Small animal CBCT (cone beam computed tomography) imaging device and method with FOV (field of view) size continuously adjustable based on super-resolution network

Abstract

The invention discloses a super-resolution network-based small animal CBCT (cone beam computed tomography) imaging device and a super-resolution network-based small animal CBCT imaging method with continuously adjustable FOV (field of view), both a radiation source and an objective table can horizontally move, and continuous adjustment of the FOV is realized through unique structural design and control mode. Meanwhile, the trained super-resolution network model is utilized to effectively optimize the image spatial resolution, and it is ensured that the system image resolution is not affected by the change of the FOV size. In an actual application scene, the equipment can flexibly adjust the size of an imaging visual field according to the size of an animal, keeps the high resolution of an image all the time, and breaks through the physical limit of conventional equipment hardware. Compared with the prior art, the method has remarkable advantages in the aspects of improving the flexibility and imaging quality of the imaging device, has extremely high practical value and innovativeness, and can be widely applied to the field of animal imaging.

Description

Technical Field

[0001] The present invention belongs to the field of diagnostic imaging technology, and in particular relates to a small animal CBCT imaging device and method with continuously adjustable FOV size based on a super-resolution network. Background Art

[0002] In vivo imaging of small animals plays a key role in life science research and disease diagnosis and treatment. Small animal CT (micro-CT) uses X-ray tomography to analyze the internal structure of animals, providing important evidence for research such as disease model validation and clinical drug development. For example, animal testing centers in universities and research institutes, pet hospitals, and large-scale breeding bases all have an urgent need for high-resolution imaging of the internal structures of small animals.

[0003] However, existing small animal CT systems use a fixed gantry and rotating platform, or a fixed X-ray source and rotating flat-panel detector. The distance between the sample and the X-ray source or X-ray machine remains fixed, which significantly limits small animal imaging due to hardware limitations. Current equipment, constrained by hardware design, suffers from the following core issues:

[0004] 1. Small imaging field of view (FOV): Most systems have a small FOV diameter, which is only suitable for imaging of ex vivo tissues or mice / rats, and cannot support in vivo research on larger animals such as rabbits. 2. The FOV size is fixed, which cannot support long-term longitudinal research on animals, limiting research. 3. Insufficient resolution: Physical limitations of the hardware make it difficult to meet the imaging accuracy requirements for analyzing fine structures such as microvessels and tumors. 4. High radiation doses limit repeated scans, hindering long-term observation. 5. The equipment is bulky and complex to operate, making it difficult to adapt to diverse experimental scenarios.

[0005] Existing small animal CT urgently needs breakthroughs in imaging quality, flexibility and functional scalability. It is urgent to develop a CBCT system optimized for small animals. Through adjustable FOV design, low-dose imaging algorithms and high-resolution reconstruction technology, it can meet the scientific research and clinical needs for accurate, safe and adaptable in vivo imaging. Summary of the Invention

[0006] To address the above technical issues, the present invention provides a small-animal CBCT imaging device and method with continuously adjustable FOV size based on a super-resolution network. Through a unique structural design and control approach, this device achieves continuously adjustable imaging field of view (FOV) size. Simultaneously, a trained super-resolution network model is utilized to effectively optimize image spatial resolution, ensuring that system image resolution is not affected by changes in FOV size. In practical applications, this device can flexibly adjust the imaging field of view size based on animal size while maintaining consistent image high resolution, surpassing the physical limitations of conventional device hardware.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect, the present invention provides a small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network, comprising a housing, a base plate, a support structure, a sliding guide rail, a ray emitting device, a flat panel detector (receiving device), a rotating object loading device, a translation device, and a control system;

[0009] The bottom plate is 130 cm long, 50 cm wide and 5 cm thick. The entire device is fixed on the bottom plate and placed inside the metal shell. The shell is provided with an openable flip cover.

[0010] The support structure comprises a fixed support with a specific structure and a connecting support;

[0011] The sliding guide rails are two parallel rails installed on the bottom plate; the guide rails are provided with limiters;

[0012] The flat panel detector is fixed to one end of the base via the fixing bracket;

[0013] The ray emitting device is mounted on the sliding guide rail via the connecting bracket. The ray emitting device includes a ray source and a high-voltage power supply. The center of the beam outlet of the ray source faces the center of the flat panel detector. The ray source can move horizontally on the guide rail.

[0014] The rotating loading device includes a loading platform, a rotating device, a lifting device, and a connecting portion, and is installed between the ray emitting device and the flat panel detector through the connecting portion, and the two ends of the connecting portion are clamped on the sliding guide rail;

[0015] The moving module comprises a ball screw and a drive motor, wherein the ball screw is parallel to the guide rail, the drive motor is fixed to one end of the base plate, and the output shaft of the drive motor is connected to the ball screw via a coupling;

[0016] The loading platform is used to place small animals and can be replaced according to the size of the animals; the connecting portion is connected to the nut of the lead screw nut, and the driving motor drives the ball screw nut to move, so that the lead screw nut drives the rotating loading device to move horizontally on the guide rail, and the movement mode can be motor-driven or manual; when the loading platform moves horizontally to the upper limit stop of the guide rail, it stops to prevent collision with the flat panel detector;

[0017] The rotating device is arranged under the loading platform, and includes a driving motor and a rotating platform. The driving motor is fixed on the side of the rotating platform. The rotating platform is marked with a scale. The rotating platform is fixed directly under the loading platform to drive the loading platform to rotate 360 ​​degrees. Preferably, the rotating platform is provided with an angle sensor for displaying the rotation angle.

[0018] The lifting device is arranged below the rotating device to control the up and down movement of the loading platform, and includes a driving motor and a ball screw. The driving motor drives the rotating device and the loading platform to move up and down.

[0019] The translation device includes a ball screw, a screw nut and a drive motor, wherein the ball screw is parallel to the guide rails in the middle of the two guide rails, the drive motor is fixed to one end of the base plate, and the output shaft of the drive motor is connected to the ball screw through a coupling;

[0020] The nut of the ball screw is connected to the lower end of the connecting part of the rotating carrier device. The driving motor drives the ball screw nut to move, so that the screw nut drives the entire rotating carrier device to move horizontally on the guide rail. The movement mode can be motor-driven or manually controlled. A limiter is provided on the guide rail, and when the carrier platform moves horizontally to the limiter, it stops to prevent it from colliding with the flat-panel detector.

[0021] The control system mainly involves hardware control, data acquisition and transmission control, imaging algorithm and user interaction software.

[0022] The hardware control includes the control of the ray emitting device, the rotating object carrying device and the translation device; it is mainly achieved through the ray source control module and the motor drive module.

[0023] The data acquisition and transmission control mainly acquires data from the flat panel detector and transmits the acquired projection data to the workstation; preferably, serial communication is used; each time the rotating platform rotates an angle, the flat panel detector is triggered to acquire projection data once.

[0024] The imaging algorithm is deployed on a workstation and is used to run the imaging algorithm, process the collected data, and realize three-dimensional reconstruction of the image.

[0025] The user interaction software provides a user interface that displays the reconstructed three-dimensional image and can store all image data. The user interaction software allows the user to operate the system and set parameters. Specifically, the user can use the software to set the X-ray source tube voltage and tube current, and control the system power on and off. The user can also use the software to precisely control the horizontal movement of the rotating stage.

[0026] By continuously and arbitrarily adjusting the horizontal position of the stage (or radiation source), the distance between the stage (or radiation source) and the detector is changed, and the geometric magnification is adjusted to achieve the effect of continuous adjustment of the imaging field of view FOV.

[0027] In a second aspect, the present invention provides a small animal CBCT imaging method with continuously adjustable FOV size based on a super-resolution network, which is applied to the aforementioned imaging device, comprising:

[0028] Step 1. Fix the object to be inspected on the stage; ensure that the rotation center of the stage and the center of the radiation source and detector are in a straight line;

[0029] Step 2: Obtain high-resolution images: Move the rotating stage horizontally until the ratio M (geometric magnification) of the distance SDD from the rotating stage to the radiation source to the distance SOD from the radiation source to the flat panel detector satisfies M=1+ When (a) the focal size of the device's ray source, d the pixel size of the flat-panel detector, the resolution is at its best, and a high-resolution projection image is obtained;

[0030] Step 3, obtaining a low-resolution image: moving the rotating object-carrying device to any other position to obtain a low-resolution projection image;

[0031] Step 4. Dataset construction: Repeat steps 1-3 above for different detection objects to obtain a sufficient number of images (about 20,000);

[0032] Step 5: Data preprocessing: The obtained image data is uniformly preprocessed, including denoising, geometric correction, and normalization; denoising: non-local mean filtering, bilateral filtering, or wavelet denoising is used to remove noise in the image, improve the signal-to-noise ratio of the image, and enhance the visual quality of the image.

[0033] Geometric correction: Due to factors such as mechanical errors in the device and the placement of objects, the captured image may have geometric distortion. Correction methods based on feature point matching are used to restore the image to its correct geometric shape. Normalization: Normalizes the image's pixel values ​​to a specific range, such as [0, 1] or [-1, 1]. This eliminates differences in brightness and contrast between images, allowing the model to better learn the image's features.

[0034] Step 6. Dataset Partitioning: The preprocessed image dataset is divided into training set, validation set, and test set according to a certain ratio. The partitioning process follows the principle of random sampling. The training set accounts for 80% of the total dataset and is used for model training. The validation set accounts for 10% and is used to evaluate the performance of the model during training and adjust the model's hyperparameters. The test set accounts for 10% and is used to finally evaluate the model's generalization ability and imaging effect.

[0035] Step 7: Build the SRGAN deep learning model: This includes a generator network and a discriminator network. The generator network learns the mapping relationship from low-resolution CBCT images to high-resolution CBCT images. The discriminator network is responsible for determining whether the input image is a true high-resolution image or a pseudo-high-resolution image generated by the generator network.

[0036] Step 8. Construct a combined loss function: a combined loss function including perceptual loss, generative adversarial loss, and mean absolute error loss, L = αL perceptual + βL adversarial + γL MAE , α, β and γ are weight coefficients;

[0037] Step 9: Model training: Use the training set and the combined loss function to iterate the parameters of the generated network.

[0038] 1. Randomly select a batch of low-resolution images and their corresponding real high-resolution images from the training set.

[0039] 2. Input the low-resolution image into the generative network to generate a pseudo high-resolution image.

[0040] 3. Input the generated pseudo high-resolution image and the real high-resolution image into the discriminator network respectively, and calculate the output of the discriminator.

[0041] 4. Calculate the loss values ​​of the generator network and the discriminator network based on the output of the discriminator and the combined loss function.

[0042] 5. Use the Adam optimizer to update the parameters of the generator and discriminator networks to minimize the loss value.

[0043] Step 10. Model Validation: Use the validation set to evaluate the model's performance, calculate the model's loss and evaluation metrics (peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM)) on the validation set, and adjust the model's hyperparameters (such as learning rate, weight coefficient, network structure, etc.) to optimize performance.

[0044] Step 11: Model testing: Use the test set to evaluate the model's generalization ability and super-resolution imaging effect; ensure that the model can stably convert low-resolution images into high-resolution images, and that the output high-resolution images have clear details and reasonable organizational structure;

[0045] Step 12: Deploy the trained model to the imaging algorithm software.

[0046] The beneficial effects of the present invention are:

[0047] The animal CT imaging device provided by this invention features a horizontally movable radiation source and stage. The animal is fixed on a rotating stage, and the stage or radiation source is continuously moved horizontally. Through a unique structural design and control method, the imaging field of view (FOV) can be continuously adjusted. A trained super-resolution network model is used to optimize image spatial resolution, ensuring that the system's image resolution is unaffected by changes in FOV size. In practical applications, the imaging field of view can be arbitrarily adjusted based on the animal's size while maintaining high image resolution. This breaks the physical limitations of conventional hardware and offers significant practical value and innovation. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A three-dimensional schematic diagram of a small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to the present invention (without the housing);

[0049] Figure 2 This is a front view of a small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to the present invention;

[0050] Figure 3 A three-dimensional schematic diagram (with housing) of a small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to the present invention;

[0051] Figure 4 This is a flow chart of a small animal CBCT imaging method with continuously adjustable FOV size based on a super-resolution network according to the present invention;

[0052] Figure 5 This is a system layout diagram of an embodiment of the present invention.

[0053] Reference numerals:

[0054] 1. Base plate; 2. Radiation source; 3. Flat-panel detector; 4. Stage; 5. High-voltage power supply; 6. Connecting bracket; 7. Fixed bracket; 8. Rotating platform; 9. Lifting platform; 10. Connecting part; 11. Screw nut; 12. Translation drive motor; 13. Sliding guide rail; 14. Ball screw; 15. Rotation drive motor; 16. Limiter; 17. Overall housing; 18. Flip cover; 19. Control system; 20. Power supply. DETAILED DESCRIPTION

[0055] The present invention will be further described below with reference to the accompanying drawings and examples.

[0056] In order to make the above-mentioned objects, features and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention.

[0057] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like to indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0058] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0059] In the present invention, unless otherwise expressly specified or limited, when a first feature is "above" or "below" a second feature, it may mean that the first and second features are in direct contact, or that the first and second features are in indirect contact through an intermediary. Furthermore, when a first feature is "above," "above," or "above" a second feature, it may mean that the first feature is directly above or diagonally above the second feature, or simply means that the first feature is at a higher level than the second feature. When a first feature is "below," "below," or "below" a second feature, it may mean that the first feature is directly below or diagonally below the second feature, or simply means that the first feature is at a lower level than the second feature.

[0060] It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it may be directly on the other element or there may be an intermediate element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be an intermediate element. The terms "vertical," "horizontal," "upper," "lower," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only implementation methods.

[0061] like Figures 1 to 3 As shown, the present invention provides a small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network, comprising a base plate, a support structure, a flat panel detector, a ray emitting device, a rotating object carrier, a mobile module and a control system, wherein:

[0062] The bracket structure fixes the ray emitting device and the flat panel detector to the two ends of the bottom plate respectively; the ray emitting device is used to emit a ray source toward the object to be imaged, and the flat panel detector is used to collect projection data passing through the object to be imaged;

[0063] The rotating object-carrying device is installed between the ray emitting device and the flat panel detector, and can rotate 360 ​​degrees and rise and fall in the vertical direction;

[0064] The moving module is used to drive the rotating loading device to slide in the horizontal direction;

[0065] The control system is used to dynamically adjust the rotating object loading device or the ray emitting device to change the imaging field of view, collect projection data of the flat panel detector in real time, and realize three-dimensional reconstruction of the image based on the projection data.

[0066] The equipment base plate 1 is 130cm long, 50cm wide and 5cm thick. It serves as the supporting base of the entire equipment. All components are fixed on the base plate 1 and placed as a whole in the overall shell 17, which has a shielding effect. The overall shell 17 is provided with an openable flip cover 18 to facilitate experimental operations.

[0067] The bracket structure consists of a fixed bracket 7 and a connecting bracket 6 of a specific structure. Two parallel sliding guide rails 13 are installed on the left and right sides of the base plate 1 respectively. The flat panel detector 3 is fixed to one end of the base plate 1 through the fixed bracket 7, and the radiation source 2 is installed on the sliding guide rail 13 through the connecting bracket 6. The radiation source 2 can move horizontally along the sliding guide rail 13.

[0068] The radiation source 2 is powered by a high-voltage power supply 5, and the center of the beam outlet of the radiation source 2 is precisely aligned with the center of the flat panel detector 3;

[0069] The imaging device includes a rotating loading device, which includes a loading platform 4, a rotating device, a lifting device, and a connecting portion 10. The connecting portion 10 is installed between the radiation source 2 and the flat panel detector 3. The ends of the connecting portion 10 are clamped on the sliding guide rails 13 to achieve flexible movement. The loading platform 4 is used to place small animals and can be replaced according to the size of the experimental animal.

[0070] A rotating device is provided below the stage 4, which consists of a rotary drive motor 15 and a rotary platform 8. The rotary drive motor 15 is fixed to the side of the rotary platform 8. The rotary platform 8 is marked with a scale and fixed directly below the stage 4, and can drive the stage 4 to rotate 360 ​​degrees. Preferably, the rotary platform 8 is provided with an angle sensor to provide real-time feedback on the rotation angle to ensure rotation accuracy.

[0071] A lifting platform 9 is provided below the rotating device, which can control the up and down movement of the loading platform 4 and can be adjusted in real time according to the size of the animal and the desired imaging part.

[0072] The translation device includes a ball screw 14, a screw nut 11 and a translation drive motor 12; the ball screw 14 is located between the two sliding guide rails 13 and is parallel to the two sliding guide rails 13. The translation drive motor 12 is fixed to one end of the base plate 1, and its output shaft is connected to the ball screw 14 through a coupling. The screw nut 11 is connected to the lower end of the connecting part 10 of the rotating load device. Driven by the motor or manually controlled, the screw nut 11 drives the entire rotating load device to move horizontally on the guide rail; when it moves to the limiter 16, it stops to prevent collision with the flat panel detector 3.

[0073] The control system 19 primarily involves hardware control, data acquisition and transmission control, imaging algorithms, and user interaction software. Hardware control includes control of the ray emission device, rotating object carrier, and translation device, primarily achieved through the ray source control module and motor drive module.

[0074] The data acquisition and transmission control mainly acquires data from the flat panel detector 3 and transmits the acquired projection data to the workstation; preferably, serial communication is used; each time the rotating platform 8 rotates an angle, the flat panel detector 3 is triggered to acquire projection data once.

[0075] The imaging algorithm is deployed on the workstation to run the imaging algorithm, process the collected data, and realize three-dimensional reconstruction of the image.

[0076] The user interaction software provides a user interface that displays the reconstructed 3D image and can store all image data. The user interaction software allows the user to operate the system and set parameters. Specifically, the user can use the software to set the X-ray source tube voltage and tube current, and control the system power supply 20 on and off. The user can also use the software to precisely control the horizontal movement of the rotating stage.

[0077] In actual applications, the operator can precisely control the translation device through software to move the stage 4 closer to or away from the flat-panel detector 3, or move the radiation source 2 to adjust the distance between it and the flat-panel detector 3, thereby achieving continuous adjustment of the imaging field of view (FOV) size, so that the object to be imaged is completely within the imaging field of view (FOV), meeting the diverse requirements of different experiments for the imaging field of view.

[0078] like Figure 4 As shown, an embodiment of the present invention provides a small animal CBCT imaging method with continuously adjustable FOV size based on a super-resolution network.

[0079] The imaging method comprises the following steps:

[0080] Step 1: Fix the object to be tested and calibrate the equipment: Fix the object to be tested on the stage; ensure that the rotation center of the stage and the center of the radiation source and detector are in a straight line;

[0081] Step 2: Obtain high-resolution images: Move the rotating stage horizontally until the ratio M (geometric magnification) of the distance SDD from the rotating stage to the radiation source to the distance SOD from the radiation source to the flat panel detector satisfies M=1+ When (a) the focal size of the device's ray source, d the pixel size of the flat-panel detector, the resolution is at its best, and a high-resolution projection image is obtained;

[0082] By horizontally moving the rotating stage or radiation source, the distance between the stage and the radiation source is changed to adjust the FOV size. However, adjusting the FOV size in this way will change the geometric magnification and the image spatial resolution. It is known that the spatial resolution is affected by the focal size a of the radiation source, the pixel size d of the flat-panel detector, and the geometric magnification M of the device:

[0083] ,

[0084] The focal size a of the device's radiation source and the pixel size d of the flat-panel detector are fixed. The geometric magnification M is the ratio of the distance between the radiation source and the flat-panel detector (SDD) to the distance between the radiation source and the rotating object device (SOD):

[0085] ,

[0086] Changing the distance SOD between the rotating stage and the radiation source changes M, thereby changing the resolution. When the rotating stage is closer to the radiation source, a smaller imaging field of view and higher spatial resolution can be obtained, while when the lifting stage is farther away from the radiation source, a larger imaging field of view and lower spatial resolution can be obtained.

[0087] After simplification, when M=1+ , the spatial resolution is the smallest, .

[0088] Step 3, obtaining a low-resolution image: moving the rotating object-carrying device to any other position to obtain a low-resolution projection image;

[0089] Step 4. Dataset construction: Repeat steps 1-3 above for different detection objects to obtain a sufficient number of images (about 20,000);

[0090] Step 5: Data preprocessing: The obtained image data is uniformly preprocessed, including denoising, geometric correction, and normalization; denoising: non-local mean filtering, bilateral filtering, or wavelet denoising is used to remove noise in the image, improve the signal-to-noise ratio of the image, and enhance the visual quality of the image.

[0091] Geometric correction: Due to factors such as mechanical errors in the device and the placement of objects, the captured image may have geometric distortion. Correction methods based on feature point matching are used to restore the image to its correct geometric shape. Normalization: Normalizes the image's pixel values ​​to a specific range, such as [0, 1] or [-1, 1]. This eliminates differences in brightness and contrast between images, allowing the model to better learn the image's features.

[0092] Step 6. Dataset Partitioning: The preprocessed image dataset is divided into training set, validation set, and test set according to a certain ratio. The partitioning process follows the principle of random sampling. The training set accounts for 80% of the total dataset and is used for model training. The validation set accounts for 10% and is used to evaluate the performance of the model during training and adjust the model's hyperparameters. The test set accounts for 10% and is used to finally evaluate the model's generalization ability and imaging effect.

[0093] Step 7: Build the SRGAN deep learning model: This includes a generator network and a discriminator network. The generator network learns the mapping relationship from low-resolution CBCT images to high-resolution CBCT images. The discriminator network is responsible for determining whether the input image is a true high-resolution image or a pseudo-high-resolution image generated by the generator network.

[0094] Step 8. Construct a combined loss function: a combined loss function including perceptual loss, generative adversarial loss, and mean absolute error loss, L = αL perceptual + βL adversarial + γL MAE , α, β and γ are weight coefficients;

[0095] Step 9: Model training: Use the training set and the combined loss function to iterate the parameters of the generated network.

[0096] 1. Randomly select a batch of low-resolution images and their corresponding real high-resolution images from the training set.

[0097] 2. Input the low-resolution image into the generative network to generate a pseudo high-resolution image.

[0098] 3. Input the generated pseudo high-resolution image and the real high-resolution image into the discriminator network respectively, and calculate the output of the discriminator.

[0099] 4. Calculate the loss values ​​of the generator network and the discriminator network based on the output of the discriminator and the combined loss function.

[0100] 5. Use the Adam optimizer to update the parameters of the generator and discriminator networks to minimize the loss value.

[0101] Step 10. Model Validation: Use the validation set to evaluate the model's performance, calculate the model's loss and evaluation metrics (peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM)) on the validation set, and adjust the model's hyperparameters (such as learning rate, weight coefficient, network structure, etc.) to optimize performance.

[0102] Step 11: Model testing: Use the test set to evaluate the model's generalization ability and super-resolution imaging effect; ensure that the model can stably convert low-resolution images into high-resolution images, and that the output high-resolution images have clear details and reasonable organizational structure;

[0103] Step 12: Deploy the trained model to the imaging algorithm software.

[0104] In summary, the present invention, through the coordinated operation of various components, not only achieves continuous adjustment of the imaging field of view (FOV) size through a unique structural design and control method, but also utilizes a trained super-resolution network model to optimize the image spatial resolution, ensuring that the system image resolution is not affected by changes in FOV size. It maintains high image resolution while being applicable to animals of different sizes, and has extremely high practical value and innovation.

[0105] like Figure 5 As shown, specifically, the system of the present invention uses a microfocus radiation source and a large flat-panel detector. The microfocus radiation source is a radiation source with a very small focal spot size a, such as the L11831-01 microfocus source from Hamamatsu, Japan, with a tube voltage of 40-90 kV, an irradiation window to focal spot distance of 11 mm, and a focal spot size of 5 μm; the CMOS flat-panel detector from Hamamatsu, Japan, has a pixel size of 200 μm.

[0106] According to the unique structural design and control method of the present invention, by precisely controlling the translation device, the stage 4 is moved closer to or away from the flat-panel detector 3, or the radiation source 2 is moved to adjust the distance between the stage 4 and the flat-panel detector 3, thereby achieving continuous adjustment of the imaging field of view (FOV) from 10 to 400 mm to meet the imaging requirements of animals of various sizes.

[0107] Then, 20,000 images with the stage at different positions (with different FOV sizes) were collected to train the super-resolution network, and the trained model was deployed on the imaging algorithm software, so that the image resolution of the system of the present invention was always maintained at the optimal level, i.e., 25 μm.

[0108] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network, characterized in that: It includes a base plate, a support structure, a flat panel detector, a ray emitting device, a rotating object carrying device, a translation device and a control system, wherein: The bracket structure fixes the ray emitting device and the flat panel detector to the two ends of the bottom plate respectively; the ray emitting device is used to emit a ray source toward the object to be imaged, and the flat panel detector is used to collect projection data passing through the object to be imaged; The rotating object-carrying device is installed between the ray emitting device and the flat panel detector, and can rotate 360 ​​degrees and rise and fall in the vertical direction; The translation device is used to drive the rotating object-carrying device to slide in the horizontal direction; The control system is used to dynamically adjust the rotating object loading device or the ray emitting device to change the imaging field of view, collect projection data of the flat panel detector in real time, and realize three-dimensional reconstruction of the image based on the projection data.

2. A small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 1, characterized in that: The support structure includes a fixing support and a connecting support; the flat panel detector is fixed to one end of the base plate through the fixing support, and the ray emitting device is installed on the other end of the base plate through the connecting support.

3. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 2, characterized in that: The ray emitting device is mounted on two parallel sliding guide rails on the base plate through the connecting bracket, and the sliding guide rails are provided with limiters.

4. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 2, characterized in that: The ray emitting device includes a ray source, and the center of the beam outlet of the ray source faces the center of the flat panel detector.

5. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 2, characterized in that: The rotating loading device includes a loading platform, a rotating device, a lifting device, and a connecting part. The loading platform is installed on the sliding guide rail through the connecting part to achieve horizontal movement.

6. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 5, characterized in that: The rotating device includes a rotating drive motor and a rotating platform. The rotating drive motor is fixed on the side of the rotating platform. The rotating platform is marked with a scale. The rotating platform is fixed just below the loading platform to drive the loading platform to rotate 360 ​​degrees.

7. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 6, characterized in that: The lifting device is arranged below the rotating device and drives the rotating device and the loading platform to move up and down.

8. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 6, characterized in that: The rotating platform is provided with an angle sensor for displaying the rotation angle.

9. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 2, characterized in that: The translation device includes a ball screw, a screw nut and a translation drive motor. The ball screw is located between the sliding guide rails and is parallel to the sliding guide rails. The translation drive motor is fixed to one end of the base plate. The output shaft of the translation drive motor is connected to the ball screw through a coupling. The connecting part of the rotating carrier device is connected to the nut of the screw nut.

10. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 1, characterized in that: The control system includes hardware control, data acquisition and transmission control, imaging algorithm and user interaction software.

11. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 10, characterized in that: The hardware control includes the control of the ray emitting device, the rotating object loading device and the translation device.

12. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 10, characterized in that: The data acquisition and transmission control is used to acquire projection data from the flat panel detector and transmit the acquired projection data to the workstation; each time the rotating platform rotates an angle, the flat panel detector is triggered to acquire projection data once.

13. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 10, characterized in that: The imaging algorithm is deployed on a workstation and is used to process the collected data to achieve three-dimensional reconstruction of the image.

14. The small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network according to claim 10, characterized in that: The user interaction software provides a user interface, displays the reconstructed three-dimensional image, sets the current and voltage of the ray source, controls the system power switch, and precisely controls the horizontal movement of the rotating stage.

15. A small animal CBCT imaging method with continuously adjustable FOV size based on a super-resolution network, characterized in that: Using a small animal CBCT imaging device with continuously adjustable FOV size based on a super-resolution network as described in any one of claims 1 to 14, the method comprises the following steps: Step 1: Fix the object to be imaged on the stage; Step 2: Obtain high-resolution images: Move the rotating stage horizontally until the ratio M of the distance SDD from the rotating stage to the radiation source to the distance SOD from the radiation source to the flat panel detector satisfies M=1+ When , a high-resolution projection image is obtained, a represents the focal size of the device's ray source, and d represents the pixel size of the flat-panel detector; Step 3, obtaining a low-resolution image: moving the rotating object-carrying device to any other position to obtain a low-resolution projection image; Step 4: Dataset construction: Repeat steps 1 to 3 above for different objects to be imaged to obtain a sufficient number of images to construct an image dataset. Step 5: Data preprocessing: preprocess the obtained image data sets uniformly; Step 6, data set division: divide the preprocessed image data set into training set, validation set and test set in proportion; Step 7: Construct an SRGAN deep learning model: The SRGAN deep learning model includes a generator network and a discriminator network; the generator network learns the mapping relationship from low-resolution CBCT images to high-resolution CBCT images; The discriminator network determines whether the input image is a real high-resolution image or a pseudo high-resolution image generated by the generator network; Step 8: Construct a combined loss function; Step 9: Model training: Use the training set and the combined loss function to iterate the parameters of the generated network. Step 10. Model Validation: Use the validation set to evaluate the performance of the SRGAN deep learning model, calculate the model loss and evaluation metrics on the validation set, and adjust the model hyperparameters to optimize performance. Step 11: Model testing: Use the test set to evaluate the model’s generalization ability and super-resolution imaging effect; Step 12: Deploy the trained model to the imaging algorithm software.

Images