Spiral CT image reconstruction method based on nerve radiation field

Abstract

The invention discloses a spiral CT image reconstruction method based on a nerve radiation field, and belongs to the field of computer vision. The method comprises the following steps: determining an original projection image of a scanned object; the original projection image is acquired by spiral CT equipment in a fan beam scanning mode; according to the sampling angle corresponding to the original projection image, classifying and splicing the original projection images with the same sampling angle in each cycle of the rotation process through the rotation direction to obtain a spliced projection image; based on the spliced projection image, calling an image reconstruction model to generate a spiral CT image; wherein the image reconstruction model is constructed and trained based on a neural radiation field. According to the method, the radiation risk of related personnel is reduced.

Description

Technical Field

[0001] The present application belongs to the field of computer vision, and specifically relates to a spiral CT image reconstruction method based on neural radiation field. Background Art

[0002] Computed tomography (CT) is a method that acquires two-dimensional radiographic images through multi-angle projections and uses mathematical reconstruction algorithms to determine the three-dimensional internal structure of the target. CT is widely used in clinical diagnosis, industrial nondestructive testing, and bioimaging. Its basic principle is to utilize the attenuation differences produced by X-rays as they penetrate an object to obtain projections. These projections are then reconstructed into voxel-level tomographic images using algorithms such as filtered back projection (FBP), algebraic reconstruction techniques (ART), or statistical iterative reconstruction (SIR).

[0003] Spiral CT technology is currently widely used. It operates by fixing the X-ray source and detector on a circular track within the spiral CT device. The X-ray source and detector rotate at a constant angular velocity around the device's Z-axis, while the scanning platform moves at a constant speed along the Z-axis. This allows the X-rays to travel along a spiral path through the object placed on the scanning platform, continuously capturing raw projection images through the detector. The object being scanned varies depending on the application. For example, in clinical diagnosis, the object being scanned is a person; in industrial nondestructive testing, the object being scanned is a workpiece.

[0004] However, related technologies typically use a high angular density approach to acquire raw projection images, posing a high radiation risk to personnel involved. Angular density refers to the number of raw projection images acquired per unit angle as the detector rotates around the scanned object. Relevant personnel include, but are not limited to, spiral CT operators and the scanned subject. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to provide a spiral CT image reconstruction method based on neural radiation field, which can solve the problem of how to reduce the radiation risk of relevant personnel.

[0006] In order to solve the above technical problems, this application is implemented as follows: In a first aspect, an embodiment of the present application provides a method for reconstructing a spiral CT image based on a neural radiation field, the method comprising: Determining an original projection image of the scanned object; the original projection image is acquired by a spiral CT device using a fan beam scanning method; According to the sampling angle corresponding to the original projection image, the original projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image; Based on the spliced ​​projection image, calling the image reconstruction model to generate a spiral CT image; The image reconstruction model is constructed and trained based on the neural radiation field.

[0007] In a second aspect, an embodiment of the present application provides a spiral CT image reconstruction device based on a neural radiation field, the spiral CT image reconstruction device based on a neural radiation field comprising: The determination module is used to determine the original projection image of the scanned object; the original projection image is acquired by a spiral CT device using a fan beam scanning method.

[0008] The splicing module is used to classify and splice the original projection images with the same sampling angle in each rotation process according to the sampling angle corresponding to the original projection image, so as to obtain a spliced ​​projection image.

[0009] A calling module is used to call an image reconstruction model based on the spliced ​​projection image to generate a spiral CT image; wherein the image reconstruction model is constructed and trained based on the neural radiation field.

[0010] In a third aspect, an embodiment of the present application provides a computer device comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the method described in the first aspect.

[0011] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented.

[0012] In a fifth aspect, an embodiment of the present application provides a chip, which includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the method described in the first aspect.

[0013] In an embodiment of this application, a spiral CT image reconstruction method based on neural radiation fields is proposed. This method uses neural radiation fields as the image reconstruction algorithm when reconstructing spiral CT images. The neural radiation field uses a neural network to model the density and color of each 3D point in space, enabling the reconstruction of high-quality 3D models from sparse inputs.

[0014] Therefore, in terms of image quality, this method takes advantage of the powerful ability of neural radiation fields to model sparse perspectives. The perspective refers to the relative angle between the detector and the scanned object when acquiring image data. Specifically, the neural radiation field can model the voxel density field through multi-angle ray projection data, achieve a high degree of restoration of detailed textures during the reconstruction process, and has good spatial consistency expression capabilities. Even in situations where the projection data is sparsely sampled, noisy, or there are slight artifacts, this method can still reconstruct high-fidelity tomographic images, effectively preserving important information such as tissue boundaries and structural contours.

[0015] Therefore, compared to acquiring raw projection images at high angular density, this method achieves high-quality reconstruction at low angular density, meaning a reduced number of viewing angles. This significantly reduces the number of X-ray exposures and total radiation dose, thereby lowering the radiation risk for those involved. In the field of clinical diagnosis, this method effectively mitigates the radiation risk to patients undergoing CT examinations while maintaining image quality, thus possessing significant clinical safety value. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flowchart of a spiral CT image reconstruction method based on neural radiation field provided by some embodiments of the present application; Figure 2 is a flowchart of model training provided by some embodiments of the present application; Figure 3a is a spliced ​​projection image provided by some embodiments of the present application without content deduplication processing; Figure 3b is another spliced ​​projection image after content deduplication processing provided by some embodiments of the present application; Figure 4a 1 is a comparison diagram of the rotated projected images when the sampling angles are 0° and 360° respectively without perspective correction provided by some embodiments of the present application; Figure 4b is another comparison diagram of the rotated projection image when the sampling angles after perspective correction are 0° and 360° respectively, provided by some embodiments of the present application; Figure 5a It is a cross-sectional view of a three-dimensional reconstruction result obtained by reconstructing the image reconstruction model without performing perspective correction on the rotated projection image, as provided in some embodiments of the present application; Figure 5b Another cross-sectional view of a three-dimensional reconstruction result obtained by reconstructing the image reconstruction model after performing perspective correction on the rotated projection image, provided in some embodiments of the present application; Figure 6is a structural block diagram of a spiral CT image reconstruction device based on neural radiation field provided by some embodiments of the present application; Figure 7 This is a diagram of the internal structure of a computer device provided in some embodiments of the present application. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0018] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects and are not used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments of this application can be implemented in an order other than those illustrated or described herein. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0019] To facilitate understanding, the following technical terms that may be involved in the embodiments of this application are explained: Fan beam: A type of X-ray source beam. X-rays diverge along a two-dimensional plane (with a certain thickness), and a linear array detector rotates around the scanned object to collect projection data.

[0020] Cone Beam: A type of X-ray source beam. X-rays are emitted from a point source and spread out in three dimensions. When collected by an area array detector, they can cover a larger volume in a single scan.

[0021] Both fan beam and cone beam can be used to scan objects in various scenarios, including clinical diagnosis, industrial non-destructive testing, and biological imaging. However, cone beams offer higher spatial resolution and faster scanning speeds, making them commonly used in oral CT and industrial CT, while fan beams are often used in clinical diagnosis.

[0022] Viewing angle height: During the scanning process, the detector rotates horizontally and also vertically displaces relative to the scanned object. Due to the limited size of the detector, after the scanned object is displaced, it also displaces within the detector's field of view. That is, the vertical position of the scanned object in the detector's field of view will change.

[0023] Viewing angle structure: The position of the scanned object in the detector’s field of view during rotation.

[0024] Rotation plane: The plane in which the X-ray source and detector rotate on a circular orbit.

[0025] Vertical direction: The direction perpendicular to the rotation plane, which is also the direction in which the scanning platform moves, is also called the Z-axis direction.

[0026] Spiral CT data: projections collected by the detector of a spiral CT device using a spiral acquisition path.

[0027] Spiral CT image: A tomographic image reconstructed by the above projections.

[0028] One rotation process: The X-ray source and detector rotate 360° on the circular orbit.

[0029] Sampling angle: The X-ray source fixed on the spiral CT device emits X-rays once every time it rotates a fixed angle, and the detector receives X-rays once to complete the sampling.

[0030] Angular density: In fan-beam CT, the number of X-ray photons received by the detector per unit angle.

[0031] Unit angle: The angular interval at which the detector samples. This can be 1°, 2°, 3°, etc. For example, if the unit angle is 1°, the X-ray source emits an X-ray every time it rotates 1°, and the detector receives an X-ray accordingly.

[0032] The initialization pose of the 360° horizontal rotation includes the poses of the detector at various sampling angles during one rotation.

[0033] Rotation direction: The direction in which the detector and X-ray source rotate, including counterclockwise and clockwise rotation.

[0034] The following describes in detail the spiral CT image reconstruction method based on the neural radiation field provided by the embodiment of the present application through specific embodiments and their application scenarios in combination with the accompanying drawings.

[0035] It should be noted that, as mentioned in the background, raw projection images are typically acquired using a high angular density method, which poses a high radiation risk to personnel involved. This high radiation risk is caused by acquiring raw projection images using a high angular density method, so to reduce this radiation risk, the angular density needs to be lowered.

[0036] In order to solve the above problems, the applicant has found that high angular density means that more original projection images are collected within a unit angle, which makes the reconstructed spiral CT image higher in quality and richer in details, but also increases the scanning time and radiation dose; low angular density means that fewer original projection images are collected within a unit angle, which makes the reconstructed spiral CT image lower in quality and loses details, but the scanning time and radiation dose will also be reduced accordingly.

[0037] Therefore, increasing angular density aims to improve quality, while reducing angular density leads to a decrease in quality, which is not worth the effort. The reason is that the image reconstruction algorithms currently used for neural radiation field-based spiral CT image reconstruction require high angular density to ensure reconstruction quality.

[0038] To this end, in one exemplary embodiment, this application proposes a spiral CT image reconstruction method based on neural radiance fields. This method uses the Neural Radiance Field (NeRF) as the image reconstruction algorithm when reconstructing spiral CT images. The Neural Radiance Field uses a neural network to model the density and color of each 3D point in space, enabling the reconstruction of high-quality 3D models from sparse inputs.

[0039] Therefore, in terms of image quality, this method takes advantage of the powerful ability of neural radiation fields to model sparse perspectives. The perspective refers to the relative angle between the detector and the scanned object when acquiring image data. Specifically, the neural radiation field can model the voxel density field through multi-angle ray projection data, achieve a high degree of restoration of detailed textures during the reconstruction process, and has good spatial consistency expression capabilities. Even in situations where the projection data is sparsely sampled, noisy, or there are slight artifacts, this method can still reconstruct high-fidelity tomographic images, effectively preserving important information such as tissue boundaries and structural contours.

[0040] Therefore, compared to acquiring raw projection images at high angular density, this method achieves high-quality reconstruction at low angular density, meaning a reduced number of viewing angles. This significantly reduces the number of X-ray exposures and total radiation dose, thereby lowering the radiation risk for those involved. In the field of clinical diagnosis, this method effectively mitigates the radiation risk to patients undergoing CT examinations while maintaining image quality, thus possessing significant clinical safety value.

[0041] However, the applicant's research has discovered that because the data collected by spiral CT equipment exhibits irregular spatial sampling—that is, its projections continuously move in the Z-axis direction and the angle of each projection image changes continuously—this leads to highly inconsistent viewing angles between projections, creating certain difficulties for the reconstruction algorithm. This highly inconsistent viewing angle refers to the inconsistent position of the scanned object in the detector's field of view in different projection images. Furthermore, because the relative motion trajectory between the detector and the scanned object is a spiral, the collected projection data approximates a one-dimensional sinogram, meaning that the width of the projection is much greater than its height.

[0042] Neural radiation fields typically use two-dimensional images with uniform viewing angles as input to ensure geometric consistency and ray consistency during neural radiation field model training. The quality of reconstruction depends heavily on the spatial uniformity of the input viewing angle and the stability of the light density distribution. A uniform viewing angle refers to capturing images at equal intervals on the same horizontal plane, such as a plane perpendicular to the Z axis, which means the viewing angle is highly consistent, and thus the viewing angle structure is consistent.

[0043] It should be noted that a uniform viewing angle can ensure the spatial uniformity of the viewing angle and the stability of the light density distribution, that is, geometric consistency and ray consistency.

[0044] Therefore, the current neural radiation field is difficult to be directly applied to the three-dimensional reconstruction task of spiral CT images.

[0045] In summary, due to the highly inconsistent viewing angles between projections and the fact that the projections are approximately one-dimensional data, spiral CT data has natural differences in data organization and viewing angle structure compared to the neural radiation field method. This makes it impossible for spiral CT data to meet the input data requirements of the neural radiation field, making it difficult for the two to work together. Ultimately, the current neural radiation field is difficult to directly apply to the three-dimensional reconstruction task of spiral CT images.

[0046] To address the above issues, this embodiment provides a solution for effectively converting the data structure of spiral CT data into a projection format suitable for neural radiation field modeling. Specifically, according to the sampling angle corresponding to the original projection image, the original projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image. Based on the spliced ​​projection image, the image reconstruction model is called to generate a spiral CT image. This solution thus achieves both compatibility with the data acquisition method of spiral CT data and full utilization of the neural radiation field reconstruction capability.

[0047] Specifically, this approach first implements a reconstructive transformation of spiral CT data (i.e., raw projection images) acquired using a fan-beam scanning method at the data structure level. By splicing the projection data from each cycle of the spiral path, the raw projection images are effectively converted into multi-angle projection images that approximate cone beams. Splicing is performed in the height dimension of the raw projection images, which improves the situation where the projection width is significantly greater than the height, thereby converting the spiral CT data from one dimension to two dimensions. This eliminates the data organization differences between spiral CT data and neural radiation field methods. Furthermore, the viewing angle height of each raw projection image acquired along the spiral path continuously changes along the Z-axis, thereby completing the scan of the scanned object. Each stitched projection image fully covers the scanned object at each sampling angle. This means that the viewing angle structure of the spiral CT data is converted to a structure that meets the input data requirements of the neural radiation field. This, in turn, constructs a spatial structure that meets the requirements of neural radiation field modeling. This conversion not only improves the compatibility between the original projection image and the neural radiation field, but also significantly broadens the scope of application of the neural radiation field in the CT field, and breaks through the technical barriers between the spiral scanning mechanism and the neural reconstruction network.

[0048] The following uses spiral CT applied in clinical diagnosis and fan beam scanning as an example to explain the various steps of this method: Reference Figure 1 The method includes steps 102 to 106. Step 102: determining an original projection image of the scanned object; the original projection image is acquired by a spiral CT device using a fan-beam scanning method.

[0049] After the patient is positioned on the scanning platform, the X-ray source and detector rotate at a constant speed around the Z-axis of the spiral CT system, while the scanning platform translates at a constant speed along the Z-axis, achieving continuous spiral scanning and acquiring a large number of fan-beam raw projection images. The acquired raw projection images are X-ray projection images in DICOM format. The detector size is H x W, meaning the raw projection image size is H x W. Here, H is the height of the raw projection image, and W is the width.

[0050] For example, the original projection images are stored in a DICOM file set arranged in the order of acquisition time, denoted as ,in, Indicates the k Assuming that the number of images collected per rotation in spiral scanning is N, the total number of images L should be M×N, where M represents a total of M rotations.

[0051] Step 104 , according to the sampling angle corresponding to the original projection image, the original projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image.

[0052] The classification splicing can be achieved by the following formula 1: (one) in, is the projected image after stitching; j is the number of the sampling angle; N is the number of the original projection images collected during one rotation, M is the number of rotations; For the M The number of the sampling angle in the circle is j The original projected image of For the The number of the sampling angle in the circle is j The original projected image of The number of the sampling angle in the first circle is j The original projected image of I The subscript of is the unique number of the angle of the original projection image during the rotation process.

[0053] For example, if the unit angle is 1°, 360 raw projection images are collected per week (N is 360), meaning that a raw projection image is collected every 1° of the detector rotation. If the number of rotations is 100, a total of 360 × 100 raw projection images are collected during the rotation. Based on this, before classifying and stitching, taking a sampling angle of 1° as an example, first take the first image (first week), then the 361st image (second week), and so on, until all 100 raw projection images corresponding to a sampling angle of 1° are collected during the sampling process and stitched together to form the stitched projection image corresponding to that sampling angle. This stitching method is then repeated for each sampling angle to obtain the stitched projection images corresponding to all sampling angles, resulting in 360 stitched projection images.

[0054] Step 106: Based on the spliced ​​projection image, call an image reconstruction model to generate a spiral CT image; wherein the image reconstruction model is constructed and trained based on the neural radiation field.

[0055] Among them, the image reconstruction model can process the input stitched projection image and the input horizontal 360° initialization pose, so as to output a spiral CT image, namely a CT reconstructed slice image, to achieve complete volume CT image reconstruction.

[0056] Among them, the initialization pose is mainly used to provide geometric constraints and perspective association information for the reconstruction of the three-dimensional scene.

[0057] During the training of the image reconstruction model: Reference Figure 2 The training process is as follows: based on the volume rendering pipeline, pose and scene modeling is first performed, and a three-dimensional density field is generated through the decoding network; then the three-dimensional density is line integrated to obtain the rendering projection; finally, the rendering projection is compared with the input projection to calculate the loss, and the network is updated until the model training requirements are met and the training is terminated.

[0058] The training process specifically includes the following steps 1 to 4: Step 1: scene modeling and light sampling.

[0059] That is, perform computer three-dimensional modeling; simulate the light source emitting light, and sample the light.

[0060] Specifically, for a given pixel point p at a specific pose and perspective, the observed pixel value is expressed as follows: (two) in, is the pixel value of point p; is the light intensity of the light source; is the equation of the light ray directed to the pixel point p, and are the starting and ending points of the ray in the integration area of ​​the reconstructed scene, respectively.

[0061] The discrete expression of formula 2 is as follows: (three) in, Indicates the number of sampling points on the light path; Indicates sampling point i arrive i +1 for the optical path length covered.

[0062] Step 2: Obtain the neural network features of the sampling points.

[0063] Convert the global coordinates of the sampling point into the local coordinates inside the corresponding node, and further extract the features of the local coordinate position And perform trilinear interpolation with the corresponding network weights through formula 4 to obtain the neural feature vector of the sampling point .

[0064] (Four) in, is the corresponding trilinear interpolation weight.

[0065] Step 3: Decoding network training.

[0066] A global decoding network shared by all sampling points And formula 5, the neural feature vector of the sampling point is converted into the required output (i.e. density ).

[0067] (five) Exemplarily, the global decoding network It can be a 3-layer Multilayer Perceptron (MLP) with 64 neurons, which has 4081 learnable parameters.

[0068] Among them, the SiLU activation function is used inside the MLP, and the SoftPlus activation function is used in the last layer. With weight Multiply them and perform discrete summation using Formula 3 in step 1 above to calculate the estimated pixel value.

[0069] Step 4: Calculate the loss function.

[0070] In order to optimize the parameters of the neural features and decoding network, this step uses the mean square error between the estimated pixel value and the true pixel value of the desired viewing angle to calculate the loss. The mean square error is expressed as follows: (six) in, is the true pixel value of pixel p, is the estimated pixel value of pixel p.

[0071] Further, refer to Figure 3a , where the dotted box is an enlarged image. After obtaining the spliced ​​projection image according to the above method, the applicant found that the continuity of the spliced ​​parts in the spliced ​​projection image was poor. This problem caused large artifacts in the subsequently reconstructed spiral CT image, thereby affecting the reconstructed resolution and clarity.

[0072] To address this issue, the applicant discovered that the problem arises from overlapping images captured by the detector during translation of the scanning bed along the Z axis. Therefore, this embodiment deduplicates the original projection image to obtain a deduplicated projection image. This deduplicated projection image is then classified and spliced. Specifically, the deduplicated projection images with the same sampling angle in each rotation are classified and spliced ​​according to the rotation direction, thereby obtaining a spliced ​​projection image. This solves the problem of poor continuity at the joints in the spliced ​​projection image.

[0073] In one embodiment, content deduplication can be achieved by performing pixel matching and overlap detection on two adjacent original projected images. Specifically, for two adjacent images in the original projected images, overlap detection is performed based on the pixel values ​​at the same coordinate position (e.g., whether the difference in pixel values ​​is within a preset threshold; if so, overlap is determined). The overlapping portions are the portions that require deduplication.

[0074] In another embodiment, the content deduplication process is cropping.

[0075] Specifically, for all original projection images, the overlapping pictures in two adjacent original projection images in the acquisition order are determined; for the overlapping pictures, each original projection image is cropped; and the remaining part of the original projection image after cropping the overlapping pictures is used as the deduplicated projection image.

[0076] For example, in a certain scanning process, the number of pixels in the overlapping images is fixed. Assuming that the overlapping images are the first w columns of the original projection images starting from the second image, it is necessary to calculate the pixel count of each original projection image. , and perform the following cropping using Formula 7: (seven) That is, take the original projected image The first w columns of , then the deduplication and projection image .

[0077] Accordingly, according to the sampling angle corresponding to the deduplicated projected image, the deduplicated projected images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain the spliced ​​projected image. The steps are: For each sampling angle, the angle number j ,in, ; A new spliced ​​projection image can be constructed , splicing can be achieved through formula eight: (eight) in, It is the projected image after stitching, that is, the "column image" after stitching, in which each column represents the same angle projection of multiple circles at a certain position; j is the number of the sampling angle; N is the number of the original projection images collected during one rotation, M is the number of rotations; For the M The number of the sampling angle in the circle is j The deduplicated projected image; For the The number of the sampling angle in the circle isj The deduplicated projected image; The number of the sampling angle in the first circle is j The deduplicated projected image; I The subscript of is the unique number of the angle of the deduplicated projected image during the rotation process.

[0078] Reference Figure 3b , Figure 3b The image below shows the projected image after deduplication (correcting the stitching width). The dotted box is an enlarged image, showing that the continuity of the projected image at the stitching joint is significantly improved after correcting the stitching width.

[0079] For example, because the rotation axis direction of spiral CT data differs from the rotation axis direction of the standard input format for neural radiation fields (horizontally and vertically), to improve the efficiency of the neural radiation field model in processing the spliced ​​projection image, in this embodiment, the spliced ​​projection image is rotated 90° to obtain a rotated projection image; based on the rotated projection image, the image reconstruction model is invoked to generate a spiral CT image.

[0080] The rotation operation can be achieved by formula nine: (Nine) After the above rotation operation, the dimension of the projected image becomes , so that the rotated projection image conforms to the standard input format of the neural radiation field.

[0081] Furthermore, the applicant found that, referring to Figure 4a Because the scanning bed is in constant motion during spiral CT data acquisition, the starting position of the rotated projection image at each sampling angle has a slight displacement in the Z-axis direction. This displacement can reduce the reconstruction quality of the model. Specifically, due to the vertical motion of the spiral motion trajectory, there is a slight vertical displacement between the first row of the rotated projection image at a sampling angle of 0° and the first row of the rotated projection image at a sampling angle of 1°. This displacement deviation increases with increasing sampling angles, resulting in a significant displacement between the first and last rotated projection images in a single rotation.

[0082] To address the above issue, this embodiment proposes a solution that performs perspective correction on each rotated projection image to eliminate the displacement deviation. Specifically, perspective correction is performed on the rotated projection image to obtain a corrected projection image. Based on this corrected projection image, an image reconstruction model is invoked to generate a spiral CT image.

[0083] Reference Figure 5a, which is a cross-sectional view of the three-dimensional reconstruction result obtained by the image reconstruction model without performing perspective correction on the rotated projection image. The artifact level in this cross-sectional view is high; refer to Figure 5b , which is a cross-sectional view of the 3D reconstruction result obtained by the image reconstruction model after perspective correction of the rotated projected image. The artifact level in this cross-sectional view is reduced to a certain extent. Therefore, this embodiment reduces the artifact level, thereby improving the resolution and clarity of the 3D reconstruction result, and further improving the resolution and clarity of the spiral CT image.

[0084] The above-mentioned perspective correction operations include but are not limited to the following operations: Operation method 1: First, the pixel offset of the rotated projection image in the Z-axis direction of the spiral CT device is determined, that is, Z-axis offset modeling is performed.

[0085] Specifically, to simulate the detector offset in the Z-axis direction during spiral scanning, when the sampling angle is 0° and the sampling angle is 360°, the maximum offset value of the projected image in the Z-axis direction after rotation is Pixels, then i The offset of the projected image after rotation can be calculated using formula 10: (ten) in, .

[0086] Secondly: In one embodiment, a correspondence is established between the pixel offset and each pixel point in each rotated projection image, and during the training of the image reconstruction model, the correspondence is input into the image reconstruction model for learning, so that the trained image reconstruction model can automatically perform perspective correction on the input rotated projection image during reconstruction.

[0087] In another embodiment, a mapping relationship between original pixel coordinates and target pixel coordinates is established, and based on the mapping relationship, the pixel values ​​at the original pixel coordinates are interpolated to the target pixel coordinates to obtain a corrected projected image.

[0088] The mapping relationship between the original pixel coordinates and the target pixel coordinates is established, including: For each column in the rotated projected image j , the original pixel coordinates are: (eleven) in, For each column in the rotated projected image j The transpose of the matrix formed by the original pixel coordinates of each pixel in .

[0089] The aligned target pixel coordinates can be expressed by Formula 12: (twelve) The range of the vertical coordinate (Z-axis direction) of the target pixel coordinate is clipped to the legal range using Formula 13: (Thirteen) Formula 13 can be used to limit the range of the vertical coordinate of the aligned target pixel coordinate, that is, to ensure that the vertical coordinate of the target pixel coordinate is within the legal interval [1, H], thereby avoiding calculation errors caused by exceeding the boundary.

[0090] The mapping relationship refers to By pixel offset and Align.

[0091] After the mapping relationship is obtained, the pixel points of the rotated projected image are interpolated and resampled column by column. That is, for each column of the pixel points of the rotated projected image j , through mapping relationships and interpolation methods (such as one-dimensional linear interpolation, proximity interpolation, etc.), each column in the rotated projected image The pixel values ​​of For example, if the interpolation method is one-dimensional linear interpolation, then each column in the rotated projection image can be calculated according to Formula 14. The pixel values ​​of superior: (fourteen) in, is the one-dimensional linear interpolation formula.

[0092] In summary, the projection image after Z-axis translation alignment (i.e., the corrected projection image) is .

[0093] Reference Figure 4b , Figure 4b is the rotated projected image after perspective correction. Figure 4a As can be seen from the dotted line, before the viewing angle correction, the displacement deviation of the projected image in the height direction is large between the sampling angle of 0° and the sampling angle of 360°. Figure 4b As can be seen at the middle dotted line, the displacement deviation is eliminated after perspective correction.

[0094] Operation method 2: Taking the pixel coordinates of each pixel point in the rotated projection image at a sampling angle of 0° as the reference coordinates, a pixel correspondence between the rotated projection image corresponding to each sampling angle and the rotated projection image at a sampling angle of 0° is established. During the training process of the image reconstruction model, the pixel correspondence is input into the image reconstruction model for learning, so that the trained image reconstruction model can automatically perform perspective correction on the input rotated projection image during reconstruction.

[0095] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0096] Based on the same inventive concept, the embodiments of the present application also provide a neural radiation field-based spiral CT image reconstruction device for implementing the neural radiation field-based spiral CT image reconstruction method involved above. The implementation solution provided by the device is similar to the implementation solution described in the above method. Therefore, the specific limitations of one or more neural radiation field-based spiral CT image reconstruction device embodiments provided below can be found in the above-mentioned limitations of the neural radiation field-based spiral CT image reconstruction method, and will not be repeated here.

[0097] In an exemplary embodiment, Figure 6 As shown, a spiral CT image reconstruction device based on neural radiation field is provided, comprising: a determination module 100, a splicing module 200 and a calling module 300, wherein: The determination module 100 is used to determine the original projection image of the scanned object; the original projection image is acquired by a spiral CT device using a fan-beam scanning method.

[0098] The splicing module 200 is configured to classify and splice the original projection images having the same sampling angle in each rotation process according to the sampling angle corresponding to the original projection image, so as to obtain a spliced ​​projection image.

[0099] The calling module 300 is used to call an image reconstruction model based on the spliced ​​projection image to generate a spiral CT image; wherein the image reconstruction model is constructed and trained based on the neural radiation field.

[0100] Optionally, the splicing module 200 is specifically used to: Performing content deduplication processing on the original projection image to obtain a deduplicated projection image; According to the sampling angle corresponding to the deduplicated projection image, the deduplicated projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image.

[0101] Optionally, the splicing module 200 is further configured to: For all original projection images, determining overlapping frames in two original projection images that are adjacent in acquisition order; For the overlapping images, cropping each original projection image; The remaining portion of the original projection image after the overlapping frames are cut out is used as a deduplicated projection image.

[0102] Optionally, the classification splicing in, is the projected image after stitching; j is the number of the sampling angle; N is the number of the original projection images collected during one rotation, M is the number of rotations; For the M The number of the sampling angle in the circle is j The deduplicated projected image; For the The number of the sampling angle in the circle is j The deduplicated projected image; The number of the sampling angle in the first circle is j The deduplicated projected image; I The subscript of is the unique number of the angle of the deduplicated projected image during the rotation process.

[0103] Optionally, the calling module 300 is specifically configured to: Rotating the spliced ​​projected image by 90° to obtain a rotated projected image; Based on the rotated projection image, an image reconstruction model is called to generate a spiral CT image.

[0104] Optionally, the calling module 300 is further configured to: Performing visual angle correction on the rotated projected image to obtain a corrected projected image; Based on the corrected projection image, an image reconstruction model is called to generate a spiral CT image.

[0105] Optionally, the calling module 300 is further configured to: Determining a pixel offset of the rotated projection image in a Z-axis direction of the spiral CT device; Based on the pixel offset, coordinate adjustment is performed on the original coordinates of the pixels in the rotated projection image to obtain a corrected projection image.

[0106] Optionally, the calling module 300 is further configured to: Based on the pixel offset, a mapping relationship between the original pixel coordinates and the target pixel coordinates is established; Based on the mapping relationship, the pixel values ​​at the original pixel coordinates are interpolated to the target pixel coordinates to obtain a corrected projected image.

[0107] Each module in the aforementioned neural radiation field-based spiral CT image reconstruction device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in the computer device in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0108] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 7As shown. The computer device includes a processor, memory, an input / output interface, a communication interface, a display unit, and an input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals via wired or wireless means, and the wireless means can be implemented via Wi-Fi, a mobile cellular network, near-field communication (NFC), or other technologies. When executed by the processor, the computer program implements a spiral CT image reconstruction method based on neural radiation fields. The display unit of the computer device is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0109] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0110] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: Determining an original projection image of the scanned object; the original projection image is acquired by a spiral CT device using a fan beam scanning method; According to the sampling angle corresponding to the original projection image, the original projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image; Based on the spliced ​​projection image, calling the image reconstruction model to generate a spiral CT image; The image reconstruction model is constructed and trained based on the neural radiation field.

[0111] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Performing content deduplication processing on the original projection image to obtain a deduplicated projection image; According to the sampling angle corresponding to the deduplicated projection image, the deduplicated projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image.

[0112] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: For all original projection images, determining overlapping frames in two original projection images that are adjacent in acquisition order; For the overlapping images, cropping each original projection image; The remaining portion of the original projection image after the overlapping frames are cut out is used as a deduplicated projection image.

[0113] In one embodiment, the classification splicing in, is the projected image after stitching; j is the number of the sampling angle; N is the number of the original projection images collected during one rotation, M is the number of rotations; For the M The number of the sampling angle in the circle is j The deduplicated projected image; For the The number of the sampling angle in the circle is j The deduplicated projected image; The number of the sampling angle in the first circle is j The deduplicated projected image; I The subscript of is the unique number of the angle of the deduplicated projected image during the rotation process.

[0114] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Rotating the spliced ​​projected image by 90° to obtain a rotated projected image; Based on the rotated projection image, an image reconstruction model is called to generate a spiral CT image.

[0115] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Performing visual angle correction on the rotated projected image to obtain a corrected projected image; Based on the corrected projection image, an image reconstruction model is called to generate a spiral CT image.

[0116] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Determining a pixel offset of the rotated projection image in a Z-axis direction of the spiral CT device; Based on the pixel offset, coordinate adjustment is performed on the original coordinates of the pixels in the rotated projection image to obtain a corrected projection image.

[0117] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Based on the pixel offset, a mapping relationship between the original pixel coordinates and the target pixel coordinates is established; Based on the mapping relationship, the pixel values ​​at the original pixel coordinates are interpolated to the target pixel coordinates to obtain a corrected projected image.

[0118] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps: Determining an original projection image of the scanned object; the original projection image is acquired by a spiral CT device using a fan beam scanning method; According to the sampling angle corresponding to the original projection image, the original projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image; Based on the spliced ​​projection image, calling the image reconstruction model to generate a spiral CT image; The image reconstruction model is constructed and trained based on the neural radiation field.

[0119] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Performing content deduplication processing on the original projection image to obtain a deduplicated projection image; According to the sampling angle corresponding to the deduplicated projection image, the deduplicated projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image.

[0120] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: For all original projection images, determining overlapping frames in two original projection images that are adjacent in acquisition order; For the overlapping images, cropping each original projection image; The remaining portion of the original projection image after the overlapping frames are cut out is used as a deduplicated projection image.

[0121] In one embodiment, the classification splicing in, is the projected image after stitching; j is the number of the sampling angle; N is the number of the original projection images collected during one rotation, M is the number of rotations; For the M The number of the sampling angle in the circle is j The deduplicated projected image; For the The number of the sampling angle in the circle is j The deduplicated projected image; The number of the sampling angle in the first circle is j The deduplicated projected image; I The subscript of is the unique number of the angle of the deduplicated projected image during the rotation process.

[0122] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Rotating the spliced ​​projected image by 90° to obtain a rotated projected image; Based on the rotated projection image, an image reconstruction model is called to generate a spiral CT image.

[0123] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Performing visual angle correction on the rotated projected image to obtain a corrected projected image; Based on the corrected projection image, an image reconstruction model is called to generate a spiral CT image.

[0124] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Determining a pixel offset of the rotated projection image in a Z-axis direction of the spiral CT device; Based on the pixel offset, coordinate adjustment is performed on the original coordinates of the pixels in the rotated projection image to obtain a corrected projection image.

[0125] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented: Based on the pixel offset, a mapping relationship between the original pixel coordinates and the target pixel coordinates is established; Based on the mapping relationship, the pixel values ​​at the original pixel coordinates are interpolated to the target pixel coordinates to obtain a corrected projected image.

[0126] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. In particular, any reference to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The databases involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the various embodiments provided herein may be, but are not limited to, general-purpose processors, central processing units (CPUs), graphics processing units (GPUs), digital signal processors (DSPs), programmable logic devices (PLDs), quantum computing-based data processing logic devices, artificial intelligence (AI) processors, and the like.

[0127] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0128] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A spiral CT image reconstruction method based on neural radiation field, characterized in that: The spiral CT image reconstruction method based on the nerve radiation field includes: Determining an original projection image of the scanned object; the original projection image is acquired by a spiral CT device using a fan beam scanning method; According to the sampling angle corresponding to the original projection image, the original projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image; Based on the spliced ​​projection image, calling the image reconstruction model to generate a spiral CT image; The image reconstruction model is constructed and trained based on the neural radiation field.

2. The method for spiral CT image reconstruction based on neural radiation field according to claim 1, characterized in that: The method classifies and splices the original projection images having the same sampling angle in each rotation process according to the sampling angle corresponding to the original projection image, and obtains the spliced ​​projection image by the rotation direction, including: Performing content deduplication processing on the original projection image to obtain a deduplicated projection image; According to the sampling angle corresponding to the deduplicated projection image, the deduplicated projection images with the same sampling angle in each rotation process are classified and spliced ​​according to the rotation direction to obtain a spliced ​​projection image.

3. The method for spiral CT image reconstruction based on neural radiation field according to claim 2, characterized in that: The performing content deduplication processing on the original projection image to obtain a deduplicated projection image includes: For all original projection images, determining overlapping frames in two original projection images that are adjacent in acquisition order; For the overlapping images, cropping each original projection image; The remaining portion of the original projection image after the overlapping frames are cut out is used as a deduplicated projection image.

4. The method for spiral CT image reconstruction based on neural radiation field according to claim 2, characterized in that: The classification splicing is achieved by the following formula: in, is the projected image after stitching; j is the number of the sampling angle; N is the number of the original projection images collected during one rotation, M is the number of rotations; For the M The number of the sampling angle in the circle is j The deduplicated projected image; For the The number of the sampling angle in the circle is j The deduplicated projected image; The number of the sampling angle in the first circle is j The deduplicated projected image; I The subscript of is the unique number of the angle of the deduplicated projected image during the rotation process.

5. The method for spiral CT image reconstruction based on neural radiation field according to any one of claims 1 to 4, characterized in that: The step of calling an image reconstruction model based on the spliced ​​projection image to generate a spiral CT image includes: Rotating the spliced ​​projected image by 90° to obtain a rotated projected image; Based on the rotated projection image, an image reconstruction model is called to generate a spiral CT image.

6. The method for spiral CT image reconstruction based on neural radiation field according to claim 5, characterized in that: The step of calling an image reconstruction model based on the rotated projection image to generate a spiral CT image includes: Performing visual angle correction on the rotated projected image to obtain a corrected projected image; Based on the corrected projection image, an image reconstruction model is called to generate a spiral CT image.

7. The method for spiral CT image reconstruction based on neural radiation field according to claim 6, characterized in that: The performing perspective correction on the rotated projected image to obtain a corrected projected image includes: Determining a pixel offset of the rotated projection image in a Z-axis direction of the spiral CT device; Based on the pixel offset, coordinate adjustment is performed on the original coordinates of the pixels in the rotated projection image to obtain a corrected projection image.

8. The method for spiral CT image reconstruction based on neural radiation field according to claim 7, characterized in that: The step of adjusting original coordinates of pixels in the rotated projected image based on the pixel offset to obtain a corrected projected image includes: Based on the pixel offset, a mapping relationship between the original pixel coordinates and the target pixel coordinates is established; Based on the mapping relationship, the pixel values ​​at the original pixel coordinates are interpolated to the target pixel coordinates to obtain a corrected projected image.

9. A computer device, characterized in that: The method comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the spiral CT image reconstruction method based on the neural radiation field as described in any one of claims 1 to 8.

10. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the spiral CT image reconstruction method based on the neural radiation field are implemented as described in any one of claims 1 to 8.

Images