De-scattering method, device and equipment for CBCT image and storage medium

Abstract

The invention discloses a CBCT (Cone Beam Computed Tomography) image scattering removal method, device and equipment and a storage medium. The CBCT image scattering removal method comprises the following steps: acquiring a plurality of CBCT original images and CT original images; performing image registration by using a registration algorithm; processing the registered image by adopting an orthographic projection algorithm to obtain a CBCT projection image and a CT projection image at different angles; inputting the CBCT projection image into a scattering correction model for training, the scattering correction model being used for performing feature extraction and analysis on the input CBCT projection image, and outputting a convolution kernel corresponding to scattering distribution; by adopting a deconvolution method, generating a scattering-removed projection image from the CBCT projection image according to a convolution kernel output by the scattering correction model; adjusting parameters of a scattering correction model according to the difference between the de-scattering projection image and the CT projection image; a scattering correction model meeting requirements is obtained; and utilizing the trained scattering correction model to carry out scattering removal processing on the CBCT original image. The scattering-removed CBCT image obtained by adopting the method disclosed by the invention does not have artifacts.

Description

Technical Field

[0001] The present invention belongs to the technical field of image processing, and in particular relates to a descattering method, device, equipment and storage medium for CBCT images. Background Art

[0002] In the field of medical imaging, CBCT (Cone Beam Computed Tomography) technology has been widely used in the fields of oral and maxillofacial surgery, orthopedics, etc. due to its unique advantages, such as high spatial resolution and low radiation dose. At the same time, CBCT also plays an important role in the field of radiotherapy. In radiotherapy, CBCT can be used for positioning before treatment, monitoring during treatment, and evaluation after treatment, providing important support for precise radiotherapy.

[0003] However, CBCT images are often affected by scattered radiation during acquisition, which can lead to reduced image contrast, increased noise, and blurred details, which seriously affects image quality and diagnostic accuracy.

[0004] At present, the methods to solve the scattering problem of CBCT images mainly include physical methods and traditional image processing methods. Physical methods usually reduce the impact of scattered radiation by improving the hardware structure of the scanning equipment, such as adding collimators and using anti-scatter grids. However, these methods often increase the cost and complexity of the equipment and have limited effects.

[0005] Traditional image processing methods, such as filtering and histogram equalization, can improve image quality to a certain extent, but they are not capable of dealing with complex problems caused by scattering. These methods are usually based on simple mathematical models and cannot accurately simulate the physical process of scattering, making it difficult to effectively remove scattering.

[0006] In addition, some existing methods may cause important information of the image to be lost or introduce new artifacts and errors when dealing with scattering problems. At the same time, these methods have high computational complexity and long processing time, making it difficult to meet the clinical real-time requirements.

[0007] In summary, the existing technology has many deficiencies in solving the scattering problem of CBCT images and is unable to effectively remove scattering and improve image quality to meet the needs of clinical diagnosis and treatment, especially in the field of radiotherapy, for high-precision images. Summary of the invention

[0008] In order to solve the above technical problems, the present invention proposes a descattering method, device, equipment and storage medium for CBCT images.

[0009] In order to achieve the above object, the technical solution of the present invention is as follows:

[0010] In a first aspect, the present invention discloses a descattering method for a CBCT image, comprising:

[0011] Step S1: Acquire a number of CBCT original images and CT original images;

[0012] Step S2: performing image registration on the acquired CBCT original image and CT original image using a registration algorithm;

[0013] Step S3: using a forward projection algorithm to process the registered CBCT original image and CT original image to obtain CBCT projection images and CT projection images at different angles;

[0014] Step S4: inputting the CBCT projection image into the scatter correction model for training. The scatter correction model is used to extract and analyze the input CBCT projection image and output a convolution kernel corresponding to the scatter distribution.

[0015] Step S5: using a deconvolution method to generate a descattered projection image by applying a convolution kernel output by the scatter correction model to the CBCT projection image;

[0016] Step S6: adjusting the parameters of the scatter correction model according to the difference between the de-scattered projection image and the CT projection image;

[0017] Step S7: repeating steps S4 to S6 until the trained scatter correction model meets the requirements;

[0018] Step S8: De-scattering the CBCT original image using the trained scatter correction model.

[0019] Based on the above technical solution, the following improvements can be made:

[0020] As a preferred solution, the scatter correction model includes:

[0021] The feature extraction part is used to convert the input CBCT projection image into feature information;

[0022] The feature prediction part is used to predict the convolution kernel that meets the characteristics of the point spread function according to the feature information after the CBCT projection image is transformed.

[0023] As a preferred solution, the feature extraction part includes: several convolution layers and pooling layers.

[0024] As a preferred solution, step S8 includes:

[0025] Step S8.1: Acquire the original CBCT image to be processed;

[0026] Step S8.2: using a forward projection algorithm to process the CBCT original image acquired in step S8.1 to obtain CBCT projection images at different angles;

[0027] Step S8.3: input the CBCT projection image obtained in step S8.2 into the trained scatter correction model, and output the convolution kernel corresponding to the scatter distribution;

[0028] Step S8.4: using a deconvolution method, the CBCT projection image obtained in step S8.2 is subjected to a convolution kernel output by the scatter correction model to generate a descattered projection image;

[0029] Step S8.5: Reconstruct the descattered projection image obtained in step S8.4 using a reconstruction algorithm to obtain a descattered CBCT image.

[0030] In a second aspect, the present invention discloses a descattering device for CBCT images, comprising:

[0031] An image acquisition module, used for acquiring a number of CBCT original images and CT original images;

[0032] A registration module, used for performing image registration on the acquired CBCT original image and CT original image by using a registration algorithm;

[0033] The orthographic projection module is used to process the registered CBCT original image and CT original image by using the orthographic projection algorithm to obtain CBCT projection images and CT projection images at different angles;

[0034] A model training module is used to input the CBCT projection image into the scatter correction model for training. The scatter correction model is used to extract and analyze the input CBCT projection image and output a convolution kernel corresponding to the scatter distribution.

[0035] A deconvolution module, for generating a descattered projection image by applying a deconvolution method to the CBCT projection image according to a convolution kernel output by a scatter correction model;

[0036] A parameter adjustment module, used for adjusting the parameters of the scatter correction model according to the difference between the descattered projection image and the CT projection image;

[0037] A repeated execution module is used to repeatedly execute the methods in the model training module, the deconvolution module, and the parameter adjustment module until the trained scatter correction model meets the requirements;

[0038] The application module is used to perform descattering processing on the CBCT original image using the trained scattering correction model.

[0039] As a preferred solution, the scatter correction model includes:

[0040] The feature extraction part is used to convert the input CBCT projection image into feature information;

[0041] The feature prediction part is used to predict the convolution kernel that meets the characteristics of the point spread function according to the feature information after the CBCT projection image is transformed.

[0042] As a preferred solution, the feature extraction part includes: several convolution layers and pooling layers.

[0043] As a preferred solution, the application module includes:

[0044] An image acquisition unit, used for acquiring the original CBCT image to be processed;

[0045] An orthographic projection unit, used to process the CBCT original image acquired in step S8.1 using an orthographic projection algorithm to obtain CBCT projection images at different angles;

[0046] A processing unit, used for inputting the CBCT projection image obtained by the forward projection unit into the trained scatter correction model, and outputting a convolution kernel corresponding to the scatter distribution;

[0047] A deconvolution unit, used for generating a descattered projection image by using a deconvolution method to convert the CBCT projection image obtained by the forward projection unit into a descattered projection image according to a convolution kernel output by a scatter correction model;

[0048] The reconstruction unit is used to reconstruct the descattered projection image obtained by the deconvolution unit by using a reconstruction algorithm to obtain a descattered CBCT image.

[0049] In a third aspect, the present invention discloses a computing device, comprising:

[0050] one or more processors;

[0051] Memory;

[0052] and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, and the one or more programs include instructions for any one of the above-mentioned descattering methods for CBCT images.

[0053] In a fourth aspect, the present invention discloses a storage medium storing one or more computer-readable programs, wherein the one or more programs include instructions suitable for being loaded by a memory and executing any of the above-mentioned descattering methods for CBCT images.

[0054] The present invention discloses a descattering method, device, equipment and storage medium for CBCT images, which have the following beneficial effects:

[0055] First, the scattering distribution is simplified to a point spread function, which is more interpretable.

[0056] Second, the descattered CBCT image obtained by the present invention has no artifacts.

[0057] Third, the present invention predicts the convolution kernel based on image feature information and has strong adaptability.

[0058] Fourthly, the scatter correction model disclosed in the present invention has fewer model parameters and faster calculation, and no additional hardware is required in the imaging device to remove scatter. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 The present invention provides a flowchart of a descattering method according to an embodiment of the present invention.

[0061] Figure 2 A schematic diagram of applying a trained scatter correction model provided by an embodiment of the present invention.

[0062] Figure 3 A schematic diagram of a descattering device provided in an embodiment of the present invention.

[0063] Figure 4 A schematic diagram of a computing device provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0065] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0066] The expression of “comprising” an element is an “open” expression, which merely means that a corresponding component or step exists, and should not be interpreted as excluding additional components or steps.

[0067] In order to achieve the purpose of the present invention, some embodiments of a CBCT image descattering method, device, apparatus and storage medium are as follows: Figure 1 As shown, the descattering method includes:

[0068] Step S101: Acquire a number of CBCT original images and CT original images;

[0069] Step S102: performing image registration on the acquired CBCT original image and CT original image using a registration algorithm;

[0070] Step S103: using the FDK forward projection algorithm to process the registered CBCT original image and CT original image to obtain CBCT projection images and CT projection images at different angles;

[0071] Step S104: inputting the CBCT projection image into a scatter correction model for training. The scatter correction model is used to extract and analyze features of the input CBCT projection image and output a convolution kernel corresponding to the scatter distribution.

[0072] Step S105: using a deconvolution method to generate a de-scattered projection image by applying a convolution kernel output by the scatter correction model to the CBCT projection image;

[0073] Step S106: adjusting the parameters of the scatter correction model according to the difference between the de-scattered projection image and the CT projection image;

[0074] Step S107: repeating steps S104 to S106 until the trained scatter correction model meets the requirements;

[0075] Step S108: using the trained scatter correction model to perform descattering processing on the CBCT original image.

[0076] Each step is described in detail below.

[0077] Step S101 is a data acquisition step, in which the CBCT original image and the CT original image are acquired, and the acquired data can be stored and managed by a computer device. The acquisition of the above CBCT original image is performed by using the imaging principle of the CBCT scanning device.

[0078] Step S102 is a data registration step, in which the corresponding CBCT original image and CT original image are collected and registered. The registration principle is to find the corresponding relationship between the CBCT original image and the CT original image through a specific algorithm, so as to generate an accurate projection image later.

[0079] Step S103 is a data forward projection step, in which the registered CBCT original image and the CT original image are processed using the FDK forward projection method to generate projection images at different angles, which are used as training sets for subsequent models. The FDK forward projection algorithm is a conventional mathematical model in the art, and can accurately generate 360° projection images that meet the requirements, and will not be elaborated here.

[0080] Steps S104 to S107 are steps for training the scatter correction model.

[0081] The input of the scatter correction model is the CBCT projection image, the scatter distribution is simplified as a point spread function, and the output is the convolution kernel corresponding to the scatter distribution.

[0082] The above scatter correction model includes:

[0083] The feature extraction part is used to convert the input CBCT projection image into feature information;

[0084] The feature prediction part is used to predict the convolution kernel that meets the characteristics of the point spread function according to the feature information after the CBCT projection image is transformed.

[0085] Furthermore, the feature extraction part includes: a number of convolutional layers and pooling layers.

[0086] The training step is performed on a computer device using a pre-trained encoder model as the backbone network for convolution kernel prediction.

[0087] When the CBCT projection image data is input into the pre-trained encoder model, it first undergoes multiple convolution operations in the convolution layer to gradually extract features at different levels of the image, gradually combining simple edge and texture information into more complex image structure features. The pooling layer downsamples the features when appropriate, retaining important features while reducing the amount of data. After multiple convolutional and pooling layers are processed alternately, the model converts the CBCT projection image into a highly abstract feature representation, captures key information related to scattering, and lays the foundation for predicting the convolution kernel.

[0088] The feature prediction part predicts the convolution kernel based on the feature information.

[0089] The pre-trained encoder model has learned a large amount of common feature representation patterns for image data. When faced with CBCT projection images, it can predict the convolution kernel corresponding to the scattering distribution based on the extracted feature information.

[0090] The model's internal weight parameters have been optimized and adjusted during the pre-training process. These weights determine the contribution of different features to the convolution kernel prediction. Specific feature combinations may be associated with specific distribution patterns scattered in the image. By learning these associations, the model can accurately output the corresponding convolution kernel based on the input CBCT projection image feature information.

[0091] The assumption that the scattering distribution is simplified to a point spread function facilitates the prediction of the convolution kernel. Based on the learning of image features and scattering characteristics, the pre-trained encoder model can predict the convolution kernel that conforms to the characteristics of the point spread function according to the characteristics of the CBCT projection image, so that the subsequent deconvolution operation can effectively remove the scattering effect and generate a de-scattered projection image.

[0092] Step S105 uses a deconvolution method to generate a descattered projection image according to the output convolution kernel. The principle of the deconvolution method is to remove the influence of scattering from the original image by inverse operation of the convolution kernel, thereby generating a descattered image. Compared with the prior art, the present invention can more accurately remove scattering and generate a high-quality descattered projection image.

[0093] Step S106 calculates the loss between the generated descattered projection image and the CT projection image, and optimizes the scatter correction model. The principle of this step is to adjust the parameters of the scatter correction model by comparing the difference between the descattered image and the CT projection, so that the model can better adapt to different CBCT images and improve the accuracy and generalization ability of the model.

[0094] like Figure 2 As shown, the above step S108 includes:

[0095] Step S108.1: Acquire the original CBCT image to be processed;

[0096] Step S108.2: using a forward projection algorithm to process the CBCT original image acquired in step S108.1 to obtain CBCT projection images at different angles;

[0097] Step S108.3: input the CBCT projection image obtained in step S108.2 into the trained scatter correction model, and output a convolution kernel corresponding to the scatter distribution;

[0098] Step S108.4: using a deconvolution method, the CBCT projection image obtained in step S108.2 is subjected to a convolution kernel output by the scatter correction model to generate a descattered projection image;

[0099] Step S108.5: Reconstruct the descattered projection image obtained in step S108.4 using a reconstruction algorithm to obtain a descattered CBCT image.

[0100] The scatter correction model can output the corresponding convolution kernel based on the CBCT image feature information, so the CBCT imaging device does not need to be equipped with additional de-scattering equipment.

[0101] In some embodiments, GPU can be used on computer devices to accelerate computing, and the processing time of each projection image can be in milliseconds. Therefore, the present invention is efficient and convenient in practical applications, and can process CBCT projection data more quickly than the prior art, thereby improving the efficiency of clinical diagnosis.

[0102] The following problems exist in the prior art:

[0103] 1) Using deep learning models with large number of parameters for prediction increases a lot of data processing time. 2) Using end-to-end methods to generate images is prone to artifacts. 3) Weak adaptability, poor versatility, and poor interpretability.

[0104] The present invention has the following effects:

[0105] First, the present invention simplifies the scattering distribution into a point spread function, which is highly interpretable and easier to understand and apply than the complex deep learning models or inaccurate physical models in the prior art.

[0106] Second, the present invention ensures that there are no artifacts in the descattered CBCT image, while the prior art uses an end-to-end method to generate images that are prone to artifacts.

[0107] Third, since the prediction of the convolution kernel of the present invention is based on image feature information, it has strong adaptability.

[0108] Fourth, compared with the prior art, which may have the problem of using a large number of model parameters, resulting in slow calculations, and may require additional hardware to remove scatter, the present invention reduces the difficulty of the task, so the model parameters are small, the calculation is fast, and no additional hardware is required in the imaging device to remove scatter.

[0109] In some embodiments, Figure 3 As shown, the present invention discloses a descattering device for CBCT images, comprising:

[0110] An image acquisition module 201 is used to acquire a number of CBCT original images and CT original images;

[0111] A registration module 202 is used to perform image registration on the collected CBCT original image and CT original image using a registration algorithm;

[0112] The orthographic projection module 203 is used to process the registered CBCT original image and CT original image by using the orthographic projection algorithm to obtain CBCT projection images and CT projection images at different angles;

[0113] A model training module 204 is used to input the CBCT projection image into the scatter correction model for training. The scatter correction model is used to extract and analyze the input CBCT projection image and output a convolution kernel corresponding to the scatter distribution.

[0114] The deconvolution module 205 is used to generate a de-scattered projection image by applying a deconvolution method to the CBCT projection image according to the convolution kernel output by the scatter correction model;

[0115] A parameter adjustment module 206, for adjusting the parameters of the scatter correction model according to the difference between the descattered projection image and the CT projection image;

[0116] A repeated execution module 207 is used to repeatedly execute the methods in the model training module, the deconvolution module, and the parameter adjustment module until the trained scatter correction model meets the requirements;

[0117] The application module 208 is used to perform descattering processing on the CBCT original image using the trained scattering correction model.

[0118] The above scatter correction model includes:

[0119] The feature extraction part is used to convert the input CBCT projection image into feature information;

[0120] The feature prediction part is used to predict the convolution kernel that meets the characteristics of the point spread function according to the feature information after the CBCT projection image is transformed.

[0121] The above feature extraction part includes: several convolutional layers and pooling layers.

[0122] Furthermore, the application module includes:

[0123] An image acquisition unit, used for acquiring the original CBCT image to be processed;

[0124] An orthographic projection unit, used to process the CBCT original image acquired in step S8.1 using an orthographic projection algorithm to obtain CBCT projection images at different angles;

[0125] A processing unit, used for inputting the CBCT projection image obtained by the forward projection unit into the trained scatter correction model, and outputting a convolution kernel corresponding to the scatter distribution;

[0126] A deconvolution unit, used for generating a descattered projection image by using a deconvolution method to convert the CBCT projection image obtained by the forward projection unit into a descattered projection image according to a convolution kernel output by a scatter correction model;

[0127] The reconstruction unit is used to reconstruct the descattered projection image obtained by the deconvolution unit by using a reconstruction algorithm to obtain a descattered CBCT image.

[0128] Further, it should be noted that: when the descattering device for CBCT images provided in the above embodiments performs descattering processing, only the division of the above functional modules is used as an example for explanation. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the descattering device for CBCT images is divided into different functional modules to complete all or part of the functions described above.

[0129] In addition, the embodiments of the CBCT image descattering device and the CBCT image descattering method provided in the above embodiments belong to the same concept, and their specific implementation processes are detailed in the method embodiments, which will not be repeated here.

[0130] In addition, in some other embodiments, Figure 4 As shown, the present invention also discloses a computing device, including:

[0131] One or more processors 301;

[0132] Memory 302;

[0133] and one or more programs, wherein the one or more programs are stored in the memory 302 and configured to be executed by the one or more processors 301, and the one or more programs include instructions for the descattering method of the CBCT image disclosed in the above embodiments.

[0134] The processor 301 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 301 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 301 may also include a main processor and a coprocessor. The main processor is a processor for processing data in an awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in a standby state. In some embodiments, the processor 301 may be integrated with a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 301 may also include an AI (Artificial Intelligence) processor, which is used to process computing operations related to machine learning.

[0135] The memory 302 may include one or more computer-readable storage media, which may be non-transitory. The memory 302 may also include a high-speed random access memory and a non-volatile memory, such as one or more disk storage devices and flash memory storage devices. In some embodiments, the non-transitory computer-readable storage medium in the memory 302 is used to store at least one instruction, which is used to be executed by the processor 301 to implement the descattering method of the CBCT image provided by the method embodiment of the present invention.

[0136] In addition, the computing device may also optionally include: a peripheral device interface and at least one peripheral device. The processor 301, the memory 302 and the peripheral device interface may be connected via a bus or a signal line. Each peripheral device may be connected to the peripheral device interface via a bus, a signal line or a circuit board. Schematically, the peripheral devices include but are not limited to: a radio frequency circuit, a touch display screen, an audio circuit, and a power supply.

[0137] Of course, the computing device may also include fewer or more components, which is not limited in this embodiment.

[0138] In addition, in some other embodiments, the present invention further discloses a storage medium, which stores one or more computer-readable programs, and the one or more programs include instructions, and the instructions are suitable for being loaded by a memory and executing the descattering method of CBCT images disclosed in the above embodiments.

[0139] The present invention discloses a descattering method, device, equipment and storage medium for CBCT images, which have the following beneficial effects:

[0140] First, the scattering distribution is simplified to a point spread function, which is more interpretable.

[0141] Second, the descattered CBCT image obtained by the present invention has no artifacts.

[0142] Third, the present invention predicts the convolution kernel based on image feature information and has strong adaptability.

[0143] Fourthly, the scatter correction model disclosed in the present invention has fewer model parameters and faster calculation, and no additional hardware is required in the imaging device to remove scatter.

[0144] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for illustrating the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which shall fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A descattering method for CBCT images, characterized in that: include: Step S1: Acquire a number of CBCT original images and CT original images; Step S2: performing image registration on the acquired CBCT original image and CT original image using a registration algorithm; Step S3: using a forward projection algorithm to process the registered CBCT original image and CT original image to obtain CBCT projection images and CT projection images at different angles; Step S4: inputting the CBCT projection image into a scatter correction model for training, wherein the scatter correction model is used to extract and analyze features of the input CBCT projection image and output a convolution kernel corresponding to the scatter distribution; Step S5: using a deconvolution method to generate a descattered projection image by applying a convolution kernel output by the scatter correction model to the CBCT projection image; Step S6: adjusting the parameters of the scatter correction model according to the difference between the de-scattered projection image and the CT projection image; Step S7: repeating steps S4 to S6 until the trained scatter correction model meets the requirements; Step S8: De-scattering the CBCT original image using the trained scatter correction model.

2. The descattering method according to claim 1, characterized in that: The scatter correction model includes: A feature extraction part, wherein the feature extraction part is used to convert the input CBCT projection image into feature information; The feature prediction part is used to predict the convolution kernel that meets the point spread function characteristics according to the feature information after the CBCT projection image is converted.

3. The descattering method according to claim 2, characterized in that: The feature extraction part includes: several convolution layers and pooling layers.

4. The descattering method according to any one of claims 1 to 3, characterized in that: The step S8 comprises: Step S8.1: Acquire the original CBCT image to be processed; Step S8.2: using a forward projection algorithm to process the CBCT original image acquired in step S8.1 to obtain CBCT projection images at different angles; Step S8.3: input the CBCT projection image obtained in step S8.2 into the trained scatter correction model, and output the convolution kernel corresponding to the scatter distribution; Step S8.4: using a deconvolution method, the CBCT projection image obtained in step S8.2 is subjected to a convolution kernel output by the scatter correction model to generate a descattered projection image; Step S8.5: Reconstruct the descattered projection image obtained in step S8.4 using a reconstruction algorithm to obtain a descattered CBCT image.

5. A descattering device for CBCT images, characterized in that: include: An image acquisition module, used for acquiring a number of CBCT original images and CT original images; A registration module, used for performing image registration on the acquired CBCT original image and CT original image by using a registration algorithm; The orthographic projection module is used to process the registered CBCT original image and CT original image by using the orthographic projection algorithm to obtain CBCT projection images and CT projection images at different angles; A model training module, used for inputting the CBCT projection image into the scatter correction model for training, wherein the scatter correction model is used for extracting and analyzing features of the input CBCT projection image and outputting a convolution kernel corresponding to the scatter distribution; A deconvolution module, for generating a descattered projection image by applying a deconvolution method to the CBCT projection image according to a convolution kernel output by a scatter correction model; A parameter adjustment module, used for adjusting the parameters of the scatter correction model according to the difference between the descattered projection image and the CT projection image; A repeated execution module is used to repeatedly execute the methods in the model training module, the deconvolution module, and the parameter adjustment module until the trained scatter correction model meets the requirements; The application module is used to perform descattering processing on the CBCT original image using the trained scattering correction model.

6. The descattering device according to claim 5, characterized in that: The scatter correction model includes: A feature extraction part, wherein the feature extraction part is used to convert the input CBCT projection image into feature information; The feature prediction part is used to predict the convolution kernel that meets the point spread function characteristics according to the feature information after the CBCT projection image is converted.

7. The descattering device according to claim 6, characterized in that: The feature extraction part includes: several convolution layers and pooling layers.

8. The descattering device according to any one of claims 5 to 7, characterized in that: The application module includes: An image acquisition unit, used for acquiring the original CBCT image to be processed; An orthographic projection unit, used to process the CBCT original image acquired in step S8.1 using an orthographic projection algorithm to obtain CBCT projection images at different angles; A processing unit, used for inputting the CBCT projection image obtained by the forward projection unit into the trained scatter correction model, and outputting a convolution kernel corresponding to the scatter distribution; A deconvolution unit, used for generating a descattered projection image by using a deconvolution method to convert the CBCT projection image obtained by the forward projection unit into a descattered projection image according to a convolution kernel output by a scatter correction model; The reconstruction unit is used to reconstruct the descattered projection image obtained by the deconvolution unit by using a reconstruction algorithm to obtain a descattered CBCT image.

9. A computing device, characterized in that include: one or more processors; Memory; and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by one or more processors, and the one or more programs include instructions for the descattering method of the CBCT image described in any one of claims 1 to 4.

10. A storage medium, characterized in that The storage medium stores one or more computer-readable programs, wherein the one or more programs include instructions, and the instructions are suitable for being loaded by the memory and executing the descattering method for CBCT images described in any one of claims 1 to 4.

Images