Method and apparatus for expanding field of view of CT image, and electronic device

Abstract

Disclosed in the present application are a method and apparatus for expanding the field of view of a CT image, and an electronic device. The method comprises: acquiring first projection data obtained by means of a CT scanner scanning a target object; on the basis of a field-of-view expansion algorithm, performing field-of-view expansion processing on the first projection data, so as to obtain an initial image; performing segmentation processing of bones and soft tissues on the initial image, so as to obtain anatomical information corresponding to the initial image; transforming the initial image and the anatomical information into a projection domain, and on the basis of the anatomical information in the projection domain, performing projection data restoration processing on the initial image, so as to obtain second projection data; and transforming the second projection data into an image domain, so as to obtain a target image. Thus, the field of view of a CT image is expanded jointly in a projection domain and an image domain, thereby improving the quality of the CT image.

Description

Field of view expansion method and device of CT image and electronic equipment

[0001] The present application claims priority to the Chinese patent application No. 2024112359339, filed on September 4, 2024, and entitled "Field of view expansion method and device of CT image and electronic equipment", which is incorporated by reference in its entirety.

TECHNICAL FIELD

[0002] The present application relates to the CT technical field, and more particularly, to a field of view expansion method and device of CT image and electronic equipment.

BACKGROUND

[0003] Computed Tomography (CT) is a medical imaging technique that uses X-ray beams to perform tomographic scanning on the human body and produces detailed images of the internal structure of the body with the help of a computer. In a CT imaging system, when the geometric structure of the CT machine and the size of its detector are determined, the maximum scanning range is also determined. Although the current CT machine has a relatively large scanning range, when encountering obese patients or patients with improper positioning, it cannot guarantee that the pre-scanning parts of the patients are within the CT machine scan field of view (sFOV). Therefore, in CT imaging technology, there is a challenge of expanding the CT field of view (FOV).

SUMMARY

[0004] In view of the above problems, the embodiments of the present application propose a field of view expansion method and device of CT image and electronic equipment.

[0005] In a first aspect, the embodiments of the present application provide a field of view expansion method of CT image, the method comprising: obtaining first projection data obtained by scanning a target object by a CT machine; performing field of view expansion processing on the first projection data based on a field of view expansion algorithm to obtain an initial image; performing segmentation processing of bone and soft tissue on the initial image to obtain anatomical information corresponding to the initial image; converting the initial image and the anatomical information to a projection domain, and performing projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data; converting the second projection data to an image domain to obtain a target image.

[0006] In a second aspect, the embodiments of the present application provide a CT image field of view expansion device, the device comprising: a first projection data acquisition module, a field of view expansion processing module, an anatomical information acquisition module, a projection domain repair module, and a CT image acquisition module. The first projection data acquisition module is configured to acquire first projection data obtained by scanning a target object using a CT machine. The field of view expansion processing module is configured to perform field of view expansion processing on the first projection data based on a field of view expansion algorithm to obtain an initial image. The anatomical information acquisition module is configured to perform bone and soft tissue segmentation processing on the initial image to obtain anatomical information corresponding to the initial image. The projection domain repair module is configured to convert the initial image and the anatomical information to a projection domain, and perform projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data. The CT image acquisition module is configured to convert the second projection data to an image domain to obtain a target image.

[0007] In a third aspect, the embodiments of the present application provide an electronic device comprising a memory and a processor, wherein the memory is coupled to the processor, and the memory stores instructions which, when executed by the processor, cause the processor to perform the above method.

[0008] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, wherein the computer readable storage medium stores program codes which can be invoked by a processor to perform the above method.

[0009] In the scheme of the present application, the first projection data obtained by scanning a target object using a CT machine is acquired, and field of view expansion processing is performed on the first projection data based on a field of view expansion algorithm to obtain an initial image, so as to preliminarily expand the CT field of view in the image domain. The initial image is subjected to bone and soft tissue segmentation processing to obtain anatomical information corresponding to the initial image, and the initial image and the anatomical information are converted to a projection domain. Projection data repair processing is performed on the initial image based on the anatomical information in the projection domain to obtain second projection data, so as to guide the recovery of the projection data based on the anatomical information in the projection domain. The second projection data is converted to an image domain to obtain a target image, so as to obtain a CT image with an expanded field of view. Thus, the CT field of view is expanded in the projection domain and the image domain, and the segmentation mechanism is introduced to constrain image recovery, so as to obtain a CT image with an expanded field of view and improve the quality of the CT image with an expanded field of view. BRIEF DESCRIPTION OF DRAWINGS

[0010] In order to make the technical solutions in the embodiments of the present application clearer, the accompanying drawings needed in the embodiment description will be briefly introduced. Obviously, the accompanying drawings in the following description only constitute some embodiments of the present application, and other accompanying drawings can be obtained by those skilled in the art without any creative effort.

[0011] FIG. 1 shows a schematic diagram of a CT image obtained based on a classical traditional algorithm WCE according to an embodiment of the present application;

[0012] FIG. 2 shows a flowchart of a field of view expansion method of a CT image according to an embodiment of the present application;

[0013] FIG. 3 shows a flowchart of a field of view expansion method of a CT image according to an embodiment of the present application;

[0014] FIG. 4 shows a flowchart of a field of view expansion method of a CT image according to an embodiment of the present application;

[0015] FIG. 5 shows a schematic diagram of simulation of obtained projection data according to an embodiment of the present application;

[0016] FIG. 6 shows a schematic diagram of projection data and truncated data corresponding to the projection data according to an embodiment of the present application;

[0017] FIG. 7 shows a schematic diagram of a reconstructed image corresponding to the projection data and a reconstructed image corresponding to the truncated data according to an embodiment of the present application;

[0018] FIG. 8 shows a flowchart of a field of view expansion method of a CT image according to an embodiment of the present application;

[0019] FIG. 9 shows a flowchart of a field of view expansion method of a CT image according to an embodiment of the present application;

[0020] FIG. 10 shows a module block diagram of a field of view expansion device of a CT image according to an embodiment of the present application;

[0021] FIG. 11 shows a block diagram of an electronic device for performing a field of view expansion method of a CT image according to an embodiment of the present application;

[0022] FIG. 12 shows a storage unit for storing or carrying program code for implementing a field of view expansion method of a CT image according to an embodiment of the present application.

DETAILED DESCRIPTION

[0023] In order to make the technical solutions in the embodiments of the present application clearer, the accompanying drawings needed in the embodiment description will be briefly introduced. Obviously, the accompanying drawings in the following description only constitute some embodiments of the present application, and other accompanying drawings can be obtained by those skilled in the art without any creative effort.

[0024] The relative positions among the detector, the light source and the object center determine the sFOV of the CT. Due to the mechanical structure design of the CT system and the cost consideration of the detector, the diameter of the sFOV is usually 40cm to 50cm, which is smaller than the patient positioning space. When the patient is obese or is positioned off-center, part of the body region is out of the sFOV region, and the projection data is partially truncated, that is, the patient scanning part cannot be fully acquired at all angles. If the truncated data is directly used to reconstruct the CT image, artifacts will be generated, and the tissue structure of the out-of-view field will be distorted and the gray scale will be shifted, thereby affecting the doctor's diagnosis.

[0025] A method of reconstructing a CT image based on physical principle modeling to complete the missing projection data is proposed in the related art; however, the premise of this method is that the out-of-view field part is soft tissue and has a regular shape, but in real life, people are different, and the out-of-view field part is not completely soft tissue or regular soft tissue. Therefore, this method has the problems of simple assumption, limited performance, large error, severe deformation, and inability to handle various complex situations (such as large-scale truncation and missing area containing foreign matter). For example, refer to FIG. 1, which shows a schematic diagram of a CT image obtained based on the classic traditional algorithm WCE proposed in an embodiment of the present application. The classic traditional algorithm WCE can partially recover the information in the extended field of view (eFOV), but the gray scale of the filled area is shifted and the shape is inaccurate. Therefore, in the related art, there is a challenge of extending the CT field of view and ensuring the quality of the CT image.

[0026] To solve the above problems, the inventors have found, after long-term research, that the CT image field of view extension method, device and electronic equipment provided in the embodiments of the present application jointly extend the FOV in the projection domain and the image domain, and introduce a segmentation mechanism to constrain image recovery, thereby obtaining a CT image with an extended field of view while improving the quality of the CT image with an extended field of view. The specific CT image field of view extension method is described in detail in subsequent embodiments.

[0027] In order to better understand the scheme of the embodiments of the present application, the technical terms used in the embodiments of the present application are explained as follows.

[0028] Peak Signal To Noise Ratio (PSNR).

[0029] Mean Absolute Error (MAE).

[0030] Classic algorithm WCE, a new reconstruction algorithm for extending the field of view of CT scanning.

[0031] CT imaging is a technique that combines X-ray scan projection data with reconstruction mathematics and computer technology to obtain medical images based on layer information.

[0032] Referring to FIG. 2, FIG. 2 shows a flowchart of a CT image field of view expansion method according to an embodiment of the present application. In specific embodiments, the CT image field of view expansion method can be applied to a CT image field of view expansion device 200 as shown in FIG. 10 and an electronic device 100 (FIG. 11) configured with the CT image field of view expansion device 200. In the following, the specific process of the present embodiment will be described taking the electronic device as an example. It should be understood that the electronic device to which the present embodiment is applied can include a desktop computer, a notebook computer, a robot, a CT machine, and the like, without limitation. The CT image field of view expansion method can specifically include the following steps with respect to the flowchart shown in FIG. 2:

[0033] Step S110: Obtain first projection data obtained by scanning a target object by a CT machine.

[0034] In some embodiments, the electronic device can obtain the projection data obtained by scanning the target object by the CT machine as the first projection data from the associated cloud or electronic device through wireless communication technology (such as Bluetooth, WiFi, zigbee technology, etc.); the electronic device can also obtain the projection data obtained by scanning the target object by the CT machine as the first projection data from the associated electronic device through a serial communication interface (such as a serial peripheral interface, a universal serial bus, etc.). Alternatively, the electronic device can also include a CT machine, and accordingly, the electronic device can obtain the first projection data by scanning the target object by the CT machine. The target object can include an object or a human body. The CT machine can scan a part of the target object to obtain the first projection data. For example, the CT machine can scan a part of the human body, such as the brain, the cervical spine, the lungs, the chest, the abdomen, etc.

[0035] In some embodiments, the scanning range of the CT machine cannot cover the scanned part of the target object, so that there is a lack of projection data of the X-rays emitted by the CT machine light source passing through the scanned part of the human body, which will affect the quality of the CT image converted from the first projection data obtained by scanning the target object by the CT machine, and thus the clinical diagnostic value of the CT image is low.

[0036] Step S120: Perform field of view expansion processing on the first projection data based on a field of view expansion algorithm to obtain an initial image.

[0037] In some embodiments, after obtaining the first projection data, the electronic device can obtain an edge value of the first projection data, and if it is determined that the edge value of the first projection data is greater than 0, it can be determined that the first projection data is truncated. It should be noted that, in the process of scanning the target object by the CT machine, because the object or the human body is placed in the center of the scanning range of the CT machine, the edge value of the normal projection data is close to zero because the air attenuation coefficient is 0. Accordingly, if it is determined that the edge value of the projection data is significantly greater than zero, it can be determined that the projection data is truncated. It can be understood that the truncated data is caused by the object being too large and exceeding the field of view of the CT machine. Accordingly, to improve the quality of the CT image, the CT field of view can be expanded for the projection data.

[0038] In some embodiments, after obtaining the first projection data, the electronic device can perform field of view expansion processing on the first projection data based on a field of view expansion algorithm to obtain an initial image. Optionally, the electronic device can be preconfigured with a field of view expansion algorithm, which can include a projection domain linear interpolation expansion algorithm, a random projection clustering algorithm, etc., without limitation. For example, the electronic device can perform field of view expansion processing on the first projection data based on the projection domain linear interpolation expansion algorithm, and convert the projection data after the field of view expansion processing to the image domain to obtain an initial image. In the process of performing field of view expansion processing on the first projection data based on the projection domain linear interpolation expansion algorithm, the electronic device can also perform FOV artifact removal processing on the first projection data, thereby obtaining an initial image with preliminary FOV artifacts removed, and improving the clarity of the expanded super field of view region image.

[0039] Step S130: performing bone and soft tissue segmentation processing on the initial image to obtain anatomical information corresponding to the initial image.

[0040] In some embodiments, after obtaining the initial image, the electronic device can perform bone and soft tissue segmentation processing on the initial image to obtain anatomical information corresponding to the initial image. The electronic device can be preconfigured with a segmentation algorithm, such as a threshold segmentation algorithm, a region growing algorithm, an edge detection algorithm, a horizontal line algorithm, a clustering algorithm, etc. For example, the electronic device can be preconfigured with a segmentation network, such as an FCN network, a SegNet network, a U-Net network, etc. Accordingly, the electronic device can perform bone and soft tissue segmentation processing on the initial image based on the segmentation algorithm to obtain anatomical information corresponding to the initial image. The anatomical information can be understood as a result of predicting the bone and soft tissue in the initial image based on the segmentation algorithm.

[0041] Step S140: converting the initial image and the anatomical information to a projection domain, and performing a projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data.

[0042] In some embodiments, after the electronic device obtains the anatomical information corresponding to the initial image, the electronic device can convert the initial image and the anatomical information to a projection domain, and perform a projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data. It can be understood that, after the electronic device converts the initial image and the anatomical information to the projection domain through orthographic projection, the electronic device guides the recovery of the projection data corresponding to the initial image according to the anatomical information obtained through segmentation processing, so as to introduce a segmentation mechanism to constrain image recovery, thereby reducing the error and deformation of the CT image with an expanded field of view.

[0043] Step S150: converting the second projection data to an image domain to obtain a target image.

[0044] In some embodiments, after the electronic device obtains the second projection data, the electronic device can convert the second projection data to an image domain to obtain a target image. Optionally, the electronic device can perform a filtered back projection on the second projection data to obtain the target image.

[0045] In some embodiments, the electronic device can be preconfigured with an image domain repair algorithm, such as an Fmm algorithm, a PatchMatch algorithm, a GAN network, etc. Accordingly, after the electronic device performs a filtered back projection on the second projection data, the electronic device can obtain a to-be-repaired image, and can perform a processing on the to-be-repaired image based on the image domain repair algorithm to obtain a target image, thereby improving the clarity and practicability of the CT image with an expanded field of view based on the image domain repair processing.

[0046] The target image can be understood as a CT image with an expanded field of view. In this embodiment, the electronic device can convert the projection data obtained by the CT machine scanning to an initial image with an expanded field of view, introduce a segmentation mechanism to extract anatomical information in the initial image, guide the information recovery of the initial image in the projection domain through orthographic projection of the anatomical information, and then convert the projection data after the projection data repair processing to an image domain, thereby jointly expanding the CT field of view in the projection domain and the image domain, improving the accuracy of the structure construction of the target object in the field of view expansion area, and improving the quality of the CT image.

[0047] The CT image field expansion method provided in an embodiment of the present application includes the following steps: obtaining first projection data obtained by scanning a target object by a CT machine; performing field expansion processing on the first projection data based on a field expansion algorithm to obtain an initial image, so as to preliminarily expand the CT field in the image domain; performing segmentation processing on the initial image to obtain anatomical information corresponding to the initial image, converting the initial image and the anatomical information to the projection domain, and performing projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data, so as to guide the recovery of the projection data based on the anatomical information in the projection domain; and converting the second projection data to the image domain to obtain a target image, so as to obtain a CT image with an expanded field, thereby expanding the CT field in the projection domain and the image domain jointly, introducing the segmentation mechanism to constrain the image recovery, and ensuring the quality of the CT image.

[0048] Referring to FIG. 3, FIG. 3 shows a flowchart of the CT image field expansion method provided in an embodiment of the present application. The method is applied to the electronic device described above, and the following will be described in detail with reference to the flowchart shown in FIG. 3. The CT image field expansion method can specifically include the following steps:

[0049] Step S210: Obtain first projection data obtained by scanning a target object by a CT machine.

[0050] For specific description of step S210, refer to the description of step S110 in the foregoing, which will not be described in detail here.

[0051] Step S220: Perform field expansion processing on the first projection data based on a linear interpolation algorithm to obtain the initial image.

[0052] In some embodiments, the electronic device can be preconfigured with the linear interpolation algorithm. Accordingly, after the electronic device obtains the first projection data, the electronic device can perform field expansion processing and reconstruction on the first projection data based on the linear interpolation algorithm to obtain the initial image. The electronic device can perform artifact elimination on the boundary and preliminary expansion of the CT field based on the linear interpolation algorithm, thereby expanding the CT field and improving the clarity of the CT image.

[0053] Step S230: Input the initial image into a target model, wherein the target model includes a segmentation network and a projection domain repair network.

[0054] In some embodiments, the electronic device can be preconfigured with the target model, wherein the target model can be composed of one or more sub-networks. The structures of the sub-networks included in the target model can be the same or different, which is not limited here. For example, the structures of the sub-networks included in the target model are the same and are all encoder-decoder structures.

[0055] In some embodiments, the target model comprises a sub-network including a segmentation network and a projection domain repair network; wherein the segmentation network can extract anatomical information of bones and soft tissues from the image, and the projection domain repair network can be used to repair the projection data. Optionally, the target model comprises a sub-network including a priori network; wherein the priori network can be used to process the image to remove visual field artifacts. Optionally, the target model comprises a sub-network including an image domain repair network; wherein the image domain repair network can be used to repair the image domain data. Accordingly, the target model can be composed of the priori network, the projection domain repair network, the image domain repair network, and the segmentation network. Accordingly, after the electronic device obtains the initial image, the initial image can be input into the target model, so as to obtain the target image output by the target model.

[0056] In some embodiments, please refer to FIG. 4, which shows a flowchart of the method for expanding the visual field of the CT image provided by the embodiments of the present application. Before performing step S230, the electronic device can further perform steps S231-S232, wherein:

[0057] Step S231: obtaining a training data set, wherein the training data set comprises a plurality of projection data and a plurality of truncated data, and the plurality of projection data correspond one-to-one to the plurality of truncated data.

[0058] In some embodiments, the electronic device can obtain a training data set; wherein the training data set can comprise a plurality of projection data and a plurality of truncated data, and the plurality of projection data correspond one-to-one to the plurality of truncated data.

[0059] Optionally, the electronic device can obtain the training data set from an associated cloud or an electronic device. Optionally, the electronic device can also pre-store the training data set in a designated area; based on this, the electronic device can obtain the training data set from the designated area. Optionally, the electronic device can also obtain a plurality of projection data from an associated cloud or an electronic device, and can obtain the training data set based on the plurality of projection data. Optionally, the electronic device can also obtain projection data obtained by scanning different objects by a CT machine, and can obtain stage data corresponding to the projection data based on the projection data, thereby obtaining the training data set. Wherein, the objects scanned by the CT machine can include human bodies, objects, etc., which are not limited here.

[0060] The projection data can be understood as non-truncated projection data, and the truncated data corresponding to the projection data can be understood as data obtained by truncating data at the edge of the projection data in the detector direction. Alternatively, the projection data can be obtained by a wide-view detector or can be obtained by simulation. For example, a complete image collected in a small FOV is enlarged to obtain an image in an ultra-wide view, and the projection data S is obtained by defining a wider detector and by orthographic projection. gt .

[0061] For example, referring to FIG. 5 and FIG. 6, FIG. 5 shows a schematic diagram of simulation of projection data according to an embodiment of the present application, and FIG. 6 shows a schematic diagram of projection data and truncated data corresponding to the projection data according to an embodiment of the present application. In the process of simulation of the projection data, the projection data can be obtained by a detector corresponding to a scanning view sFOV, or the projection data can be obtained by a wider detector collected in full. The projection data S gt is obtained by truncating the edge of the projection data S gt . FOV .

[0062] In step S232, the initial model is trained based on the training data set to obtain the target model. The initial model includes a segmentation network, a projection domain repair network, a prior network, and an image domain repair network. The segmentation network is used for segmentation processing of bones and soft tissues in an image. The projection domain repair network is used for repairing projection data. The prior network is used for field-of-view artifact removal processing of an image. The image domain repair network is used for repairing image domain data.

[0063] In some embodiments, after obtaining the training data set, the electronic device can train the initial model based on the training data set to obtain the target model. The initial model can include a segmentation network, a projection domain repair network, a prior network, and an image domain repair network. The segmentation network can be used for segmentation processing of bones and soft tissues in an image. The projection domain repair network can be used for repairing projection data. The prior network can be used for field-of-view artifact removal processing of an image. The image domain repair network can be used for repairing image domain data.

[0064] In some embodiments, after obtaining the training data set, the electronic device can convert the projection data and the truncated data corresponding one-to-one in the training data set to the image domain respectively to obtain a reconstructed image corresponding to the projection data and a reconstructed image corresponding to the truncated data. For example, referring to FIG. 7, FIG. 7 shows a schematic diagram of a reconstructed image corresponding to the projection data and a reconstructed image corresponding to the truncated data according to an embodiment of the present application. The reconstructed image I FOV corresponding to the projection data S FOVcorresponding reconstructed image, I gt for the projection data S gt corresponding reconstructed image.

[0065] Wherein, the electronic device can respectively perform filtered back-projection processing on the plurality of projection data and then convert to image domain reconstruction to obtain a plurality of sample images corresponding to the plurality of projection data respectively. Wherein, the projection data can be understood as the projection data without truncation, and correspondingly, the sample image corresponding to the projection data can be understood as the image data without field of view FOV missing.

[0066] In some embodiments, after the electronic device obtains the plurality of sample images corresponding to the plurality of projection data respectively, the electronic device can perform anatomical information labeling on the plurality of sample images corresponding to the plurality of projection data respectively to obtain a plurality of segmentation labels corresponding to the plurality of projection data respectively. Wherein, the segmentation label can at least include the segmentation label of bone and soft tissue.

[0067] In some embodiments, after the electronic device obtains the plurality of sample images corresponding to the plurality of projection data respectively and the plurality of segmentation labels corresponding to the plurality of projection data respectively, the electronic device can train an initial model according to the plurality of projection data, the plurality of truncated data corresponding to the plurality of projection data, the plurality of sample images corresponding to the plurality of projection data, and the plurality of segmentation labels corresponding to the plurality of projection data to obtain a target model.

[0068] In some embodiments, the electronic device can respectively input the plurality of truncated data into the initial model, respectively perform the view field artifact removal processing on the plurality of truncated data through the prior network, obtain the third image corresponding to each of the plurality of truncated data, respectively perform the bone and soft tissue segmentation processing on the third image corresponding to each of the plurality of truncated data through the segmentation network, obtain the anatomical information corresponding to each of the plurality of truncated data, convert the third image corresponding to each of the plurality of truncated data and the anatomical information corresponding to each of the plurality of truncated data to the projection domain to obtain the fourth projection data corresponding to each of the plurality of truncated data, perform the projection data repair processing on the fourth projection data corresponding to each of the plurality of truncated data through the projection domain repair network to obtain the fifth projection data corresponding to each of the plurality of truncated data, convert the fifth projection data corresponding to each of the plurality of truncated data to the image domain to obtain the fourth image corresponding to each of the plurality of truncated data, perform the image domain repair processing on the fourth image corresponding to each of the plurality of truncated data through the image domain repair network to obtain the fifth image corresponding to each of the plurality of truncated data output by the image domain repair network, and obtain the target loss value corresponding to the loss function of the initial model according to the third image corresponding to each of the plurality of truncated data, the sample image corresponding to each of the plurality of truncated data, the fifth projection data corresponding to each of the plurality of truncated data, the projection data corresponding to each of the plurality of truncated data, the fifth image corresponding to each of the plurality of truncated data, the segmentation label corresponding to each of the plurality of truncated data, and the anatomical information corresponding to each of the plurality of truncated data. If it is determined that the loss function does not converge according to the target loss value, the parameters of the initial model can be adjusted until the loss function converges, and the target model is obtained.

[0069] Exemplarily, refer to FIG. 8, which shows a flowchart of a CT image view field expansion method provided by an embodiment of the present application. The initial model can be composed of a prior network (Prior-Net), a projection domain repair network (SR-Net), an image domain repair network (IR-Net), and a segmentation network (Segmentor). The truncated data S FOV The corresponding reconstructed image I FOV The initial model is input, and the prior network is used to process the reconstructed image I FOV The view field artifact removal processing is performed to obtain the truncated data S FOV The corresponding third image I prior The segmentation network is used to perform the bone and soft tissue segmentation processing on the third image I prior The corresponding third image I FOV The anatomical information y, and the third image I prior and the anatomical information y can be converted to the projection domain through forward projection (FP) to obtain the fourth projection data corresponding to the truncated data S FOV The projection data repair processing is performed on the fourth projection data through the projection domain repair network to obtain the fifth projection data corresponding to the truncated data SFOV corresponding fifth projection data S R , and the fifth projection data S R converted to the image domain by a filtered back projection (FBP) process to obtain truncated data S FOV corresponding fourth image, and the fourth image can be subjected to an image domain inpainting process by an image domain inpainting network to obtain truncated data S FOV corresponding fifth image I out .

[0070] In some embodiments, in the process of obtaining the target loss value corresponding to the initial model according to the third image corresponding to each of the plurality of truncated data, the sample image corresponding to each of the plurality of truncated data, the fifth projection data corresponding to each of the plurality of truncated data, the projection data corresponding to each of the plurality of truncated data, the fifth image corresponding to each of the plurality of truncated data, the segmentation label corresponding to each of the plurality of truncated data, and the anatomical information corresponding to each of the plurality of truncated data, a first loss value corresponding to each of the plurality of truncated data can be obtained based on the third image corresponding to each of the plurality of truncated data and the sample image corresponding to each of the plurality of truncated data; a second loss value corresponding to each of the plurality of truncated data can be obtained based on the fifth projection data corresponding to each of the plurality of truncated data and the projection data corresponding to each of the plurality of truncated data; a third loss value corresponding to each of the plurality of truncated data can be obtained based on the fifth image corresponding to each of the plurality of truncated data and the third image corresponding to each of the plurality of truncated data; a fourth loss value corresponding to each of the plurality of truncated data can be obtained based on the anatomical information corresponding to each of the plurality of truncated data and the segmentation label corresponding to each of the plurality of truncated data; an image gradient of the fifth projection data corresponding to each of the plurality of truncated data can be obtained as a first image gradient corresponding to each of the plurality of truncated data, an image gradient of the projection data corresponding to each of the plurality of truncated data can be obtained as a second image gradient corresponding to each of the plurality of truncated data, and a fifth loss value corresponding to each of the plurality of truncated data can be obtained based on the first image gradient corresponding to each of the plurality of truncated data and the second image gradient corresponding to each of the plurality of truncated data; and the target loss value corresponding to the loss function of the initial model is obtained according to the first loss value corresponding to each of the plurality of truncated data, the second loss value corresponding to each of the plurality of truncated data, the third loss value corresponding to each of the plurality of truncated data, the fourth loss value corresponding to each of the plurality of truncated data, and the fifth loss value corresponding to each of the plurality of truncated data.

[0071] The electronic device can determine a first loss value based on the similarity between the third image corresponding to the truncated data and the sample image corresponding to the truncated data; the electronic device can determine a second loss value based on the similarity between the fifth projection data corresponding to the truncated data and the projection data corresponding to the truncated data; the electronic device can determine a third loss value based on the similarity between the fifth image corresponding to the truncated data and the third image corresponding to the truncated data; the electronic device can determine a fourth loss value based on the similarity between the anatomical information corresponding to the truncated data and the segmentation label corresponding to the truncated data; and the electronic device can determine a fifth loss value based on the similarity between the first image gradient corresponding to the truncated data and the second image gradient corresponding to the truncated data.

[0072] Optionally, the electronic device can determine the sum of the first loss value, the second loss value, the third loss value, the fourth loss value, and the fifth loss value as the target loss value, or can determine the maximum value, the minimum value, the median value, etc. of the first loss value, the second loss value, the third loss value, the fourth loss value, and the fifth loss value as the target loss value. Optionally, the electronic device can be pre-configured with a weight factor corresponding to each of the first loss value, the second loss value, the third loss value, the fourth loss value, and the fifth loss value, and accordingly, the electronic device can determine the target loss value based on the first loss value, the second loss value, the third loss value, the fourth loss value, and the fifth loss value and the weight factor corresponding to each of the first loss value, the second loss value, the third loss value, the fourth loss value, and the fifth loss value.

[0073] For example, the electronic device determines the truncated data S FOV The corresponding reconstructed image I FOV The initial model is input to obtain the truncated data S FOV The corresponding third image I prior The truncated data S FOV The corresponding anatomical information y, the truncated data S FOV The corresponding fifth projection data S R The truncated data S FOV The corresponding fifth image I out Accordingly, the truncated data S FOV The corresponding sample image I gt The truncated data S FOV The corresponding projection data S gt The truncated data S FOV The corresponding segmentation label y gt The electronic device is configured with a weight factor a1 of the first loss value, a weight factor a2 of the second loss value, a weight factor a3 of the third loss value, a weight factor a4 of the fourth loss value, and a weight factor a5 of the fifth loss value. Accordingly, the truncated data S FOV The corresponding fifth projection data SR Image gradient as truncated data S FOV The corresponding first image gradient Get truncated data S FOV Corresponding projection data S gt Image gradient as truncated data S FOV The corresponding second image gradient Among them, the operation of image gradient It can include Sobel operators, Prewitt operators, Canny operators, Scharr operators, etc., without limitation.

[0074] Accordingly, the electronic device can obtain the target loss value based on the network training loss function. The loss function of the initial model can be:

[0075] Where L can represent the target loss value, ||I prior -I gt ||2 can represent the first loss value, ||S R -S gt ||2 can represent the second loss value, ||I out -I gt ||2 can represent the third loss value, Dice(y, y gt This can characterize the fourth loss value. This can represent the fifth loss value. Where, Dice(y, y) gt () can be the loss function used in the segmentation network, or it can be... Here, σ can be a minimum value to prevent division by zero.

[0076] In some implementations, the process of training an initial model using an electronic device can be understood as continuously minimizing the target loss value of the loss function, or as adjusting the parameters of the initial model until the loss function converges. Specifically, adjusting the model weight parameters through gradient backpropagation error can gradually reduce the loss value. Therefore, the process of minimizing the loss function using an electronic device can include: a pre-set loss threshold in the electronic device; correspondingly, after obtaining the target loss value, the electronic device can compare the target loss value with the loss threshold; if the target loss value is greater than the loss threshold, the parameters of the initial model can be adjusted until the target loss value is less than or equal to the loss threshold, thus obtaining the target model. Optionally, the process of minimizing the loss function using an electronic device can also include: a pre-set number of training rounds and a preset range in the electronic device; correspondingly, the electronic device can determine that the loss function has converged and obtained the target model after adjusting the parameters of the initial model to the required number of training rounds and the target loss value of the loss function has decreased to the preset range.

[0077] In the process of adjusting the parameters of the initial model by the electronic device, the parameters of the prior network in the initial model, and / or the parameters of the projection domain restoration network, and / or the parameters of the image domain restoration network, and / or the parameters of the segmentor can be adjusted until it is determined that the loss function of the initial model converges, and the target model is obtained.

[0078] Step S240: performing bone and soft tissue segmentation processing on the initial image by the segmentor to obtain the anatomical information.

[0079] In some embodiments, after the initial image is input into the target model, the electronic device can perform bone and soft tissue segmentation processing on the initial image by the segmentor in the target model to obtain the anatomical information. The anatomical information can include the prediction results of the bones and soft tissues included in the initial image.

[0080] As an implementable manner, the target model can further include a prior network. Accordingly, the electronic device can perform the view field artifact removal processing on the initial image by the prior network to obtain a first image, and can perform bone and soft tissue segmentation processing on the first image by the segmentor to obtain the anatomical information. Thus, the image is subjected to artifact removal processing after the FOV is preliminarily expanded, the image clarity is improved, and the anatomical information in the image after the FOV is preliminarily expanded is extracted to constrain the image restoration, thereby improving the structural accuracy of the target object recovered beyond the scanning view field.

[0081] Step S250: converting the initial image and the anatomical information to the projection domain to obtain third projection data.

[0082] In some embodiments, the target model can further include a projection operator. Accordingly, after the target model obtains the anatomical information corresponding to the initial image, the initial image and the anatomical information can be orthogonally projected to the projection domain by the projection operator included in the target model to obtain the third projection data.

[0083] Step S260: performing projection data restoration processing on the third projection data by the projection domain restoration network to obtain the second projection data.

[0084] In some embodiments, after the target model obtains the third projection data corresponding to the initial image, the projection data restoration processing can be performed on the third projection data by the projection domain restoration network in the target model to obtain the second projection data.

[0085] Step S270: converting the second projection data to the image domain to obtain a target image.

[0086] In some implementations, the target model may also include filtering operators and backprojection operators; accordingly, after the target model obtains the second projection data corresponding to the initial image, it can perform filtering and backprojection processing on the second projection data through the filtering operators and backprojection operators included in the target model to obtain the target image.

[0087] In some implementations, the target model may also include an image domain inpainting network. Accordingly, the electronic device can convert the second projection data into the image domain after filtering and backprojection to obtain a second image, and perform image domain inpainting processing on the second image through the image domain inpainting network to obtain the target image output by the image domain inpainting network, thereby expanding the CT field of view and ensuring the quality of the CT image with expanded field of view.

[0088] For example, please refer to Figure 9, which illustrates a flowchart of a CT image field-of-view expansion method provided in an embodiment of this application. The target model may include a Prior-Net, a Projection Domain Inpainting Network (SR-Net), an Image Domain Inpainting Network (IR-Net), and a Segmentation Network. The electronic device can acquire first projection data S obtained by scanning the target object using a CT scanner. FOV Wherein, if the first projection data S FOV If the edge values ​​are significantly greater than 0, then the first projection data S can be determined. FOV To truncate the data. For example, if the first projection data S... FOV If the difference between the edge value and 0 is greater than the preset deviation, then the first projection data S can be determined. FOV To truncate the data; the preset deviation can be pre-set in the electronic device, can be set by the user, or can be obtained from third-party experimental data, which is not limited here.

[0089] Accordingly, the electronic device can use a field-of-view expansion algorithm to process the first projection data S. FOV Preprocessing is performed to obtain the initial image I0. This preprocessing may include field-of-view expansion processing; optionally, the electronic device may initially expand the first projection data S based on linear interpolation of the projection domain. FOV The field of view, and the first projection data S after expanding the field of view. FOV Artifact removal and reconstruction are performed to obtain the initial image I0 in the image domain. Specifically, basic FOV expansion is performed on the input target model image, improving the performance of CT field of view expansion for the target model.

[0090] The electronic device can input the initial image view I0 into the target model, and can further process the initial image view I0 to remove view artifacts through a prior network to obtain the first image I. prior And the first image I can be segmented using a segmentation network.prior The segmentation processing of the bone and soft tissue is performed to obtain the anatomical information y. The prior network is added before the dual-domain network to perform the image domain processing, and the performance of the target model is improved.

[0091] The projection operator can also be included in the target model, and the projection operator in the target model can also convert the first image I prior to the projection domain through forward projection (FP) to obtain third projection data, and the projection data repair network can be used to perform projection data repair processing on the third projection data to obtain second projection data S R . In addition, the model of anatomical structure segmentation and image domain recovery guided by segmentation information is added to the target model, and the performance of the CT field of view expansion is improved.

[0092] The filtering and back-projection operator can also be included in the target model, and the filtering and back-projection operator in the target model can perform filtering and back-projection processing (FBP) on the second projection data S R to convert to the image domain to obtain the second image, and the image domain repair network can be used to perform image domain repair processing on the second image to obtain the target image I out expanded by the image domain repair network.

[0093] It can be understood that in the embodiment, the performance of the dual-domain network of the target model with the projection domain first and then the image domain is better than that of the single-domain network; the initial image input into the target model is the de-FOV artifact image after linear interpolation, the network architecture of the target model is a dual-domain network framework of image domain-projection domain-image domain cascade, the FOV is expanded in the projection domain and the image domain, and the segmentation mechanism is introduced to further constrain the image recovery, thereby improving the structural accuracy of the target object recovered beyond the scanning field of view, improving the clarity of the image through the de-artifact processing, expanding the field of view, and guiding the projection data recovery through the anatomical structure information obtained by segmentation through forward projection, thereby reducing the reconstruction error and deformation of the structure of the target object in the expanded field of view, improving the diagnostic value of the CT image, and improving the user experience.

[0094] Compared with the CT image field of view expansion method shown in FIG. 2, the CT image field of view expansion method provided by the embodiment can further perform field of view expansion processing on the first projection data based on a linear interpolation algorithm to obtain an initial image, thereby preliminarily expanding the CT field of view, and can input the initial image into a target model, wherein the target model includes a segmentation network and a projection domain repair network. The segmentation network is used for performing segmentation processing on the initial image to obtain anatomical information, the initial image and the anatomical information are converted to a projection domain to obtain third projection data, the projection domain repair network is used for performing projection data repair processing on the third projection data to obtain second projection data, and the second projection data is converted to an image domain to obtain a target image. Therefore, the target model using the network architecture of the dual domain is used to jointly expand the FOV in the projection domain and the image domain, and the anatomical structure information obtained by segmentation is used to guide the recovery of the projection data, so that the quality of the CT image is ensured, the diagnostic value of the CT image is improved, and the experience of the user is improved.

[0095] Please refer to FIG. 10, which shows a module block diagram of the CT image field of view expansion device provided by the embodiment of the present application. The CT image field of view expansion device 200 is applied to the electronic device described above. The CT image field of view expansion device 200 shown in FIG. 10 will be described in detail below. The CT image field of view expansion device 200 can include a first projection data acquisition module 210, a field of view expansion processing module 220, an anatomical information obtaining module 230, a projection domain repair module 240, and a CT image obtaining module 250, wherein:

[0096] The first projection data acquisition module 210 is configured to acquire first projection data obtained by scanning a target object by a CT machine.

[0097] The field of view expansion processing module 220 is configured to perform field of view expansion processing on the first projection data based on a field of view expansion algorithm to obtain an initial image.

[0098] The anatomical information obtaining module 230 is configured to perform segmentation processing on the initial image to obtain anatomical information corresponding to the initial image.

[0099] The projection domain repair module 240 is configured to convert the initial image and the anatomical information to a projection domain, and perform projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data.

[0100] The CT image obtaining module 250 is configured to convert the second projection data to an image domain to obtain a target image.

[0101] Further, the CT image field of view expansion device 200 can further include an image input model unit, an image segmentation unit, a projection domain conversion unit, and a projection data repair unit, wherein:

[0102] The image input model unit is configured to input the initial image into a target model, wherein the target model includes a segmentation network and a projection domain repair network.

[0103] The image segmentation unit is configured to perform bone and soft tissue segmentation processing on the initial image by using the segmentation network to obtain the anatomical information.

[0104] The projection domain conversion unit is configured to convert the initial image and the anatomical information into a projection domain to obtain third projection data.

[0105] The projection data repair unit is configured to perform projection data repair processing on the third projection data by using the projection domain repair network to obtain the second projection data.

[0106] Further, the target model further includes a prior network, and the image segmentation unit can include a de-artifact unit and an image segmentation sub-unit, wherein:

[0107] The de-artifact unit is configured to perform de-field-of-view artifact processing on the initial image by using the prior network to obtain a first image.

[0108] The image segmentation sub-unit is configured to perform bone and soft tissue segmentation processing on the first image by using the segmentation network to obtain the anatomical information.

[0109] Further, the target model further includes an image domain repair network, and the CT image obtaining module 250 can include an image domain conversion unit and an image domain repair unit, wherein:

[0110] The image domain conversion unit is configured to convert the second projection data into an image domain to obtain a second image.

[0111] The image domain repair unit is configured to perform image domain repair processing on the second image by using the image domain repair network to obtain the target image output by the image domain repair network.

[0112] Further, before the initial image is input into the target model, the CT image field of view expansion device 200 can further include a training data set acquisition unit and a model training unit, wherein:

[0113] The training data set acquisition unit is configured to acquire a training data set, wherein the training data set includes a plurality of projection data and a plurality of truncated data, and the plurality of projection data and the plurality of truncated data correspond one-to-one.

[0114] a model training unit, configured to train an initial model based on the training data set to obtain the target model, wherein the initial model comprises a segmentation network, a projection domain repair network, a prior network, and an image domain repair network, the segmentation network is configured to perform segmentation processing on bones and soft tissues in an image, the projection domain repair network is configured to repair projection data, the prior network is configured to perform field-of-view artifact removal processing on an image, and the image domain repair network is configured to repair image domain data.

[0115] Further, the model training unit can comprise a sample image acquisition unit, a segmentation label acquisition unit, and a model training subunit, wherein:

[0116] The sample image acquisition unit is configured to perform filtered back-projection processing on the plurality of projection data respectively to obtain sample images corresponding to the plurality of projection data respectively.

[0117] The segmentation label acquisition unit is configured to perform anatomical information labeling on the sample images corresponding to the plurality of projection data respectively to obtain segmentation labels corresponding to the plurality of projection data respectively, wherein the segmentation labels comprise at least labels of bones and soft tissues.

[0118] The model training subunit is configured to train the initial model according to the plurality of projection data, the truncated data corresponding to the plurality of projection data, the sample images corresponding to the plurality of projection data, and the segmentation labels corresponding to the plurality of projection data to obtain the target model.

[0119] Further, the model training subunit can comprise a truncated data input unit, a third image acquisition unit, a truncated data corresponding anatomical information acquisition unit, a fourth projection data acquisition unit, a fifth projection data acquisition unit, a fourth image acquisition unit, a fifth image acquisition unit, a target loss value acquisition unit, and a target model acquisition unit, wherein:

[0120] The truncated data input unit is configured to input the plurality of truncated data into the initial model respectively.

[0121] The third image acquisition unit is configured to perform field-of-view artifact removal processing on the plurality of truncated data by the prior network to obtain third images corresponding to the plurality of truncated data respectively.

[0122] The truncated data corresponding anatomical information acquisition unit is configured to perform segmentation processing on bones and soft tissues on the third images corresponding to the plurality of truncated data respectively by the segmentation network to obtain anatomical information corresponding to the plurality of truncated data respectively.

[0123] A fourth projection data obtaining unit is configured to convert the third image corresponding to each of the plurality of truncated data and the anatomical information corresponding to each of the plurality of truncated data to a projection domain to obtain fourth projection data corresponding to each of the plurality of truncated data.

[0124] A fifth projection data obtaining unit is configured to perform projection data repairing processing on the fourth projection data corresponding to each of the plurality of truncated data by using the projection domain repairing network to obtain fifth projection data corresponding to each of the plurality of truncated data.

[0125] A fourth image obtaining unit is configured to convert the fifth projection data corresponding to each of the plurality of truncated data to an image domain to obtain a fourth image corresponding to each of the plurality of truncated data.

[0126] A fifth image obtaining unit is configured to perform image domain repairing processing on the fourth image corresponding to each of the plurality of truncated data by using the image domain repairing network to obtain fifth image output by the image domain repairing network corresponding to each of the plurality of truncated data.

[0127] A target loss value obtaining unit is configured to obtain a target loss value corresponding to a loss function of the initial model according to the third image corresponding to each of the plurality of truncated data, the sample image corresponding to each of the plurality of truncated data, the fifth projection data corresponding to each of the plurality of truncated data, the projection data corresponding to each of the plurality of truncated data, the fifth image corresponding to each of the plurality of truncated data, the segmentation label corresponding to each of the plurality of truncated data, and the anatomical information corresponding to each of the plurality of truncated data.

[0128] A target model obtaining unit is configured to adjust parameters of the initial model until the loss function converges if it is determined that the loss function does not converge according to the target loss value, and obtain a target model.

[0129] Further, the target loss value obtaining unit can include a first loss value obtaining unit, a second loss value obtaining unit, a third loss value obtaining unit, a fourth loss value obtaining unit, a first image gradient obtaining unit, a second image gradient obtaining unit, a fifth loss value obtaining unit, and a target loss value obtaining subunit, wherein:

[0130] The first loss value obtaining unit is configured to obtain a first loss value corresponding to each of the plurality of truncated data based on the third image corresponding to each of the plurality of truncated data and the sample image corresponding to each of the plurality of truncated data.

[0131] The second loss value obtaining unit is configured to obtain a second loss value corresponding to each of the plurality of truncated data based on the fifth projection data corresponding to each of the plurality of truncated data and the projection data corresponding to each of the plurality of truncated data.

[0132] The third loss value obtaining unit is configured to obtain a third loss value corresponding to each of the plurality of truncated data based on the fifth image corresponding to each of the plurality of truncated data and the third image corresponding to each of the plurality of truncated data.

[0133] The fourth loss value obtaining unit is configured to obtain a fourth loss value corresponding to each of the plurality of truncated data based on the anatomical information corresponding to each of the plurality of truncated data and the segmentation label corresponding to each of the plurality of truncated data.

[0134] The first image gradient obtaining unit is configured to obtain an image gradient of the fifth projection data corresponding to each of the plurality of truncated data as a first image gradient corresponding to each of the plurality of truncated data.

[0135] The second image gradient obtaining unit is configured to obtain an image gradient of the projection data corresponding to each of the plurality of truncated data as a second image gradient corresponding to each of the plurality of truncated data.

[0136] The fifth loss value obtaining unit is configured to obtain a fifth loss value corresponding to each of the plurality of truncated data based on the first image gradient corresponding to each of the plurality of truncated data and the second image gradient corresponding to each of the plurality of truncated data.

[0137] The target loss value obtaining sub-unit is configured to obtain the target loss value corresponding to the loss function of the initial model according to the first loss value corresponding to each of the plurality of truncated data, the second loss value corresponding to each of the plurality of truncated data, the third loss value corresponding to each of the plurality of truncated data, the fourth loss value corresponding to each of the plurality of truncated data and the fifth loss value corresponding to each of the plurality of truncated data.

[0138] Further, the field of view expansion processing module 220 includes a field of view expansion processing sub-unit, and the field of view expansion processing sub-unit includes:

[0139] The field of view expansion processing sub-unit is configured to perform field of view expansion processing on the first projection data based on a linear interpolation algorithm to obtain the initial image.

[0140] Those skilled in the art can clearly understand the specific working process of the above-described devices and modules for the convenience and brevity of description, which can refer to the corresponding process in the foregoing method embodiments, and will not be described here.

[0141] In several embodiments provided in the present application, the coupling between the modules can be electrical, mechanical or other forms of coupling.

[0142] In addition, each function module in each embodiment of the present application can be integrated in one processing module, or each module can be physically present alone, or two or more modules can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software function module.

[0143] Referring to FIG. 11, a structural block diagram of an electronic device 100 is shown. The electronic device 100 can be a vehicle, a tablet computer, a robot, or the like, which has processing capability. The electronic device 100 in the present application can include one or more of the following components: a processor 110, a memory 120, and one or more application programs, wherein the one or more application programs can be stored in the memory 120 and configured to be executed by the one or more processors 110, and the one or more programs are configured to perform the method as described in the foregoing method embodiments.

[0144] The processor 110 can include one or more processing cores. The processor 110 connects various parts in the entire electronic device 100 through various interfaces and lines, performs various functions of the electronic device 100 and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 120, and calling data stored in the memory 120. Alternatively, the processor 110 can be realized in at least one of the hardware forms of a digital signal processor (DSP), a field-programmable gate array (FPGA), and a programmable logic array (PLA). The processor 110 can integrate a combination of one or several of a central processing unit (CPU), a graphics processing unit (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, and application programs, etc.; the GPU is responsible for rendering and drawing the content to be displayed; and the modem is used for processing wireless communication. It can be understood that the above-mentioned modem can also not be integrated into the processor 110, but can be realized by a separate communication chip.

[0145] The memory 120 can include a random access memory (RAM) and can also include a read-only memory (ROM). The memory 120 can be used to store instructions, programs, codes, code sets, or instruction sets. The memory 120 can include a program storage area and a data storage area, where the program storage area can store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing each of the method embodiments described below, and the like. The data storage area can also store data created by the electronic device 100 in use (such as a phone book, audio and video data, chat record data, etc.).

[0146] Referring to FIG. 12, a structural block diagram of a computer readable storage medium according to an embodiment of the present application is shown. The computer readable storage medium 300 stores program codes therein, which can be invoked by a processor to execute the methods described in the above method embodiments.

[0147] The computer readable storage medium 300 can be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk, or a ROM. Alternatively, the computer readable storage medium 300 includes a non-transitory computer readable medium. The computer readable storage medium 300 has a storage space for program codes 310 for executing any of the method steps described above. These program codes can be read from or written to one or more computer program products. The program codes 310 can be compressed in an appropriate form, for example.

[0148] Finally, it should be noted that the above embodiments are merely used to illustrate the technical solutions of the present application, rather than limit the same; even though the above-mentioned embodiments are described in detail, those skilled in the art should understand that the technical solutions recorded in the above-mentioned embodiments can be modified, or some technical features can be replaced equivalently; and these modifications or replacements do not cause the technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of field of view extension of a CT image, characterized by, The method comprises: obtaining first projection data obtained by scanning a target object by a CT machine; performing field-of-view expansion processing on the first projection data based on a field-of-view expansion algorithm to obtain an initial image; performing segmentation processing of bone and soft tissue on the initial image to obtain anatomical information corresponding to the initial image; converting the initial image and the anatomical information to a projection domain, and performing projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data; converting the second projection data to an image domain to obtain a target image.

2. The method of claim 1, wherein, The segmentation processing of bone and soft tissue on the initial image to obtain anatomical information corresponding to the initial image, the conversion of the initial image and the anatomical information to a projection domain, and the projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data comprise: inputting the initial image into a target model, wherein the target model comprises a segmentation network and a projection domain repair network; performing segmentation processing of bone and soft tissue on the initial image by the segmentation network to obtain the anatomical information; converting the initial image and the anatomical information to a projection domain to obtain third projection data; performing projection data repair processing on the third projection data by the projection domain repair network to obtain the second projection data.

3. The method of claim 2, wherein, The target model further comprises a prior network, and the segmentation processing of bone and soft tissue on the initial image by the segmentation network to obtain the anatomical information comprises: performing field-of-view artifact removal processing on the initial image by the prior network to obtain a first image; performing segmentation processing of bone and soft tissue on the first image by the segmentation network to obtain the anatomical information.

4. The method of claim 3, wherein, The target model further comprises an image domain repair network, and the conversion of the second projection data to an image domain to obtain a target image comprises: converting the second projection data to an image domain to obtain a second image; performing image domain repair processing on the second image by the image domain repair network to obtain the target image output by the image domain repair network.

5. The method of claim 4, wherein, Before the initial image is input into the target model, the method further comprises: obtaining a training data set, wherein the training data set comprises a plurality of projection data and a plurality of truncated data, and the plurality of projection data correspond one-to-one to the plurality of truncated data; training an initial model based on the training data set to obtain the target model, wherein the initial model comprises a segmentation network, a projection domain repair network, a prior network, and an image domain repair network, the segmentation network is used for segmentation processing of bone and soft tissue in an image, the projection domain repair network is used for repairing projection data, the prior network is used for field-of-view artifact removal processing on an image, and the image domain repair network is used for repairing image domain data.

6. The method of claim 5, wherein, The training of the initial model based on the training data set to obtain the target model comprises: performing filter back projection processing on the plurality of projection data respectively to obtain sample images corresponding to the plurality of projection data respectively; Anatomical information is labeled on each sample image corresponding to the plurality of projection data, and a segmentation label corresponding to each of the plurality of projection data is obtained, wherein the segmentation label at least includes labels of bones and soft tissues; The initial model is trained according to the plurality of projection data, the plurality of truncated data corresponding to each of the plurality of projection data, the plurality of sample images corresponding to each of the plurality of projection data, and the plurality of segmentation labels corresponding to each of the plurality of projection data, and the target model is obtained.

7. The method of claim 6, wherein, The initial model is trained according to the plurality of projection data, the plurality of truncated data corresponding to each of the plurality of projection data, the plurality of sample images corresponding to each of the plurality of projection data, and the plurality of segmentation labels corresponding to each of the plurality of projection data, and the target model is obtained. The plurality of truncated data is respectively input into the initial model; The plurality of truncated data is respectively processed by the prior network to obtain a third image corresponding to each of the plurality of truncated data; The third image corresponding to each of the plurality of truncated data is segmented by the segmentation network to obtain anatomical information corresponding to each of the plurality of truncated data; The third image corresponding to each of the plurality of truncated data and the anatomical information corresponding to each of the plurality of truncated data are converted to the projection domain to obtain fourth projection data corresponding to each of the plurality of truncated data; The fourth projection data corresponding to each of the plurality of truncated data is processed by the projection domain repair network to obtain fifth projection data corresponding to each of the plurality of truncated data; The fifth projection data corresponding to each of the plurality of truncated data is converted to the image domain to obtain a fourth image corresponding to each of the plurality of truncated data; The fourth image corresponding to each of the plurality of truncated data is processed by the image domain repair network to obtain fifth images corresponding to each of the plurality of truncated data output by the image domain repair network; A target loss value corresponding to a loss function of the initial model is obtained according to the third image corresponding to each of the plurality of truncated data, the sample image corresponding to each of the plurality of truncated data, the fifth projection data corresponding to each of the plurality of truncated data, the projection data corresponding to each of the plurality of truncated data, the fifth image corresponding to each of the plurality of truncated data, the segmentation label corresponding to each of the plurality of truncated data, and the anatomical information corresponding to each of the plurality of truncated data; If it is determined according to the target loss value that the loss function does not converge, the parameters of the initial model are adjusted until the loss function converges, and the target model is obtained.

8. The method of claim 7, wherein, A target loss value corresponding to a loss function of the initial model is obtained according to the third image corresponding to each of the plurality of truncated data, the sample image corresponding to each of the plurality of truncated data, the fifth projection data corresponding to each of the plurality of truncated data, the projection data corresponding to each of the plurality of truncated data, the fifth image corresponding to each of the plurality of truncated data, the segmentation label corresponding to each of the plurality of truncated data, and the anatomical information corresponding to each of the plurality of truncated data, obtaining a first loss value corresponding to each of the plurality of truncated data based on the third image corresponding to each of the plurality of truncated data and the sample image corresponding to each of the plurality of truncated data; obtaining a second loss value corresponding to each of the plurality of truncated data based on the fifth projection data corresponding to each of the plurality of truncated data and the projection data corresponding to each of the plurality of truncated data; obtaining a third loss value corresponding to each of the plurality of truncated data based on the fifth image corresponding to each of the plurality of truncated data and the third image corresponding to each of the plurality of truncated data; obtaining a fourth loss value corresponding to each of the plurality of truncated data based on the anatomical information corresponding to each of the plurality of truncated data and the segmentation label corresponding to each of the plurality of truncated data; obtaining a first image gradient corresponding to each of the plurality of truncated data as the image gradient of the fifth projection data corresponding to each of the plurality of truncated data; obtaining a second image gradient corresponding to each of the plurality of truncated data as the image gradient of the projection data corresponding to each of the plurality of truncated data; obtaining a fifth loss value corresponding to each of the plurality of truncated data based on the first image gradient corresponding to each of the plurality of truncated data and the second image gradient corresponding to each of the plurality of truncated data; obtaining the target loss value corresponding to the loss function of the initial model according to the first loss value corresponding to each of the plurality of truncated data, the second loss value corresponding to each of the plurality of truncated data, the third loss value corresponding to each of the plurality of truncated data, the fourth loss value corresponding to each of the plurality of truncated data, and the fifth loss value corresponding to each of the plurality of truncated data.

9. The method according to any one of claims 1 to 8, characterized in that, The field of view expansion algorithm is used to perform field of view expansion processing on the first projection data to obtain an initial image, including: The linear interpolation algorithm is used to perform field of view expansion processing on the first projection data to obtain the initial image.

10. A field of view extension apparatus for CT images, characterized by, The device includes: A first projection data acquisition module is configured to acquire first projection data obtained by scanning a target object by a CT machine. A field of view expansion processing module is configured to perform field of view expansion processing on the first projection data based on a field of view expansion algorithm to obtain an initial image. An anatomical information obtaining module is configured to perform segmentation processing of bones and soft tissues on the initial image to obtain anatomical information corresponding to the initial image. A projection domain repair module is configured to convert the initial image and the anatomical information to a projection domain, and perform projection data repair processing on the initial image based on the anatomical information in the projection domain to obtain second projection data. A CT image obtaining module is configured to convert the second projection data to an image domain to obtain a target image.

11. An electronic device, comprising: One or more processors; Memory; One or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs are configured to perform the method of any one of claims 1-9. The computer readable storage medium stores program code, which can be called and executed by the processor to perform the method of any one of claims 1-9.

12. A computer readable storage medium, characterized in that, ​

Images