A CBCT large field of view reconstruction method and system

Abstract

This invention provides a CBCT large field-of-view reconstruction method and system, relating to the field of medical imaging technology. The CBCT large field-of-view reconstruction system includes a scanning module, a data acquisition and correction module, a reference frame extraction module, a global parameter calculation module, a projection domain fusion module, a geometric correction parameter readjustment module, and a large field-of-view image reconstruction module. The scanning module acquires imaging commands and performs a first rotational scan on the target object according to a preset timing sequence to acquire the lower region projection dataset. Then, under the same exposure parameters, it raises the detector and performs a second rotational scan on the target object to acquire the upper region projection dataset. This invention achieves integrated reconstruction of large field-of-view three-dimensional image data using a lifting mechanical structure and an imaging stage, obtaining a larger imaging field of view in the longitudinal direction with low-cost modifications to existing CBCT scanning equipment.

Description

Technical Field

[0001] This invention relates to the field of medical imaging technology, specifically to a CBCT large field-of-view reconstruction method and system. Background Technology

[0002] Dental CBCT (Cone Beam Computer Tomography) is widely used in the medical field due to its compact size, lightweight design, excellent imaging resolution and low radiation dose. However, due to the limitations of the magnification and detector height of the CBCT system, the imaging field in the longitudinal direction of CBCT is difficult to meet the needs of oral fields such as orthodontics, maxillofacial surgery and full mouth implantation.

[0003] Among existing methods, spiral-based CBCT can solve the problem of insufficient longitudinal imaging field of view, but it involves significant costs in modifying existing CBCT scanning equipment, and the stability of the scanning equipment requires strict control. CBCT based on a stitching method acquires two sets of data (upper and lower) by raising and lowering the rotating arm of the scanning equipment, and then registers and fuses the reconstructed images. This approach is less expensive, but requires that the lifting mechanism not be linked to the imaging stage; that is, the scanning gantry must have independent lifting capabilities.

[0004] Therefore, this invention proposes a CBCT large field of view reconstruction method and system, which effectively expands the longitudinal imaging field of view without making major modifications to existing CBCT scanning equipment and without the scanning gantry having independent lifting function, thus becoming a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] To address the shortcomings of existing technologies, a method and system for large field-of-view reconstruction in CBCT is provided. This method solves the problem of insufficient longitudinal imaging field of view when the scanning gantry of existing CBCT equipment does not have an independent lifting function. It achieves a larger longitudinal imaging field of view while reducing equipment modification costs and motion control requirements.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A CBCT large field-of-view reconstruction system includes a scanning module, a data acquisition and correction module, a reference frame extraction module, a global parameter calculation module, a projection domain fusion module, a geometric correction parameter readjustment module, and a large field-of-view image reconstruction module. The scanning module is used to acquire imaging instructions, perform a first rotational scan on the target object according to a preset time sequence to acquire the lower region projection dataset, and perform a second rotational scan on the target object after raising the detector under the same exposure parameter conditions to acquire the upper region projection dataset. The data acquisition and correction module is used to acquire air and preset upper and lower region projection datasets of the calibration phantom, and to solve the geometric correction parameters of the CBCT scanning device using a preset geometric correction algorithm based on the lower region projection dataset of the calibration phantom. The reference frame extraction module is used to extract a first reference projection frame at a preset scanning angle from the lower region projection dataset of the calibration phantom, and to extract a second reference projection frame at a reverse scanning angle from the upper region projection dataset of the calibration phantom. The global parameter calculation module is used to calculate the global displacement distance and global scanning error angle of the second reference projection frame relative to the first reference projection frame based on the first reference projection frame and the second reference projection frame using a projection matrix decomposition algorithm. The projection domain fusion module is used to construct a projection domain fusion model using global displacement distance and global scanning error angle, and to perform fusion and stitching processing on the upper and lower region projection datasets through the projection domain fusion model to obtain an expanded projection dataset. The geometric correction parameter readjustment module is used to perform displacement adjustment processing on the geometric correction parameters of the CBCT scanning device using global displacement distance, so as to obtain the geometric correction parameters corresponding to the expanded projection dataset. The large field-of-view image reconstruction module is used to obtain large field-of-view three-dimensional image data based on the extended projection dataset and the geometric correction parameters of the extended projection dataset, using a preset FDK reconstruction algorithm.

[0007] A CBCT large field-of-view reconstruction method, characterized in that: the method is implemented using the CBCT large field-of-view reconstruction system described in claim 1, and the technical solution of the method is as follows: Step 1: Perform a first rotational scan on the target object according to the preset timing sequence to collect the lower region projection dataset, and then perform a second rotational scan on the target object after raising the detector under the same exposure parameters to collect the upper region projection dataset. Step 2: According to the above imaging instructions, acquire air and preset upper and lower region projection datasets of the calibration phantom. Based on the lower region projection dataset of the calibration phantom, use the preset geometric correction algorithm to solve the geometric correction parameters of the CBCT scanning device. Step 3: Extract the first reference projection frame at the preset scanning angle from the lower region projection dataset of the calibration phantom, and extract the second reference projection frame at the reverse scanning angle from the upper region projection dataset of the calibration phantom. Step 4: Based on the first and second reference projection frames, calculate the global displacement distance and global scanning error angle of the second reference projection frame relative to the first reference projection frame using the projection matrix decomposition algorithm; Step 5: Construct a projection domain fusion model using global displacement distance and global scanning error angle, and perform fusion and stitching processing on the upper and lower region projection datasets through the projection domain fusion model to obtain the expanded projection dataset; Step 6: Perform displacement adjustment processing on the geometric correction parameters of the CBCT scanning device using the global displacement distance to obtain the geometric correction parameters corresponding to the expanded projection dataset; Step 7: Based on the extended projection dataset and the geometric correction parameters of the extended projection dataset, obtain large field-of-view 3D image data through the preset FDK reconstruction algorithm; The above technical solution involves raising the detector to collect projection data of the upper and lower regions of the target object, and using projection matrix decomposition and image fusion techniques to obtain expanded projection data, thus compensating for the lack of imaging data in the vertical direction. Then, the geometric correction parameters are readjusted to adapt to the expanded projection data, and finally, the preset FDK algorithm is used to reconstruct the large field-of-view 3D image data, which can meet the needs of a larger imaging field of view at a lower cost.

[0008] Further, step 1, which involves performing a first rotational scan on the target object according to a preset timing sequence to acquire the lower region projection dataset, and then performing a second rotational scan on the target object to acquire the upper region projection dataset after raising the detector under the same exposure parameters, specifically includes: The first rotational scan starts from the pre-acceleration point and performs forward acceleration. When the preset exposure point is reached, the exposure is turned on, and the exposure and rotational scan stop when the scan angle is detected to have reached the preset constant threshold. After completing the first rotational scan, the detector is first raised to the preset position, and then the scanning position is controlled to retreat along the reverse trajectory to the preparatory acceleration point that has passed the starting position of the first rotational scan, while maintaining the same exposure parameters and starting the second reverse rotational scan when the scanning position reaches the starting position. The above technical solution enables the acquisition of data from the upper and lower regions of the target object in the longitudinal direction through two rotational scans in both directions and detector lifting, so as to facilitate subsequent geometric correction and projection domain fusion processing.

[0009] Further, step 4, which calculates the global displacement distance and global scanning error angle of the second reference projection frame relative to the first reference projection frame using a projection matrix decomposition algorithm based on the first and second reference projection frames, specifically includes: After performing physical corrections on the first and second reference projection frames respectively, the two-dimensional image coordinates of the steel ball in the first and second reference projection frames are extracted by combining the adaptive threshold segmentation algorithm and the fixed threshold segmentation algorithm. Based on the above two-dimensional image coordinates, the elliptical angle sorting method is used to sort the steel balls, and the average pixel displacement distance of the y coordinate of the steel balls after the first and second upper-level sorting is calculated, which is the global displacement distance. The first and second projection matrices are calculated based on the sorted two-dimensional image coordinates and the preset three-dimensional spatial coordinates of the steel ball, respectively. The first and second rotational scanning angles are calculated by the projection matrix decomposition algorithm to obtain the global scanning error angle. By performing physical correction, steel ball sorting, and projection matrix decomposition on the reference projection frame, the global displacement distance of the detector lift and the global scanning error angle of the two forward and reverse rotation scans are obtained, thereby improving the accuracy of subsequent projection domain fusion and the resolution of the reconstructed image.

[0010] Further, step 6 constructs a projection domain fusion model using global displacement distance and global scanning error angle, and performs fusion and stitching processing on the upper and lower region projection datasets through the projection domain fusion model to obtain the expanded projection dataset, specifically including: Physical corrections are performed on the projection data of the first and second rotating scans, and the projection data of the second rotating scan is reversed. After calculating the number of error frames for forward and reverse rotation scanning based on the global scanning error angle, the forward and reverse rotation scanning projection data alignment process is performed. The centerline coordinates of the projection domain fusion are determined based on the global displacement distance, and the fusion weights are calculated using a piecewise function. Based on the aligned forward and reverse rotation scan projection data and fusion weights, the preset overlapping regions are fused and stitched to obtain the expanded projection dataset. The above technical solution obtains the specific overlapping area and fusion weight through global displacement distance, aligns the forward and reverse rotation scanning projection data through global scanning error angle, and performs fusion and stitching processing to obtain an expanded projection dataset to make up for the problem of insufficient projection data in the vertical direction.

[0011] This invention provides a method and system for large-field-of-view reconstruction in CBCT. It has the following beneficial effects: 1. This invention provides a CBCT large field-of-view reconstruction method and system, which realizes the integrated reconstruction of large field-of-view three-dimensional image data by integrating the lifting mechanical structure and the imaging object stage. Under the premise of low-cost modification of existing CBCT scanning equipment, a larger imaging field of view in the longitudinal direction is obtained.

[0012] 2. This invention provides a CBCT large field-of-view reconstruction method and system, which corrects the motion error of the CBCT scanning equipment by calculating the global displacement distance and the global scanning error angle, thereby reducing the requirements of this scheme on the motion control of the CBCT scanning equipment. Attached Figure Description

[0013] Figure 1 This invention specifically describes the detailed process of the large field-of-view reconstruction system performing the large field-of-view reconstruction method; Figure 2 A simplified flowchart illustrating the large field-of-view reconstruction method performed by the large field-of-view reconstruction system of the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Example 1: This application discloses a CBCT large field-of-view reconstruction method (hereinafter referred to as the large field-of-view reconstruction method), the execution subject of which is a CBCT large field-of-view reconstruction processing system (hereinafter referred to as the large field-of-view reconstruction system). The following will describe it in conjunction with the attached... Figure 1 This section details the process of implementing the large-field-of-view reconstruction method in the large-field-of-view reconstruction system. Step 1: Perform a first rotational scan on the target object according to the preset timing sequence to collect the lower region projection dataset, and then perform a second rotational scan on the target object after raising the detector under the same exposure parameters to collect the upper region projection dataset. This step aims to construct the raw data foundation required for large-field-of-view reconstruction images through time-division scanning and detector lifting strategies. Specifically, for the same target object, the field of view required for large-field-of-view reconstruction images is constructed sequentially using a single-tube imaging system and a detector lifting system. In conjunction with rotational scanning motion, the first set of projection data and the second set of projection data corresponding to the upper and lower regions of the target object are acquired respectively, thereby providing source data for subsequent projection domain fusion and geometric correction parameter readjustment. In one possible implementation, according to a preset timing sequence, a first rotational scan is performed on the target object to acquire a lower region projection dataset, and a second rotational scan is performed on the target object to acquire an upper region projection dataset after the detector is raised under the same exposure parameters. Specifically, the first rotational scan starts from a pre-acceleration point and performs forward acceleration. When the preset exposure point is reached, exposure is activated, and exposure and rotational scanning are stopped when the scanning angle is detected to reach a preset constant threshold. After the first rotational scan is completed, the detector is first raised to a preset position, and then the scanning position is controlled to retreat along a reverse trajectory to the pre-acceleration point that has passed the starting position of the first rotational scan. The same exposure parameters are maintained, and a second reverse rotational scan is performed when the scanning position reaches the starting position. In this embodiment, after the preset exposure parameters are set, the first rotational scan is performed. The acceleration point is prepared so that the first rotational scan has a stable scanning angular velocity when it reaches the preset exposure point (such as the 0-degree position). The first set of projection data is collected from the preset exposure point, and the exposure and rotational scan are stopped when the preset constant threshold (such as the 360-degree position) is reached, thus completing the collection of the first set of projection data. After completing the first rotational scan data acquisition, the detector is raised to a preset position (e.g., raised 5cm upwards) by a stepper motor; then, a reverse drive command is generated according to the same exposure parameters to control the scanning position to retreat along the original trajectory; in the retreat logic, according to the pre-acceleration point rule of the previous scan, the scanning frame is controlled to start reverse acceleration from the reverse pre-acceleration point, and exposure is started from the reverse preset exposure point. Exposure and rotational scanning are stopped when the preset constant threshold is reached, thus completing the acquisition of the second set of projection data; Step 2: According to the above imaging instructions, acquire air and preset upper and lower region projection datasets of the calibration phantom. Based on the lower region projection dataset of the calibration phantom, use the preset geometric correction algorithm to solve the geometric correction parameters of the CBCT scanning device. According to the aforementioned imaging instructions, air projection data and the upper and lower regions of a preset calibration phantom are collected respectively. In this embodiment, the collected air projection data is used for physical correction processing of the target object. Since the detector is displaced, it is necessary to collect air projection data of the upper and lower regions, and perform physical correction processing on the first and second sets of projection data respectively. The preset calibration phantom consists of a cylinder and steel balls. Specifically, the cylinder has an outer diameter of 120mm, an inner diameter of 114mm, and a height of 70mm. Three circular planes (top, middle, and bottom) are embedded with steel balls. The top and middle planes are 20mm apart, and the middle and bottom planes are 30mm apart. Each plane evenly embeds 12 steel balls, with the smaller balls having a diameter of 1.5mm and the larger balls (highlighted in red) having a diameter of 2.5mm, for subsequent projection of the steel balls' two-dimensional coordinates. Physical correction specifically includes dead pixel and dead line correction, dark field correction, and gain correction. Geometric correction parameters include the detector's rotation angle (…). Translation and displacement () , ), Source-to-detector distance (SDD), Source-to-object distance (SOD), Scan angle ( ), Magnification factor When there are errors in the geometric correction parameters, image artifacts will appear in the large field-of-view reconstructed image.

[0016] Furthermore, physical correction is performed on the first set of projection data of the pre-calibrated phantom based on the collected first set of projection data of the air, and the geometric correction parameters of the CBCT scanning equipment are solved using the pre-defined geometric correction algorithm. The geometric correction parameters can be used for subsequent geometric correction parameter readjustment and large field-of-view image reconstruction. Step 3: Extract the first reference projection frame at the preset scanning angle from the lower region projection dataset of the calibration phantom, and extract the second reference projection frame at the reverse scanning angle from the upper region projection dataset of the calibration phantom. A reference projection frame is a two-dimensional X-ray projection image with a clearly defined geometric location, extracted from a continuously acquired sequence of projection data. It is used for subsequent calculations of global displacement distance and global scanning error angle. For example, a reference projection frame could be the projection image corresponding to the starting position of the rotational scan (i.e., the 0-degree position). Based on the system's preset alignment strategy, a preset scanning angle (e.g., 0 degrees) is determined. The first set of projection datasets is traversed, and the image frame corresponding to the scanning angle is extracted as the first reference projection frame. Simultaneously, the second set of projection datasets is traversed in the same logic, and the image frame corresponding to the scanning angle is extracted in the reverse direction as the second reference projection frame. Through this process, a pair of reference images that initially correspond in theoretical geometric viewpoint is obtained, providing source data input for subsequent calculations of global displacement distance and global scanning error angle. Step 4: Based on the first and second reference projection frames, calculate the global displacement distance and global scanning error angle of the second reference projection frame relative to the first reference projection frame using the projection matrix decomposition algorithm; This step aims to calculate the displacement distance and scanning error angle of the two rotational scans after the visual quantization detector is lifted. Specifically, this step calculates the projection matrix of the preset three-dimensional spatial coordinates of the steel ball onto the segmented and extracted two-dimensional image coordinates of the steel ball, and decomposes the projection matrix to obtain the first and second rotational scan angles, thereby obtaining the global scanning error angle; the segmented and extracted two-dimensional image coordinates of the steel ball can be further used to calculate the average pixel displacement distance in the y-coordinate direction, which is the global displacement distance, in order to subsequently construct the projection domain fusion model; In one possible implementation, the global displacement distance and global scanning error angle of the second reference projection frame relative to the first reference projection frame are calculated using a projection matrix decomposition algorithm based on the first and second reference projection frames. Specifically, this includes: performing physical correction on the first and second reference projection frames respectively, and then extracting the two-dimensional image coordinates of the steel balls in the first and second reference projection frames using an adaptive threshold segmentation algorithm and a fixed threshold segmentation algorithm; performing steel ball sorting processing using an elliptical angle sorting method based on the above two-dimensional image coordinates, and calculating the average pixel displacement distance of the y-coordinates of the steel balls after the first and second upper-level sorting, which is the global displacement distance; calculating the first and second projection matrices based on the above sorted two-dimensional image coordinates and the preset three-dimensional spatial coordinates of the steel balls respectively, and calculating the first and second rotational scanning angles using a projection matrix decomposition algorithm to obtain the global scanning error angle; After physical correction is performed on the first and second reference projection frames respectively, compensated first and second projection images are generated. The first and second projection images are converted into first and second projection grayscale images, and the two-dimensional image coordinates of the first and second steel balls are extracted from the first and second projection grayscale images by combining adaptive threshold segmentation algorithm and fixed threshold segmentation algorithm respectively. This combined segmentation algorithm can effectively avoid the steel balls being missed in segmentation and has better stability. Furthermore, the two-dimensional image coordinates of the first and second steel balls are classified into layers (based on the upper, middle, and lower layers of the calibration phantom). Then, the ellipse fitting method is used to calculate the ellipse center of each layer of steel balls in the two-dimensional image coordinates of the first and second steel balls, and the angle between the line connecting the center point of each layer to each steel ball point in its respective layer and the horizontal line is calculated. The angles calculated for each layer are sorted from smallest to largest, and the three-dimensional and two-dimensional coordinates of the steel balls in each projection data are aligned according to the coordinates of the red steel balls and the preset index (e.g., the index of the red steel balls in the upper layer is number 1 and the index of the red steel balls in the lower layer is number 7), thus completing the mapping from three-dimensional to two-dimensional coordinates. Finally, the pixel deviation in the y-coordinate direction of the corresponding index of the steel balls after sorting in the first and second upper layers is calculated, and the final average value is taken as the global displacement distance. Based on the mapping from 3D to 2D coordinates, the projection matrix is ​​calculated using Singular Value Decomposition (SVD) on the aligned 3D and 2D coordinates. Projection matrix The calculation formula is as follows: ; in,( Let (u, v) represent the target coordinates, (u, v) represent the detector coordinates, and w represent the range weighting factor. The projection matrix projects the three-dimensional coordinates of the target onto the two-dimensional coordinates of the detector, and represents the geometric characteristics of the system. ; ; ; ; in( Represents the coordinates of the X-ray source Two-dimensional coordinates are orthogonally projected onto the detector; K indicates The matrix contains the source-to-detector distance (SDD) and the orthogonal projection points of the source coordinates; R is represented by three Euler angles. Rotation matrix; It is the rotation angle of the detector around the X-axis. y is the rotation angle of the detector about the y-axis, and y is the rotation angle of the detector about the z-axis; I am The formulas for calculating the identity matrix and the rotation matrix R are as follows: ;

[0017] The formula for solving using three Euler angles is as follows:

[0018] Based on the above projection matrix calculation method, the first and second projection matrices are obtained, and then the obtained y Euler angles are decomposed and used as the first and second rotation scanning angles, respectively. The absolute value of the difference between the two is used as the global scanning error angle. Step 5: Construct a projection domain fusion model using global displacement distance and global scanning error angle, and perform fusion and stitching processing on the upper and lower region projection datasets through the projection domain fusion model to obtain the expanded projection dataset; This step aims to extend the calculated global displacement distance and global scan error angle to the full sequence of projection data, eliminating positional errors in forward scanning and reverse scanning after detector lift-up through mathematical modeling. Specifically, the error frame number of forward and reverse scanning is corrected using the global scan error angle, the displacement distance after detector lift-up is corrected using the global displacement distance, and the projection data from the two scans are fused using weighted fusion to obtain expanded projection data, thereby meeting the need for a larger reconstructed field of view. A projection domain fusion model is constructed using global displacement distance and global scan error angle. This model is then used to perform fusion and stitching processing on the upper and lower region projection datasets to obtain an expanded projection dataset. Specifically, this involves: aligning the projection data from the second rotational scan to the projection data from the first rotational scan using the global scan error angle; determining the centerline coordinates of the fusion stitching using the global displacement distance, and calculating the fusion weights based on the centerline coordinates and a preset piecewise function; and performing fusion and stitching processing on the projection data from both scans based on the weighted weights to obtain the expanded projection data. Specifically, the rotation angle of each frame is calculated based on the total angle of rotation and the total number of frames in one scan. The global scan error frame number is obtained by dividing the global scan error angle by the rotation angle of each frame. Since the second rotation scan is a reverse scan, the projection data of the second rotation scan needs to be inverted sequentially. Then, the corresponding projection data is extracted based on the global scan error frame number to obtain the aligned projection data of the second rotation scan. Furthermore, the centerline y-coordinate for fusion and stitching is determined based on the global displacement distance. A weighted average is then calculated using a pre-defined piecewise function to determine the region to be fused. The pre-defined piecewise function is as follows: ; in This represents the coordinate of the i-th row in the y-coordinate direction. The y-coordinate of the centerline is represented by D, the distance of the preset linear fusion region is represented by w, and the weighting weight is represented by w. Based on the above weighting, the projection data of the first rotation scan and the second rotation scan after the sequence is reversed and aligned are merged and spliced ​​to obtain the expanded projection data; Step 6: Perform displacement adjustment processing on the geometric correction parameters of the CBCT scanning device using the global displacement distance to obtain the geometric correction parameters corresponding to the expanded projection dataset; This step aims to adjust the geometric correction parameters obtained from the first rotational scan using the global displacement distance, thereby adapting them to the expanded projection data to address the issue of geometric artifacts during subsequent reconstruction.

[0019] Specifically, the fusion and stitching distance is determined by the global displacement distance. Since the scanning device only performs detector lifting, it can be physically assumed that the geometric parameters of the scanning device are only adjusted in the v direction. Based on this theory, the translation displacement values ​​in the geometric parameters are iterated and readjusted. The calculation formula is as follows: ; in This represents the translational displacement value for readjustment. This represents the original translation displacement value, and d represents the distance of the merged splicing. Step 7: Based on the extended projection dataset and the geometric correction parameters of the extended projection dataset, obtain large field-of-view 3D image data through the preset FDK reconstruction algorithm; This step aims to obtain large-field-of-view 3D image data by executing a reconstruction procedure according to the preset FDK reconstruction algorithm based on the expanded projection data and adjusted geometric correction parameters. As a further extension, this step can also add image enhancement functions such as image post-processing (e.g., image denoising, image sharpening, etc.) and artifact removal.

[0020] The following points should be noted in this article: 1. The accompanying drawings of the embodiments disclosed herein only relate to the structures involved in the embodiments disclosed herein; other structures can be referred to in general design.

[0021] 2. Where there is no conflict, the embodiments of this disclosure and the features in the embodiments can be combined with each other to obtain new embodiments.

[0022] Although embodiments of the present invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

Claims

1. A CBCT large field-of-view reconstruction system, comprising a scanning module, a data acquisition and correction module, a reference frame extraction module, a global parameter calculation module, a projection domain fusion mode, a geometric correction parameter readjustment module, and a large field-of-view image reconstruction module, characterized in that: The scanning module is used to acquire imaging instructions, perform a first rotational scan on the target object according to a preset timing sequence to collect the lower region projection dataset, and perform a second rotational scan on the target object after raising the detector under the same exposure parameter conditions to collect the upper region projection dataset. The data acquisition and correction module is used to acquire air and preset upper and lower region projection datasets of the calibration phantom, and to solve the geometric correction parameters of the CBCT scanning device using a preset geometric correction algorithm based on the lower region projection dataset of the calibration phantom. The reference frame extraction module is used to extract a first reference projection frame at a preset scanning angle from the lower region projection dataset of the calibration phantom, and to extract a second reference projection frame at a reverse scanning angle from the upper region projection dataset of the calibration phantom. The global parameter calculation module is used to calculate the global displacement distance and global scanning error angle of the second reference projection frame relative to the first reference projection frame based on the first reference projection frame and the second reference projection frame using a projection matrix decomposition algorithm. The projection domain fusion module is used to construct a projection domain fusion model using global displacement distance and global scanning error angle, and to perform fusion and stitching processing on the upper and lower region projection datasets through the projection domain fusion model to obtain an expanded projection dataset. The geometric correction parameter readjustment module is used to perform displacement adjustment processing on the geometric correction parameters of the CBCT scanning device using global displacement distance, so as to obtain the geometric correction parameters corresponding to the expanded projection dataset. The large field-of-view image reconstruction module is used to obtain large field-of-view three-dimensional image data based on the extended projection dataset and the geometric correction parameters of the extended projection dataset, using a preset FDK reconstruction algorithm.

2. A CBCT large field-of-view reconstruction method, characterized in that: The method is implemented using the CBCT large field-of-view reconstruction system described in claim 1, and the technical solution of the method is as follows: Step 1: Perform a first rotational scan on the target object according to the preset timing sequence to collect the lower region projection dataset, and then perform a second rotational scan on the target object after raising the detector under the same exposure parameters to collect the upper region projection dataset. Step 2: According to the above imaging instructions, acquire air and preset upper and lower region projection datasets of the calibration phantom. Based on the lower region projection dataset of the calibration phantom, use the preset geometric correction algorithm to solve the geometric correction parameters of the CBCT scanning device. Step 3: Extract the first reference projection frame at the preset scanning angle from the lower region projection dataset of the calibration phantom, and extract the second reference projection frame at the reverse scanning angle from the upper region projection dataset of the calibration phantom. Step 4: Based on the first and second reference projection frames, calculate the global displacement distance and global scanning error angle of the second reference projection frame relative to the first reference projection frame using the projection matrix decomposition algorithm; Step 5: Construct a projection domain fusion model using global displacement distance and global scanning error angle, and perform fusion and stitching processing on the upper and lower region projection datasets through the projection domain fusion model to obtain the expanded projection dataset; Step 6: Perform displacement adjustment processing on the geometric correction parameters of the CBCT scanning device using the global displacement distance to obtain the geometric correction parameters corresponding to the expanded projection dataset; Step 7: Based on the extended projection dataset and the geometric correction parameters of the extended projection dataset, obtain large-field-of-view 3D image data through the preset FDK reconstruction algorithm.

3. The CBCT large field-of-view reconstruction method according to claim 2, characterized in that: Step 1, which involves performing a first rotational scan on the target object according to a preset timing sequence to acquire the lower region projection dataset, and then performing a second rotational scan on the target object to acquire the upper region projection dataset after raising the detector under the same exposure parameters, specifically includes: The first rotational scan starts from the pre-acceleration point and performs forward acceleration. When the preset exposure point is reached, the exposure is turned on, and the exposure and rotational scan stop when the scan angle is detected to have reached the preset constant threshold. After completing the first rotational scan, the detector is first raised to the preset position, and then the scanning position is controlled to retreat along the reverse trajectory to the preparatory acceleration point that has passed the starting position of the first rotational scan. The same exposure parameters are maintained, and the second reverse rotational scan is started when the scanning position reaches the starting position.

4. The CBCT large field-of-view reconstruction method according to claim 2, characterized in that: Step 4, which calculates the global displacement distance and global scanning error angle of the second reference projection frame relative to the first reference projection frame using a projection matrix decomposition algorithm based on the first and second reference projection frames, specifically includes: After performing physical corrections on the first and second reference projection frames respectively, the two-dimensional image coordinates of the steel ball in the first and second reference projection frames are extracted by combining the adaptive threshold segmentation algorithm and the fixed threshold segmentation algorithm. Based on the above two-dimensional image coordinates, the elliptical angle sorting method is used to sort the steel balls, and the average pixel displacement distance of the y coordinate of the steel balls after the first and second upper-level sorting is calculated, which is the global displacement distance. The first and second projection matrices are calculated based on the sorted two-dimensional image coordinates and the preset three-dimensional spatial coordinates of the steel ball, respectively. The first and second rotational scanning angles are then calculated using a projection matrix decomposition algorithm to obtain the global scanning error angle.

5. A CBCT large field-of-view reconstruction method according to claim 2, characterized in that: Step 6 constructs a projection domain fusion model using global displacement distance and global scanning error angle, and performs fusion and stitching processing on the upper and lower region projection datasets through the projection domain fusion model to obtain the expanded projection dataset, which specifically includes: Physical corrections are performed on the projection data of the first and second rotating scans, and the projection data of the second rotating scan is reversed. After calculating the number of error frames for forward and reverse rotation scanning based on the global scanning error angle, the forward and reverse rotation scanning projection data alignment process is performed. The centerline coordinates of the projection domain fusion are determined based on the global displacement distance, and the fusion weights are calculated using a piecewise function. Based on the aligned forward and reverse rotation scan projection data and fusion weights, the preset overlapping regions are fused and stitched to obtain the expanded projection dataset.

Images