divide the medical image to be corrected into different sub-images according to a preset size, input each sub-image into the preset correction model to remove crosstalk artifacts from each sub-image, and then combine the crosstalk artifact-free sub-images according to the order in which the medical image to be corrected was divided to obtain the corrected medical image.

[0065] In this embodiment, the computer device inputs the medical image to be corrected into a preset correction model. Through the preset correction model, crosstalk artifacts in the medical image to be corrected can be accurately removed, thereby obtaining a more accurate corrected medical image and improving the accuracy of the obtained corrected medical image.

[0066] In one embodiment, if a computer device obtains correction parameters using a correction model, and then uses these parameters to correct the medical image to be corrected, a corrected medical image is obtained, such as... Figure 3As shown, S202 above includes:

[0067] S301, Input the preset matrix into the calibration model, and obtain the calibration matrix corresponding to the medical image through the calibration model; wherein, the size of the preset matrix is ​​the same as the size of the medical image, and all elements in the preset matrix are the same.

[0068] Specifically, the computer device inputs a preset matrix into the aforementioned correction model, and through the correction model, obtains the correction matrix corresponding to the medical image to be corrected; wherein, the size of the preset matrix is ​​the same as the size of the medical image to be corrected, and all elements in the preset matrix are identical. Optionally, all elements in the preset matrix can be one. It is understood that the correction matrix output by the correction model is a parameter matrix, which can be used to correct crosstalk artifacts in the medical image to be corrected. For example, as shown... Figure 3a As shown, Figure 3a a in 11 ,a 12 ,L,a 14 This refers to the values ​​of each element in the correction matrix corresponding to the medical image to be corrected. It can be understood that the aforementioned preset correction model is a pre-trained model used to remove crosstalk artifacts from the medical image to be corrected. The parameters in the model are fixed. Therefore, by inputting a preset matrix with all elements equal to one into the correction model, the parameters of the correction model can be used to adjust the element values ​​of the preset matrix, thereby obtaining the correction matrix corresponding to the medical image to be corrected.

[0069] S302, Correct the medical image according to the correction matrix to obtain the corrected medical image.

[0070] Specifically, the computer device corrects the medical image to be corrected based on the correction matrix obtained above, thereby obtaining a corrected medical image. Optionally, the computer device can multiply the obtained correction matrix with the medical image to be corrected to obtain the corrected medical image.

[0071] In this embodiment, the computer device inputs a preset matrix into a preset correction model. Through the correction model, the correction matrix corresponding to the medical image to be corrected can be obtained. Since the preset correction model is a pre-trained model, the correction matrix corresponding to the medical image to be corrected can be accurately obtained through the correction model, thereby improving the accuracy of the obtained correction matrix corresponding to the medical image to be corrected. Based on the obtained correction matrix, the medical image to be corrected can be accurately corrected, thereby improving the accuracy of the obtained corrected medical image.

[0072] In the scenario described above where medical images are corrected according to a preset correction model, the correction model is a pre-trained model. In one embodiment, such as... Figure 4 As shown, the above correction model was trained in the following way:

[0073] S401, acquire the sample medical image to be corrected and the corresponding gold standard image; the sample medical image is an image including sample crosstalk artifacts; the corresponding gold standard image is an image after removing sample crosstalk artifacts.

[0074] Specifically, the computer device acquires the sample medical image to be calibrated and its corresponding gold standard image. The sample medical image to be calibrated includes sample crosstalk artifacts, and the corresponding gold standard image is the image after removing the crosstalk artifacts. Optionally, the computer device can acquire an image without crosstalk artifacts from the PACS server as the gold standard image corresponding to the sample medical image, and then construct artifacts on this image to obtain the sample medical image to be calibrated. Alternatively, it can acquire an image with crosstalk artifacts from the PACS server as the sample medical image to be calibrated, and then remove the crosstalk artifacts to obtain the corresponding gold standard image.

[0075] S402, input the sample medical image into the preset initial calibration model to obtain the sample calibration image.

[0076] Specifically, the computer device inputs the acquired sample medical image to be corrected into the aforementioned preset initial correction model. Through the initial correction model, a sample corrected image is obtained, wherein the sample corrected image is an image in which crosstalk artifacts have been removed from the sample medical image to be corrected. Optionally, the network structure of the initial correction model can be any neural network structure.

[0077] S403, the initial calibration model is trained based on the sample calibration image and the gold standard image corresponding to the sample image to obtain the calibration model.

[0078] Specifically, the computer device compares the obtained sample-corrected image with the corresponding gold standard image to obtain the loss function of the initial correction model. This loss function is typically, but not limited to, an L1 or L2 loss function. The initial correction model is trained by iterating through multiple iterations to reduce the value of the loss function, thus obtaining the aforementioned correction model. Optionally, the computer device can determine the initial correction model as the one whose loss function reaches a stable value or a minimum value.

[0079] In this embodiment, the computer device first acquires the sample medical image to be calibrated and the corresponding gold standard image. Then, it inputs the sample medical image into a preset initial calibration model to obtain the sample calibration image. The initial calibration model can then be trained based on the difference between the sample calibration image and the corresponding gold standard image. In this way, by using a large number of sample medical images to be calibrated and the corresponding gold standard images, the initial calibration model can be accurately trained, thereby obtaining a more accurate calibration model and improving the accuracy of the obtained calibration model.

[0080] In the scenario described above, where a computer device acquires a sample medical image to be calibrated and its corresponding gold standard image, it can either acquire a crosstalk-free medical image using a forward acquisition method, and then construct it into a crosstalk-inducing image according to a preset construction method, or acquire a crosstalk-inducing medical image using a reverse acquisition method, and then remove the crosstalk artifacts from the medical image to obtain the sample medical image to be calibrated and its corresponding gold standard image. The following details the specific implementation methods for acquiring the sample medical image to be calibrated and its corresponding gold standard image using different approaches.

[0081] In one embodiment, if a computer device obtains the sample medical image to be calibrated and the corresponding gold standard image of the sample medical image through a forward acquisition method, such as Figure 5 As shown, S401 includes:

[0082] S501, acquire medical images without crosstalk artifacts.

[0083] Specifically, the computer equipment acquires crosstalk-free medical images. These crosstalk-free medical images are those reconstructed by the detector system during the acquisition of medical image data without crosstalk occurring. Optionally, the computer equipment can acquire crosstalk-free medical images from a PACS server.

[0084] S502, according to the preset construction method, construct the medical image without crosstalk artifacts into the sample medical image to be corrected.

[0085] Specifically, the computer device constructs the medical image without crosstalk artifacts into the sample medical image to be corrected according to a preset construction method. Optionally, the computer device can first obtain the projection data corresponding to the medical image without crosstalk artifacts based on the feature information of the medical image without crosstalk artifacts, wherein the feature information of the medical image without crosstalk artifacts includes the attenuation correction coefficient and physical coefficient of the medical image without crosstalk artifacts, and then multiply the obtained projection data corresponding to the medical image without crosstalk artifacts by a preset crosstalk artifact coefficient to obtain the projection data with crosstalk artifacts, and then reconstruct the projection data with crosstalk artifacts to obtain the sample medical image to be corrected. The computer device can determine the preset crosstalk artifact coefficient in the following way: (1) Assuming that crosstalk occurs between a pixel of a certain detector and its adjacent pixels, for example, such as Figure 5a As shown, assuming the current pixel only experiences crosstalk with its adjacent previous pixel, firstly, as... Figure 5d As shown in the first figure, the crosstalk coefficient of each pixel is set to c. If crosstalk occurs, that is, the nth pixel multiplies the received signal strength with the crosstalk coefficient c, and then transmits the multiplied signal strength to the (n-1)th adjacent pixel through crosstalk. Based on this, a fixed parameter c can be determined as the preset crosstalk artifact coefficient; (2) as Figure 5b and Figure 5c As shown, Figure 5b and Figure 5c This is a schematic diagram of the detector configuration. It is assumed that for each detector module, the crosstalk artifact coefficient is the same within the module, but the crosstalk artifact signal is suppressed at the interface between two detector modules. Based on this, a crosstalk artifact coefficient can be determined for different detector modules, such as... Figure 5d In the second image, 'c' represents the crosstalk artifact coefficient corresponding to a detector, thus obtaining the preset crosstalk artifact coefficient. For example, assuming that the medical image is obtained by data acquisition from two detector modules a and b, the crosstalk artifact coefficients inside detector module a can be set to a1, and the crosstalk artifact coefficients inside detector module b can be set to b1. a1 and b1 are determined as the preset crosstalk artifact coefficients. (3) Assuming that the crosstalk artifact coefficients between pixels of each medical image with crosstalk artifacts are random, but the crosstalk artifact coefficient of each pixel is in the range of [-0.07, 0.07], such as Figure 5d As shown, the crosstalk artifact coefficient of pixel x12 can be 0.05 (meaning that the signal strength of pixel x11 is added to x11), and the crosstalk artifact coefficient of pixel x13 can be -0.06 (meaning that the signal strength of pixel x13 will be added to x14).

[0086] S503 uses medical images without crosstalk artifacts as the gold standard images corresponding to the sample images.

[0087] Specifically, the computer device uses the acquired crosstalk-free medical image as the gold standard image corresponding to the sample medical image. It is understood that the calibration model is used to remove crosstalk artifacts from the medical image to be calibrated. Therefore, the gold standard image used when training the calibration model is a crosstalk-free medical image. In this embodiment, the computer device can directly use the acquired crosstalk-free medical image as the gold standard image corresponding to the sample medical image to be calibrated.

[0088] In this embodiment, the computer device first acquires a crosstalk-free medical image and uses it as the gold standard image corresponding to the sample medical image to be calibrated. Then, according to a preset construction method, the acquired crosstalk-free medical image is constructed as the sample medical image to be calibrated. Since the acquired crosstalk-free medical image is directly used as the gold standard image corresponding to the sample medical image to be calibrated, the accuracy of the gold standard image corresponding to the obtained sample medical image to be calibrated is guaranteed, ensuring that the gold standard image corresponding to the obtained sample medical image to be calibrated is a crosstalk-free medical image. In addition, the sample medical image to be calibrated is constructed from the acquired crosstalk-free medical image according to the preset construction method, and the process of constructing the acquired crosstalk-free medical image according to the preset construction method is relatively simple, improving the efficiency of obtaining the sample medical image to be calibrated.

[0089] In one embodiment, if a computer device obtains the sample medical image to be corrected and the corresponding gold standard image of the sample medical image through a reverse acquisition method, such as Figure 6 As shown, S401 includes:

[0090] S601, acquire a medical image including sample crosstalk artifacts, and use the medical image including sample crosstalk artifacts as a sample medical image.

[0091] Specifically, the computer device acquires a medical image including sample crosstalk artifacts and uses this medical image, including the sample crosstalk artifacts, as the sample medical image to be corrected. Optionally, the computer device can acquire the medical image including sample crosstalk artifacts from a PACS server. Optionally, the sample crosstalk artifacts can be strip-shaped artifacts, ring-shaped artifacts, or artifacts of other shapes.

[0092] S602 uses a preset crosstalk correction method to remove sample crosstalk artifacts and obtain the gold standard image corresponding to the sample medical image.

[0093] Specifically, the computer device uses a preset crosstalk correction method to remove sample crosstalk artifacts from the medical image containing the aforementioned sample crosstalk artifacts, thereby obtaining the gold standard image corresponding to the sample medical image. It should be noted that the crosstalk correction method in this embodiment can be any available crosstalk correction method, such as fitting smoothing, etc., and this embodiment is not limited thereto. It is understood that using the preset crosstalk correction method corrects the medical image containing sample crosstalk artifacts to an equivalent medical image without sample crosstalk artifacts.

[0094] In this embodiment, the computer device first acquires a medical image including sample crosstalk artifacts and uses it as a sample medical image. This allows for accurate identification of the sample medical image. Then, using a preset fitting method, the sample crosstalk artifacts in the acquired medical image can be removed quickly and accurately, thereby improving the efficiency and accuracy of the gold standard image corresponding to the obtained sample medical image. In addition, the computer device directly uses the acquired medical image including sample crosstalk artifacts as the sample medical image, which allows for rapid identification of the sample medical image and improves the efficiency of obtaining the sample medical image.

[0095] To facilitate understanding by those skilled in the art, the artifact correction method provided in this disclosure is described in detail below. This method may include:

[0096] S1, acquire the sample medical image to be corrected and the corresponding gold standard image; the sample medical image is an image including sample crosstalk artifacts; the corresponding gold standard image is an image after removing sample crosstalk artifacts.

[0097] S2, input the sample medical image into the preset initial calibration model to obtain the sample calibration image.

[0098] S3. Based on the sample correction image and the gold standard image corresponding to the sample image, train the initial correction model to obtain the correction model.

[0099] S4, acquire the medical image to be corrected; the medical image is an image with crosstalk artifacts; execute S5, or execute S6-S7.

[0100] S5. Input the medical image into the calibration model, and obtain the calibrated medical image through the calibration model; wherein, the calibration model is used to remove crosstalk artifacts in the medical image; the calibrated medical image is the image after removing crosstalk artifacts.

[0101] S6. Input the preset matrix into the calibration model, and obtain the calibration matrix corresponding to the medical image through the calibration model; wherein, the size of the preset matrix is ​​the same as the size of the medical image.

[0102] S7. Based on the correction matrix, the medical image is corrected to obtain the corrected medical image.

[0103] In step S1, which involves acquiring the sample medical image to be corrected and its corresponding gold standard image, the computer device can either acquire a crosstalk-free medical image using a forward acquisition method, and then construct it into a crosstalk-inducing medical image according to a preset construction method, or acquire a crosstalk-inducing medical image using a reverse acquisition method, and then remove the crosstalk artifacts to obtain the sample medical image to be corrected and its corresponding gold standard image. The specific implementation methods of these two approaches are detailed below:

[0104] (i) If a computer device first acquires a medical image without crosstalk artifacts through a forward acquisition method, and then constructs the medical image without crosstalk artifacts into a medical image with crosstalk artifacts according to a preset construction method, then the above S1 includes:

[0105] D1, acquire medical images without crosstalk artifacts.

[0106] D2, based on the feature information of the crosstalk-free medical image, obtain the projection data corresponding to the crosstalk-free medical image; the feature information includes the attenuation correction coefficient and physical coefficient of the crosstalk-free medical image.

[0107] D3 multiplies the projection data with the preset crosstalk artifact coefficient to obtain projection data with crosstalk artifacts.

[0108] D4 reconstructs the projection data with crosstalk artifacts to obtain the sample medical image to be corrected.

[0109] (ii) If a computer device first acquires a medical image with crosstalk artifacts through a reverse acquisition method, and then removes the crosstalk artifacts from the medical image to obtain a sample medical image to be corrected and the corresponding gold standard image, then S1 above includes:

[0110] H1, acquire medical images including sample crosstalk artifacts, and use the medical images including sample crosstalk artifacts as sample medical images.

[0111] H2 uses a preset fitting method to remove sample crosstalk artifacts and obtain the gold standard image corresponding to the sample medical image.

[0112] It should be noted that the descriptions of S1-S7, D1-D4, and H1-H2 above can be found in the relevant descriptions in the above embodiments, and their effects are similar. Therefore, they will not be repeated here.

[0113] It should be understood that, although Figure 2-6 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 2-6 At least some of the steps in the process may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but may be executed at different times. The execution order of these steps or stages is not necessarily sequential, but may be executed in turn or alternately with other steps or at least some of the steps or stages in other steps.

[0114] In one embodiment, such as Figure 7 As shown, an artifact correction device is provided, comprising: a first acquisition module and a correction module, wherein:

[0115] The first acquisition module is used to acquire the medical image to be corrected; the medical image is an image with crosstalk artifacts.

[0116] The correction module is used to correct medical images according to a preset correction model to obtain corrected medical images; wherein, the correction model is used to remove crosstalk artifacts in medical images; the corrected medical image is the image after removing crosstalk artifacts.

[0117] The artifact correction device provided in this embodiment can perform the above-described method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0118] Based on the above embodiments, optionally, the correction module includes: a first correction unit, wherein:

[0119] The first calibration unit is used to input the medical image into the calibration model, and obtain the calibrated medical image through the calibration model.

[0120] The artifact correction device provided in this embodiment can perform the above-described method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0121] Based on the above embodiments, optionally, the correction module includes: a first acquisition unit and a second correction unit, wherein:

[0122] The first acquisition unit is used to input a preset matrix into the correction model and obtain the correction matrix corresponding to the medical image through the correction model; wherein, the size of the preset matrix is ​​the same as the size of the medical image, and all elements in the preset matrix are one.

[0123] The second correction unit is used to correct the medical image according to the correction matrix to obtain the corrected medical image.

[0124] The artifact correction device provided in this embodiment can perform the above-described method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0125] Based on the above embodiments, optionally, the apparatus further includes: a second acquisition module, a third acquisition module, and a training module, wherein:

[0126] The second acquisition module is used to acquire the sample medical image to be corrected and the corresponding gold standard image; the sample medical image is an image including sample crosstalk artifacts; the corresponding gold standard image is an image after removing sample crosstalk artifacts.

[0127] The third acquisition module is used to input the sample medical image into a preset initial calibration model to obtain the sample calibration image.

[0128] The training module is used to train the initial calibration model based on the sample calibration image and the gold standard image corresponding to the sample image, so as to obtain the calibration model.

[0129] The artifact correction device provided in this embodiment can perform the above-described method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0130] Based on the above embodiments, optionally, the second acquisition module includes: a second acquisition unit, a construction unit, and a third acquisition unit, wherein:

[0131] The second acquisition unit is used to acquire medical images free of crosstalk artifacts.

[0132] The construction unit is used to construct a sample medical image to be corrected from a medical image without crosstalk artifacts according to a preset construction method.

[0133] The third acquisition unit is used to use the medical image without crosstalk artifacts as the gold standard image corresponding to the sample medical image.

[0134] The artifact correction device provided in this embodiment can perform the above-described method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0135] Based on the above embodiments, optionally, the above-mentioned construction unit is specifically used to obtain the projection data corresponding to the medical image without crosstalk artifacts based on the feature information of the medical image without crosstalk artifacts; the feature information includes the attenuation correction coefficient and physical coefficient of the medical image without crosstalk artifacts.

[0136] The projection data is multiplied by a preset crosstalk artifact coefficient to obtain projection data with crosstalk artifacts; the projection data with crosstalk artifacts is reconstructed to obtain the sample medical image to be corrected.

[0137] The artifact correction device provided in this embodiment can perform the above-described method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0138] Based on the above embodiments, optionally, the second acquisition module includes: a fourth acquisition unit and a fifth acquisition unit, wherein:

[0139] The fourth acquisition unit is used to acquire medical images including sample crosstalk artifacts and use the medical images including sample crosstalk artifacts as sample medical images.

[0140] The fifth acquisition unit is used to remove sample crosstalk artifacts using a preset fitting method to obtain the gold standard image corresponding to the sample medical image.

[0141] The artifact correction device provided in this embodiment can perform the above-described method embodiment, and its implementation principle and technical effect are similar, so it will not be described again here.

[0142] Specific limitations regarding the artifact correction device can be found in the limitations of the artifact correction method described above, and will not be repeated here. Each module in the aforementioned artifact correction device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in hardware or independently of the processor in the computer device, or stored in software in the memory of the computer device, so that the processor can call and execute the corresponding operations of each module.

[0143] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0144] Acquire the medical image to be corrected; the medical image is an image with crosstalk artifacts;

[0145] The medical image is corrected according to the preset correction model to obtain the corrected medical image; the correction model is used to remove crosstalk artifacts in the medical image; the corrected medical image is the image after removing crosstalk artifacts.

[0146] The computer device provided in the above embodiments has similar implementation principles and technical effects to the above method embodiments, and will not be described again here.

[0147] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0148] Acquire the medical image to be corrected; the medical image is an image with crosstalk artifacts;

[0149] The medical image is corrected according to the preset correction model to obtain the corrected medical image; the correction model is used to remove crosstalk artifacts in the medical image; the corrected medical image is the image after removing crosstalk artifacts.

[0150] The computer-readable storage medium provided in the above embodiments has similar implementation principles and technical effects to the above method embodiments, and will not be described again here.

[0151] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, or optical storage, etc. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0152] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0153] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An artifact correction method, characterized in that, The method includes: Acquire a medical image to be corrected; the medical image is an image with crosstalk artifacts; the crosstalk artifacts are artifacts caused by crosstalk between two adjacent pixels of the detector during the acquisition of the medical image; the shape of the crosstalk artifacts is related to the location where the crosstalk occurs in the detector; A preset matrix is ​​input into a preset correction model. Through the correction model, a correction matrix corresponding to the medical image is obtained. The correction matrix is ​​multiplied by the medical image to obtain the corrected medical image. The size of the preset matrix is ​​the same as the size of the medical image, and all elements in the preset matrix are the same. The corrected medical image is the image after removing the crosstalk artifacts. The correction model is trained based on the sample medical image to be corrected and the gold standard image corresponding to the sample medical image. The sample medical image is determined based on the projection data corresponding to the obtained crosstalk artifact-free medical image and the preset crosstalk artifact coefficients. The crosstalk artifact coefficients are determined based on the detector. The gold standard image is the crosstalk artifact-free medical image.

2. The method according to claim 1, characterized in that, The correction model is trained in the following manner: Obtain the sample medical image to be calibrated and the corresponding gold standard image; the sample medical image is an image including sample crosstalk artifacts; the corresponding gold standard image is an image after removing the sample crosstalk artifacts; The sample medical image is input into a preset initial calibration model to obtain a sample calibration image; The initial calibration model is trained based on the sample calibration image and the gold standard image corresponding to the sample medical image to obtain the calibration model.

3. The method according to claim 2, characterized in that, The process of acquiring the sample medical image to be corrected and the corresponding gold standard image includes: Acquire medical images free of crosstalk artifacts; According to the preset construction method, the crosstalk-free medical image is constructed as the sample medical image to be corrected; The medical image without crosstalk artifacts is used as the gold standard image corresponding to the sample medical image.

4. The method according to claim 3, characterized in that, The step of constructing the crosstalk-free medical image into the sample medical image to be corrected according to a preset construction method includes: Based on the feature information of the crosstalk-free medical image, the projection data corresponding to the crosstalk-free medical image is obtained; the feature information includes the attenuation correction coefficient and physical coefficient of the crosstalk-free medical image. Multiply the projection data by a preset crosstalk artifact coefficient to obtain projection data with crosstalk artifacts. The projection data with crosstalk artifacts is reconstructed to obtain the sample medical image to be corrected.

5. An artifact correction device, characterized in that, The device includes: The first acquisition module is used to acquire a medical image to be corrected; the medical image is an image with crosstalk artifacts; the crosstalk artifacts are artifacts caused by crosstalk between two adjacent pixels of the detector during the acquisition of the medical image; the shape of the crosstalk artifacts is related to the location where the crosstalk occurs in the detector; A calibration module is used to input a preset matrix into a preset calibration model, obtain a calibration matrix corresponding to the medical image through the calibration model, and multiply the calibration matrix with the medical image to obtain a calibrated medical image. The preset matrix has the same size as the medical image, and all elements in the preset matrix are identical. The calibrated medical image is the image after removing crosstalk artifacts. The calibration model is trained based on a sample medical image to be calibrated and a corresponding gold standard image. The sample medical image is determined based on the projection data corresponding to an acquired crosstalk artifact-free medical image and preset crosstalk artifact coefficients. The crosstalk artifact coefficients are determined based on the detector. The gold standard image is the crosstalk artifact-free medical image.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 4.

Images