Computed tomography image processing method and system

Abstract

The invention provides a computed tomography image processing method and system, and the method comprises the steps: carrying out dual-energy CT scanning and enhanced CT scanning on the same body position of the same patient, so as to correspondingly obtain a dual-energy CT image and an enhanced CT image; generating a corresponding CT image pair according to the dual-energy CT image and the enhanced CT image based on a preset rule, and generating a corresponding target training set according to the CT image pair; performing model training on a preset neural network through the target training set to generate a corresponding target image processing model; a three-energy-level image is generated according to the target dual-energy CT image of the patient, an imitated and enhanced CT image corresponding to the patient is generated in real time according to the three-energy-level image through a target image processing model, and the imitated and enhanced CT image has uniqueness. According to the method, the required simulated and enhanced CT image can be quickly and effectively obtained, and the working efficiency is correspondingly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a computer tomography image processing method and system. Background Art

[0002] With the development of the times and the advancement of science and technology, Computed Tomography (CT) technology has been widely used in the medical field. It uses a rotating X-ray tube and detectors arranged in a rack to measure the absorption of X-rays by different tissues in the body. After computer processing, it can generate cross-sectional images of the body, which can be used by doctors for diagnosis and treatment planning.

[0003] Dual-energy CT and enhanced CT are two common methods of computed tomography. Dual-energy CT utilizes two different X-ray energies to scan the human body, acquiring rich tissue information and accurate data. Enhanced CT, on the other hand, uses intravenous contrast agents to more clearly visualize blood vessels, organs, and lesions, improving diagnostic accuracy.

[0004] However, in the actual application of enhanced CT, since contrast agents need to be injected into patients, it may cause allergic reactions in patients and even severe anaphylactic shock. In addition, the iodide in the contrast agent injected into the human body needs to be excreted through the human kidneys, so it may further damage the renal function of patients with impaired renal function. Therefore, enhanced CT has certain limitations in use and is not conducive to large-scale use.

[0005] Based on this, it is necessary to provide a computerized tomography image processing method that can display rich information about blood vessels, organs and diseased tissues without using contrast agents. Summary of the Invention

[0006] Based on this, the purpose of the present invention is to provide a computerized tomography image processing method and system, so as to provide a computerized tomography image processing method that can display rich information about blood vessels, organs and diseased tissues.

[0007] The first aspect of the embodiment of the present invention proposes:

[0008] A method for processing a computed tomography image, wherein the method comprises:

[0009] Performing a dual-energy CT scan and an enhanced CT scan on the same body position of the same patient to obtain a dual-energy CT image and an enhanced CT image respectively;

[0010] generating a corresponding CT image pair according to the dual-energy CT image and the enhanced CT image based on a preset rule, and generating a corresponding target training set according to the CT image pair;

[0011] Performing model training on a preset neural network using the target training set to generate a corresponding target image processing model;

[0012] A three-level image is generated according to the target dual-energy CT image of the patient, and a simulated enhanced CT image corresponding to the patient is generated in real time according to the three-level image through the target image processing model, wherein the simulated enhanced CT image is unique.

[0013] The beneficial effect of the present invention is that by setting the dual-energy CT images and enhanced CT images acquired in real time into corresponding CT image pairs, a target training set for subsequent training can be further generated. Based on this, the pre-set neural network model can be further trained through the target training set, and the required target image processing model can be obtained. Based on this, the target image processing model can be used to quickly and accurately obtain simulated enhanced CT images adapted to the current patient, thereby correspondingly improving work efficiency.

[0014] Furthermore, the step of generating a corresponding CT image pair according to the dual-energy CT image and the enhanced CT image based on a preset rule includes:

[0015] detecting in real time the low-energy image and the high-energy image respectively contained in the dual-energy CT image, and linearly combining the low-energy image and the high-energy image to generate a corresponding medium-energy image;

[0016] The medium-energy image and the dual-energy CT image are combined to generate a three-channel three-energy-level image, and the enhanced CT image is paired with the three-energy-level image to generate the CT image pair, each of which is unique.

[0017] Furthermore, the step of linearly combining the low-energy image and the high-energy image to generate a corresponding medium-energy image includes:

[0018] When the low-energy image and the high-energy image are respectively acquired, setting corresponding target weight factors for the low-energy image and the high-energy image;

[0019] The low-energy image and the high-energy image are fused according to the target weight factor to generate the corresponding medium-energy image, wherein the target weight factor is randomly selected from the uniform distribution U, and the expression of the algorithm for fusing the medium-energy image in real time is:

[0020] E mid =E low +w×(E high -E low )

[0021] Among them, E mid represents the medium energy image, ω represents the target weight factor, E low Represents the low energy image, E high represents the high energy image, wherein the uniform distribution U is between 0.2 and 0.8.

[0022] Furthermore, the step of generating a corresponding target training set according to the CT image pair includes:

[0023] The CT image pairs are subjected to normalization, image slicing, and data amplification processing in sequence, and the target training set is generated according to the processed CT image pairs, wherein the target training set includes specific parameters.

[0024] Furthermore, the step of performing model training on a preset neural network using the target training set to generate a corresponding target image processing model includes:

[0025] An EDCNN network is retrieved, and a trainable Sobel convolution is employed in an edge enhancement module in the EDCNN network, wherein the Sobel convolution is composed of a Sobel operator with defined learnable parameters;

[0026] According to the Sobel convolution, edge information corresponding to the target training set is extracted in real time, and the model training of the preset neural network is completed according to the edge information to generate the target image processing model accordingly.

[0027] Furthermore, the step of generating the target image processing model based on the target network architecture includes:

[0028] A corresponding composite loss is generated according to the MSE loss and the multi-scale perceptual loss, and the target network architecture is trained by the composite loss and the target training set to generate the target image processing model accordingly.

[0029] The second aspect of the embodiment of the present invention proposes:

[0030] A computerized tomography image processing system, wherein the system comprises:

[0031] a scanning module, configured to perform a dual-energy CT scan and an enhanced CT scan on the same body position of the same patient, respectively, to obtain a dual-energy CT image and an enhanced CT image respectively;

[0032] a processing module, configured to generate a corresponding CT image pair from the dual-energy CT image and the enhanced CT image based on a preset rule, and generate a corresponding target training set from the CT image pair;

[0033] A training module, configured to perform model training on a preset neural network using the target training set to generate a corresponding target image processing model;

[0034] An execution module is used to generate a three-level image based on the target dual-energy CT image of the patient, and to generate a simulated enhanced CT image corresponding to the patient in real time based on the three-level image through the target image processing model, wherein the simulated enhanced CT image is unique.

[0035] Furthermore, the processing module is specifically configured to:

[0036] detecting in real time the low-energy image and the high-energy image respectively contained in the dual-energy CT image, and linearly combining the low-energy image and the high-energy image to generate a corresponding medium-energy image;

[0037] The medium-energy image and the dual-energy CT image are combined to generate a three-channel three-energy-level image, and the enhanced CT image is paired with the three-energy-level image to generate the CT image pair, each of which is unique.

[0038] Furthermore, the processing module is specifically configured to:

[0039] When the low-energy image and the high-energy image are respectively acquired, setting corresponding target weight factors for the low-energy image and the high-energy image;

[0040] The low-energy image and the high-energy image are fused according to the target weight factor to generate the corresponding medium-energy image, wherein the target weight factor is randomly selected from the uniform distribution U, and the expression of the algorithm for fusing the medium-energy image in real time is:

[0041] E mid =E low +w×(E high -E low )

[0042] Among them, E midrepresents the medium energy image, ω represents the target weight factor, E low Represents the low energy image, E high represents the high energy image, wherein the uniform distribution U is between 0.2 and 0.8.

[0043] Furthermore, the processing module is specifically configured to:

[0044] The CT image pairs are subjected to normalization, image slicing, and data amplification processing in sequence, and the target training set is generated according to the processed CT image pairs, wherein the target training set includes specific parameters.

[0045] Furthermore, the training module is specifically used to:

[0046] An EDCNN network is retrieved, and a trainable Sobel convolution is employed in an edge enhancement module in the EDCNN network, wherein the Sobel convolution is composed of a Sobel operator with defined learnable parameters;

[0047] According to the Sobel convolution, edge information corresponding to the target training set is extracted in real time, and the model training of the preset neural network is completed according to the edge information to generate the target image processing model accordingly.

[0048] Furthermore, the training module is specifically used to:

[0049] A corresponding composite loss is generated according to the MSE loss and the multi-scale perceptual loss, and the target network architecture is trained by the composite loss and the target training set to generate the target image processing model accordingly.

[0050] The third aspect of the embodiment of the present invention proposes:

[0051] A computer comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer tomography image processing method as described above is implemented when the processor executes the computer program.

[0052] The fourth aspect of the embodiments of the present invention proposes:

[0053] A readable storage medium stores a computer program thereon, wherein when the program is executed by a processor, the computer tomography image processing method as described above is implemented.

[0054] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned by practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 A flowchart of a method for processing a computed tomography image provided by a first embodiment of the present invention;

[0056] Figure 2 This is a structural block diagram of a computed tomography image processing system provided by a fifth embodiment of the present invention.

[0057] The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0058] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to the accompanying drawings. The drawings illustrate several embodiments of the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and comprehensive understanding of the present invention.

[0059] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be an intermediate element. When an element is referred to as being "connected to" another element, it may be directly connected to the other element or there may be an intermediate element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only.

[0060] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0061] See also Figure 1 , shown is the computed tomography image processing method provided by the first embodiment of the present invention. The computed tomography image processing method provided by this embodiment can quickly and accurately obtain simulated enhanced CT images adapted to the patient through a target processing model constructed in real time, thereby correspondingly improving the doctor's work efficiency.

[0062] Specifically, this embodiment provides:

[0063] A method for processing a computed tomography image comprises the following steps:

[0064] Step S10, performing a dual-energy CT scan and an enhanced CT scan on the same body position of the same patient, to obtain a dual-energy CT image and an enhanced CT image respectively;

[0065] Step S20, generating a corresponding CT image pair from the dual-energy CT image and the enhanced CT image based on a preset rule, and generating a corresponding target training set from the CT image pair;

[0066] Step S30, performing model training on a preset neural network using the target training set to generate a corresponding target image processing model;

[0067] Step S40, generating a three-level image based on the target dual-energy CT image of the patient, and generating a simulated enhanced CT image corresponding to the patient in real time based on the three-level image through the target image processing model, wherein the simulated enhanced CT image is unique.

[0068] Specifically, in this embodiment, it should be noted that in order to complete patient imaging without the use of contrast agents, the present invention first requires performing a dual-energy CT scan and an enhanced CT scan on the same body part of the same patient. It should be noted that both scanning methods can be performed using existing technologies. Furthermore, after performing both scanning methods, the desired dual-energy CT image and enhanced CT image can be obtained.

[0069] Furthermore, in order to automatically complete image processing during subsequent work, it is necessary to first construct a corresponding image processing model. Based on this, in order to accurately obtain the required training samples, the present invention will immediately pair the currently generated dual-energy CT images and enhanced CT images according to pre-set rules to generate corresponding CT image pairs. Preferably, each CT image pair contains two images. Based on this, a corresponding target training set can be directly produced based on the currently acquired CT image pairs. On this basis, the present invention will immediately use the current target training set to perform corresponding model training on the pre-set EDCNN (edge-enhanced densely connected convolutional neural network), and can further train the required target image processing model. Furthermore, it is necessary to further process the target dual-energy CT image of the patient, preferably, to further convert it into the required three-level image. On this basis, finally, only the current three-level image needs to be processed using the target image processing model to immediately generate a simulated enhanced CT image corresponding to the current patient. This allows the patient to be scanned and the corresponding CT image to be generated without the use of contrast agents, eliminating safety risks and improving the patient's user experience.

[0070] Second embodiment

[0071] Furthermore, the step of generating a corresponding CT image pair according to the dual-energy CT image and the enhanced CT image based on a preset rule includes:

[0072] detecting in real time the low-energy image and the high-energy image respectively contained in the dual-energy CT image, and linearly combining the low-energy image and the high-energy image to generate a corresponding medium-energy image;

[0073] The medium-energy image and the dual-energy CT image are combined to generate a three-channel three-energy-level image, and the enhanced CT image is paired with the three-energy-level image to generate the CT image pair, each of which is unique.

[0074] Specifically, in this embodiment, it should be noted that after the required dual-energy CT image and enhanced CT image are obtained through the above steps, in order to objectively and accurately generate the required CT image pair, the present invention further performs matching processing on the current dual-energy CT image and the enhanced CT image. Preferably, since each CT image contains different energy densities and can be specifically divided into low energy, medium energy, and high energy, based on this, the present invention first detects the corresponding low-energy image and high-energy image contained in the current dual-energy CT image, and simultaneously immediately performs corresponding linear combination processing on the current low-energy image and high-energy image to generate the corresponding medium-energy image.

[0075] Furthermore, the present invention will further perform secondary combination processing on the current medium-energy image and the current dual-energy CT image, and can generate the required three-channel three-energy level image accordingly. Based on this, the present invention will immediately perform one-to-one pairing processing on each current enhanced CT image and each current three-energy level image, and further match several required CT image pairs. Among them, since each CT image pair contains different content, each CT image pair is unique, which is convenient for subsequent processing.

[0076] Furthermore, the step of linearly combining the low-energy image and the high-energy image to generate a corresponding medium-energy image includes:

[0077] When the low-energy image and the high-energy image are respectively acquired, setting corresponding target weight factors for the low-energy image and the high-energy image;

[0078] The low-energy image and the high-energy image are fused according to the target weight factor to generate the corresponding medium-energy image, wherein the target weight factor is randomly selected from the uniform distribution U, and the expression of the algorithm for fusing the medium-energy image in real time is:

[0079] E mid =E low +w×(E high -E low )

[0080] Among them, E mid represents the medium energy image, ω represents the target weight factor, E low Represents the low energy image, E high represents the high energy image, wherein the uniform distribution U is between 0.2 and 0.8.

[0081] Specifically, in this embodiment, it should be noted that after the required low-energy image and high-energy image are obtained respectively through the above steps, in order to quickly and effectively complete the linear combination of the current two images. Based on this, the current low-energy image and high-energy image are fused in real time according to the amount determined by the current target weight factor, so that the required medium-energy image can be generated for subsequent processing. It should be noted that the above method can highly simulate the process of doctors making personalized adjustments according to different patient conditions and diagnostic needs in clinical practice. In actual clinical use, doctors will flexibly select appropriate energy parameters according to specific cases. The random weight factor allows the model to be exposed to various possible energy combinations, thereby improving the robustness of the model. At the same time, through random sampling, the model avoids overfitting to specific weight factors, allowing the model to learn a wider range of feature representations and enhance its generalization ability in different scenarios. In addition, it should be noted that the consensus of a large number of clinical experts and relevant medical literature can also be referred to to understand the image feature requirements of different scanning parts during the diagnosis process. For example, in the diagnosis of lung diseases, clearly demonstrating the morphology and course of pulmonary vessels is crucial for determining the location and nature of lesions. Therefore, a preference for low energy is necessary to enhance vascular contrast. For the diagnosis of skeletal diseases, minimizing soft tissue interference and highlighting skeletal structure and detail are crucial, so a preference for high energy is recommended. Simultaneously, the scanned regions of the training set images are analyzed to determine optimal weighting factors based on the anatomical and physiological characteristics of each region. A weighting factor database can be established to store recommended weighting factors for different regions and regularly updated based on the latest clinical research and practical experience. Specifically, for ease of understanding and as an example, not limitation, consider the chest: a weighting factor of 0.35 is used. This weighting favors low energy, enhancing pulmonary vascular contrast and making vessels appear more clearly in the image, aiding in the detection of pulmonary vascular lesions such as pulmonary embolism and vascular malformations. The abdomen: a weighting factor of 0.45 is used to balance the contrast between the liver parenchyma and vascular structures, enabling clear visualization of the liver's internal structure and accurate observation of the liver's vascular conditions. This is crucial for the diagnosis of liver diseases such as liver cancer and hemangiomas. Bones: The weight factor is set to 0.8, which is biased towards high energy. It can effectively reduce the interference of soft tissue, highlight the morphology, density and structure of bones, and facilitate the diagnosis of bone diseases such as fractures, bone tumors, and osteoporosis. In this way, the training data can cover typical clinical scenarios, allowing the model to learn the image features of different parts under specific clinical needs. By using the training data generated by these fixed weight factors, the model can better adapt to various situations in actual clinical diagnosis and improve the accuracy and reliability of diagnosis. At the same time, it provides a clear learning goal for the model, which helps the model converge and optimize faster and improves training efficiency;

[0082] It's also worth noting that another feasible approach combines the random weighting method with a clinically driven fixed weighting method. During training, the training data generated using these two methods is distributed according to a certain ratio. For example, 70% of the training data can be generated using the random weighting method, and 30% using the clinically driven fixed weighting method. This approach ensures that the model learns a wide range of feature representations, improving robustness, while allowing the model to focus on typical clinical scenarios and improve adaptability to specific sites.

[0083] Third embodiment

[0084] Furthermore, the step of generating a corresponding target training set according to the CT image pair includes:

[0085] The CT image pairs are subjected to normalization, image slicing, and data amplification processing in sequence, and the target training set is generated according to the processed CT image pairs, wherein the target training set includes specific parameters.

[0086] In addition, in this embodiment, it should be noted that after the required CT image pairs are acquired in real time through the above steps, subsequent model training can be accurately performed.

[0087] Furthermore, before conducting formal model training, in order to correspondingly improve the speed of model training, the present invention will further perform normalization, image slicing and data amplification processing on each current CT image pair, that is, perform corresponding standardization processing on each current CT image, and finally generate the required target training set based on the processed CT image pairs to facilitate subsequent processing.

[0088] Furthermore, the step of performing model training on a preset neural network using the target training set to generate a corresponding target image processing model includes:

[0089] An EDCNN network is retrieved, and a trainable Sobel convolution is employed in an edge enhancement module in the EDCNN network, wherein the Sobel convolution is composed of a Sobel operator with defined learnable parameters;

[0090] According to the Sobel convolution, edge information corresponding to the target training set is extracted in real time, and the model training of the preset neural network is completed according to the edge information to generate the target image processing model accordingly.

[0091] In addition, in this embodiment, it should be noted that after the required target training set is further obtained through the above steps, targeted model training can be carried out at this time. Preferably, the present invention will call out the pre-set EDCNN network, and detect the edge enhancement model contained in the current EDCNN network. Specifically, the present invention will set a trainable Sobel convolution inside the edge enhancement module. Specifically, the Sobel convolution provided by the present invention is specifically composed of a Sobel operator with defined learnable parameters. Furthermore, the present invention will further extract the edge information corresponding to the current target training set in real time through the current Sobel convolution. Based on this, the current edge information is further transmitted to a densely connected architecture with eight convolution blocks, so that a required overall network architecture can be preliminarily constructed. On this basis, the current overall network architecture is further trained with the above-mentioned target training set to perform corresponding model training, so that the required target image processing model can be finally trained for subsequent processing.

[0092] Fourth embodiment

[0093] Furthermore, the step of generating the target image processing model based on the target network architecture includes:

[0094] A corresponding composite loss is generated according to the MSE loss and the multi-scale perceptual loss, and the target network architecture is trained by the composite loss and the target training set to generate the target image processing model accordingly.

[0095] Among them, in this embodiment, it should be pointed out that after the required target network architecture is finally constructed through the above steps, before starting training, the corresponding loss function needs to be determined for subsequent verification. Based on this, the present invention will respectively call out the required MES loss and multi-scale perception loss, and immediately fuse the current two losses to fuse the required composite loss. Based on this, the current target network architecture training is finally completed through the current target training set and the composite loss, and the required target image processing model is finally generated for subsequent processing.

[0096] See also Figure 2 , the fifth embodiment of the present invention provides:

[0097] A computerized tomography image processing system, wherein the system comprises:

[0098] a scanning module, configured to perform a dual-energy CT scan and an enhanced CT scan on the same body position of the same patient, respectively, to obtain a dual-energy CT image and an enhanced CT image respectively;

[0099] a processing module, configured to generate a corresponding CT image pair from the dual-energy CT image and the enhanced CT image based on a preset rule, and generate a corresponding target training set from the CT image pair;

[0100] A training module, configured to perform model training on a preset neural network using the target training set to generate a corresponding target image processing model;

[0101] An execution module is used to generate a three-level image based on the target dual-energy CT image of the patient, and to generate a simulated enhanced CT image corresponding to the patient in real time based on the three-level image through the target image processing model, wherein the simulated enhanced CT image is unique.

[0102] Furthermore, the processing module is specifically configured to:

[0103] detecting in real time the low-energy image and the high-energy image respectively contained in the dual-energy CT image, and linearly combining the low-energy image and the high-energy image to generate a corresponding medium-energy image;

[0104] The medium-energy image and the dual-energy CT image are combined to generate a three-channel three-energy-level image, and the enhanced CT image is paired with the three-energy-level image to generate the CT image pair, each of which is unique.

[0105] Furthermore, the processing module is specifically configured to:

[0106] When the low-energy image and the high-energy image are respectively acquired, setting corresponding target weight factors for the low-energy image and the high-energy image;

[0107] The low-energy image and the high-energy image are fused according to the target weight factor to generate the corresponding medium-energy image, wherein the target weight factor is randomly selected from the uniform distribution U, and the expression of the algorithm for fusing the medium-energy image in real time is:

[0108] E mid =E low +w×(E high -E low )

[0109] Among them, E mid represents the medium energy image, ω represents the target weight factor, E low Represents the low energy image, E high represents the high energy image, wherein the uniform distribution U is between 0.2 and 0.8.

[0110] Furthermore, the processing module is specifically configured to:

[0111] The CT image pairs are subjected to normalization, image slicing, and data amplification processing in sequence, and the target training set is generated according to the processed CT image pairs, wherein the target training set includes specific parameters.

[0112] Furthermore, the training module is specifically used to:

[0113] An EDCNN network is retrieved, and a trainable Sobel convolution is employed in an edge enhancement module in the EDCNN network, wherein the Sobel convolution is composed of a Sobel operator with defined learnable parameters;

[0114] According to the Sobel convolution, edge information corresponding to the target training set is extracted in real time, and the model training of the preset neural network is completed according to the edge information to generate the target image processing model accordingly.

[0115] Furthermore, the training module is specifically used to:

[0116] A corresponding composite loss is generated according to the MSE loss and the multi-scale perceptual loss, and the target network architecture is trained by the composite loss and the target training set to generate the target image processing model accordingly.

[0117] A sixth embodiment of the present invention provides a computer, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the computed tomography image processing method as described above when executing the computer program.

[0118] A seventh embodiment of the present invention provides a readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the computed tomography image processing method as described above.

[0119] In summary, the computed tomography image processing method and system provided by the above embodiments of the present invention can quickly and accurately obtain simulated enhanced CT images adapted to the patient through the target image processing model constructed in real time, thereby correspondingly improving work efficiency.

[0120] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0121] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0122] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0123] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0124] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0125] The above-described embodiments merely illustrate several embodiments of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make various modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A method for processing a computed tomography image, characterized in that: The method comprises: Performing a dual-energy CT scan and an enhanced CT scan on the same body position of the same patient to obtain a dual-energy CT image and an enhanced CT image respectively; generating a corresponding CT image pair according to the dual-energy CT image and the enhanced CT image based on a preset rule, and generating a corresponding target training set according to the CT image pair; Performing model training on a preset neural network using the target training set to generate a corresponding target image processing model; A three-level image is generated according to the target dual-energy CT image of the patient, and a simulated enhanced CT image corresponding to the patient is generated in real time according to the three-level image through the target image processing model, wherein the simulated enhanced CT image is unique.

2. The method for processing computed tomography images according to claim 1, wherein: The step of generating a corresponding CT image pair according to the dual-energy CT image and the enhanced CT image based on a preset rule includes: detecting in real time the low-energy image and the high-energy image respectively contained in the dual-energy CT image, and linearly combining the low-energy image and the high-energy image to generate a corresponding medium-energy image; The medium-energy image and the dual-energy CT image are combined to generate a three-channel three-energy-level image, and the enhanced CT image is paired with the three-energy-level image to generate the CT image pair, each of which is unique.

3. The method for processing computed tomography images according to claim 2, wherein: The step of linearly combining the low-energy image and the high-energy image to generate a corresponding medium-energy image includes: When the low-energy image and the high-energy image are respectively acquired, setting corresponding target weight factors for the low-energy image and the high-energy image; The low-energy image and the high-energy image are fused according to the target weight factor to generate the corresponding medium-energy image, wherein the target weight factor is randomly selected from the uniform distribution U, and the expression of the algorithm for fusing the medium-energy image in real time is: AND mid =And low +w×(E high -AND low ) Among them, E mid represents the medium energy image, ω represents the target weight factor, E low Represents the low energy image, E high represents the high energy image, wherein the uniform distribution U is between 0.2 and 0.

8.

4. The method for processing computed tomography images according to claim 1, wherein: The step of generating a corresponding target training set according to the CT image pair comprises: The CT image pairs are subjected to normalization, image slicing, and data amplification processing in sequence, and the target training set is generated according to the processed CT image pairs, wherein the target training set includes specific parameters.

5. The method for processing computed tomography images according to claim 2, wherein: The step of performing model training on a preset neural network using the target training set to generate a corresponding target image processing model includes: An EDCNN network is retrieved, and a trainable Sobel convolution is employed in an edge enhancement module in the EDCNN network, wherein the Sobel convolution is composed of a Sobel operator with defined learnable parameters; According to the Sobel convolution, edge information corresponding to the target training set is extracted in real time, and the model training of the preset neural network is completed according to the edge information to generate the target image processing model accordingly.

6. The method for processing computed tomography images according to claim 5, wherein: The step of generating the target image processing model based on the target network architecture includes: A corresponding composite loss is generated according to the MSE loss and the multi-scale perceptual loss, and the target network architecture is trained by the composite loss and the target training set to generate the target image processing model accordingly.

7. A computer tomography image processing system, characterized in that: The system comprises: a scanning module, configured to perform a dual-energy CT scan and an enhanced CT scan on the same body position of the same patient, respectively, to obtain a dual-energy CT image and an enhanced CT image, respectively; a processing module, configured to generate a corresponding CT image pair from the dual-energy CT image and the enhanced CT image based on a preset rule, and generate a corresponding target training set from the CT image pair; A training module, configured to perform model training on a preset neural network using the target training set to generate a corresponding target image processing model; An execution module is used to generate a three-level image based on the target dual-energy CT image of the patient, and to generate a simulated enhanced CT image corresponding to the patient in real time based on the three-level image through the target image processing model, wherein the simulated enhanced CT image is unique.

8. A computer comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the computer tomography image processing method according to any one of claims 1 to 6 is implemented.

9. A readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the computer tomography image processing method according to any one of claims 1 to 6 is implemented.

Images