On-chip integrated coded aperture cone beam computed tomography method and system

Abstract

The application discloses a method and system for on-chip integrated coded aperture cone-beam computed tomography, and relates to the field of cone-beam computed tomography, comprising: designing a coded mask plate according to the scale of an X-ray image sensor and an expected magnification, and integrating the coded mask plate on the X-ray image sensor; and constructing an on-chip integrated coded aperture X-ray image sensor imaging model. A two-dimensional high-resolution image is obtained by reconstructing a collected two-dimensional coded image through an SCI algorithm, and a three-dimensional high-resolution voxel image is obtained by reconstructing through an FDK algorithm. The application reconstructs a cone-beam computed tomography device, constructs an X-ray imaging model for mapping a two-dimensional image to a three-dimensional space, and can use natural images or X-ray images for training, reconstruct a two-dimensional high-resolution image from a two-dimensional coded image, and finally obtain a three-dimensional high-resolution voxel image through an FDK algorithm. The application has the characteristics of compact structure, high reconstruction quality, good generalization, high robustness and the like.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cone-beam computed tomography, in particular to a method and system for on-chip integrated coded aperture cone-beam computed tomography. BACKGROUND

[0002] The resolution capability of the X-ray image sensor carried in the cone-beam computed tomography system is one of the key factors determining the resolution capability of the reconstructed voxel. At present, limited by the imaging mechanism of the X-ray image sensor and the level of semiconductor technology, the resolution capability of the X-ray image sensor gradually reaches a bottleneck and is difficult to further improve. At the same time, the image super-resolution technology based on deep learning lacks theoretical basis and medical basis, and it is difficult to avoid misdiagnosis of artifacts, and it is difficult to apply in the field of medical imaging. SUMMARY

[0003] In view of the above problems, the present application is proposed.

[0004] Therefore, the technical problem solved by the present application is how to improve the imaging resolution capability of the existing X-ray image sensor without changing the hardware structure of the X-ray image sensor, so as to obtain a high-resolution three-dimensional tomographic image.

[0005] To solve the above technical problems, the present application provides the following technical scheme: a method for on-chip integrated coded aperture cone-beam computed tomography, comprising the following steps,

[0006] According to the cone-beam computed tomography device, the size of the coded pixels in the coded pixel array of the coded mask plate and the size of the single coded pixel are determined;

[0007] Randomly code the coded pixels of the coded pixel array as "0" and "1", and output the coded mask plate;

[0008] The cone-beam computed tomography device is modified, and based on the X-ray imaging principle, the X-ray image sensor and the coded mask plate in the cone-beam computed tomography device are designed and manufactured on-chip;

[0009] According to the X-ray transmission equation, an on-chip integrated coded aperture X-ray image sensor imaging model is constructed;

[0010] According to the cone-beam computed tomography principle, an on-chip integrated coded aperture high-resolution cone-beam computed tomography imaging model is constructed;

[0011] A group of two-dimensional coded images obtained by the computer tomography device in the working mode is reconstructed to obtain a group of two-dimensional high-resolution projection images by the Snapshot-Compressive-Imaging open source method;

[0012] A set of two-dimensional high-resolution projection images obtained by reconstruction is used to reconstruct a high-resolution three-dimensional voxel image by the Feldkamp-Davis-Kress open-source method.

[0013] As a preferred scheme of the on-chip integrated coded aperture cone beam computed tomography method, the size of the coded pixels in the coded aperture mask and the size of the individual coded pixels are determined according to the size of the area array of the X-ray image sensor, the size of the individual pixels, and the size and the resolution of the high-resolution reconstructed voxels.

[0014] If the size of the area array of the X-ray image sensor is M1×N1 pixels, the size of the individual sensor pixels is m×n, the coded aperture mask contains M2×N2 pixels, the size of the individual coded aperture mask pixels is a×b, and the magnification of the desired high-resolution image in the x-axis direction and the y-axis direction is z1 and z2, respectively, which are natural numbers greater than or equal to 2, the x-axis direction is perpendicular to the scanning direction of the X-ray image sensor, and the y-axis direction is parallel to the scanning direction of the X-ray image sensor.

[0015] The size of the area array of the X-ray image sensor and the coded mask should satisfy M2=z1M1 and N2=z2N1, and the size of the individual sensor pixels and the size of the individual coded aperture mask pixels should satisfy m=z1a and n=z2b, and each pixel in the X-ray image sensor corresponds to z1×z2 pixels in the coded mask.

[0016] As a preferred scheme of the on-chip integrated coded aperture cone beam computed tomography method, the pattern of the coded mask depends on the coded matrix composed of closed elements and non-closed elements, and only the closed elements and the non-closed elements exist in the M2×N2 pixels of the coded mask, which are denoted as “0” and “1”, respectively.

[0017] The coded mask is integrated on the X-ray image sensor, and the imaging model of the coded mask determines the construction of the coded matrix.

[0018] As a preferred scheme of the on-chip integrated coded aperture cone beam computed tomography method, the modification of the cone beam computed tomography device is made according to the structure of the X-ray image sensor carried by the computed tomography imaging device to be modified.

[0019] If the X-ray image sensor carried by the computed tomography imaging device is an X-ray image sensor containing a scintillator, there are two modification methods, including:

[0020] The encoding mask plate is close to the scintillator surface of the X-ray image sensor, and the encoding mask is close to the scintillator surface in a way of on-chip integration, and the distance between the encoding mask and the scintillator is 0, wherein 0 in the current encoding mask plate is an X-ray non-penetrable part, and 1 is an X-ray penetrable part.

[0021] The encoding mask plate is close to the photonic image sensor surface of the X-ray image sensor, and the encoding mask is close to the photonic image sensor surface in a way of on-chip integration, and the distance between the encoding mask and the photonic image sensor is 0, wherein 0 in the current encoding mask plate is a light non-penetrable part emitted by the scintillator after absorbing X-rays, and 1 is a light penetrable part emitted by the scintillator after absorbing X-rays.

[0022] If the X-ray image sensor carried by the computer tomography device is a scintillator-free X-ray photon counting detector, the encoding mask plate is close to the photonic image sensor surface of the scintillator-free X-ray image sensor, and the encoding mask plate is close to the photonic image sensor surface in a way of on-chip integration, and the distance between the encoding mask plate and the photonic image sensor is 0, wherein 0 in the current encoding mask plate is an X-ray non-penetrable part, and 1 is an X-ray penetrable part.

[0023] As a preferred scheme of the on-chip integrated encoding aperture cone beam computer tomography method, the on-chip integrated encoding aperture X-ray image sensor imaging model is constructed by mapping z1×z2 pixels in a two-dimensional space corresponding to each pixel in a two-dimensional encoding image to a three-dimensional space, denoted as a three-dimensional vector 1×1×z1z2, and a high-resolution image is expressed as

[0024] For each X-ray two-dimensional encoding image collected and obtained, one encoding pixel corresponding to the corresponding position is sequentially taken from each X-ray image sensor pixel corresponding to z1z2 encoding pixels to re-form a two-dimensional vector corresponding to the data of the encoding pixel, to obtain z1z2 two-dimensional vectors; and the z1z2 two-dimensional vectors are stacked to form a three-dimensional vector, to obtain a three-dimensional encoding image.

[0025] Each pixel of the X-ray image sensor corresponds to z1×z2 pixels in the encoding mask plate, and for the measurement value L(x,y) of the (x,y) pixel in the low-resolution image with a resolution of X×Y collected by the X-ray image sensor in the working mode, the expression is:

[0026]

[0027] wherein F(x i ,y j ) is the (x i ,y i) pixel position, taking value "0" or "1", S1(x i ,y j ) is the image pixel value scanned by the X-ray image sensor at the desired high resolution;

[0028] The expression of the coded mask imaging model is:

[0029] y = Fx

[0030] wherein, is the vectorized representation of the image obtained by scanning with the X-ray image sensor, is the vectorized representation of the desired high resolution image , is the coded perception matrix of the coded mask;

[0031] The expression of the coded perception matrix is:

[0032]

[0033] wherein the entire coded perception matrix covers all the pixels on the X-ray image sensor surface array, i.e. the entire coded mask is composed of the coded perception matrix .

[0034] As a preferred scheme of the on-chip integrated coded aperture cone-beam computed tomography method, the construction of the on-chip integrated coded aperture high resolution cone-beam computed tomography model comprises that, when the modified computed tomography device is in the working mode, the X-ray source emits X-rays to penetrate different parts of the object at different angles to produce different degrees of absorption attenuation, and forms a two-dimensional projection image on the X-ray image sensor. Meanwhile, the X-ray image sensor and the X-ray source rotate around the object to obtain a group of two-dimensional projection images. The current obtained image is an X-ray two-dimensional coded image modulated by the coded mask, and all the images obtained by one scanning are stored as a group of two-dimensional coded images to be reconstructed.

[0035] As a preferred scheme of the on-chip integrated coded aperture cone-beam computed tomography method, the construction of the on-chip integrated coded aperture high resolution cone-beam computed tomography model comprises that, when the modified computed tomography device is in the working mode, the X-ray source emits X-rays to penetrate different parts of the object at different angles to produce different degrees of absorption attenuation, and forms a two-dimensional projection image on the X-ray image sensor. Meanwhile, the X-ray image sensor and the X-ray source rotate around the object to obtain a group of two-dimensional projection images. The current obtained image is an X-ray two-dimensional coded image modulated by the coded mask, and all the images obtained by one scanning are stored as a group of two-dimensional coded images to be reconstructed.

[0036] When reconstructing the two-dimensional coded image, first map each pixel of the two-dimensional coded image to a three-dimensional space, denoted as 1x1xz1z2, and then take the three-dimensional coded image as the input of the reconstruction model;

[0037] Convert the reconstruction into a mathematical convergence problem, and the expression is:

[0038]

[0039] wherein, represents the reconstruction result, y is a three-dimensional matrix, represents the input of the reconstruction, ||·|| represents a norm, X1 and X2 are positive integers;

[0040] The result of the reconstruction is a three-dimensional high-resolution image with mapping the three-dimensional high-resolution image to a two-dimensional space to obtain an M2xN2 two-dimensional high-resolution image;

[0041] The reconstruction includes reconstructing according to an on-chip integrated coded aperture X-ray image sensor imaging model through a Snapshot-Compressive-Imaging open source method; the training data required in the reconstruction process is independent of the scene and modality of tomographic imaging applications, and different modality data sets such as natural images, X-ray images, infrared images, etc. are selected for model training in a cross-modality or multi-modality manner;

[0042] In the reconstruction process by the FDK open source algorithm, a set of two-dimensional high-resolution images are taken as inputs, and the projection data of each two-dimensional high-resolution projection image is pre-weighted, one-dimensional filtered and back projected in three steps to obtain a high-resolution three-dimensional voxel image;

[0043] The two-dimensional projection data is weighted and filtered, and the expression is:

[0044]

[0045] wherein g(a) is a one-dimensional filter, R is an orbit radius, β is a source position, also referred to as a projection angle, λ is a fan angle, k is a cone angle, and p(β, a, b) is projection data collected at a detector array coordinate (a, b);

[0046] The weighted projection data is back projected for reconstruction:

[0047]

[0048] U(x, y, β) = R + xcosβ + ysinβ

[0049] wherein f FDK (x, y, z) is back projected data, and U(x, y, β) is the coordinate position of a point in a two-dimensional projection plane in a three-dimensional space.

[0050] Another object of the present application is to provide an on-chip integrated coded aperture cone beam computed tomography system, which can improve the imaging accuracy and resolution by optimizing the direct integration design of the coded mask plate and the X-ray image sensor, and solve the problems of image distortion and alignment caused by the physical separation of the sensor and the mask plate in the prior art.

[0051] To solve the above technical problems, the present application provides the following technical solutions: an on-chip integrated coded aperture cone beam computed tomography system, comprising: a coded mask plate generation module, an integrated design module, a data acquisition system, and an image reconstruction module.

[0052] The coded mask plate generation module is a coded pixel array designed and optimized for the coded mask plate, and the size and layout of the coded pixels are determined according to the required imaging resolution and the characteristics of the X-ray image sensor, and random '0' and '1' coding is performed.

[0053] The integrated design module is to physically integrate the coded mask plate and the X-ray image sensor, so as to accurately align the mask plate and the sensor.

[0054] The data acquisition system is to control the scanning process of the imaging device and acquire the coded two-dimensional projection image.

[0055] The image reconstruction module is to use open source or customized algorithms to reconstruct the coded two-dimensional projection image into a high-resolution two-dimensional and three-dimensional image.

[0056] A computer device, comprising a memory and a processor, the memory stores a computer program, and the processor implements the steps of the on-chip integrated coded aperture cone beam computed tomography method as described above when executing the computer program.

[0057] A computer readable storage medium, which stores a computer program, the computer program is executed by a processor to implement the steps of the on-chip integrated coded aperture cone beam computed tomography method as described above.

[0058] The present application has the following beneficial effects: without changing the original X-ray image sensor resolution, by modifying the cone beam computed tomography, the coded mask plate is integrated on the X-ray image sensor, the two-dimensional coded image obtained by scanning is reconstructed to obtain a high-resolution two-dimensional image, and then the high-resolution two-dimensional image is reconstructed to obtain a high-resolution three-dimensional voxel image, thereby improving the image resolution. BRIEF DESCRIPTION OF DRAWINGS

[0059] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort on the basis of these drawings. Among them:

[0060] Figure 1 The overall flow chart of the on-chip integrated encoding aperture cone beam computed tomography method provided for the first embodiment of the present application;

[0061] Figure 2 The structure block diagram and the encoding mask plate diagram in the on-chip integrated encoding aperture cone beam computed tomography method provided for the first embodiment of the present application;

[0062] Figure 3 The process schematic diagram of the conversion of the two-dimensional encoding image into a three-dimensional image in the on-chip integrated encoding aperture cone beam computed tomography method provided for the first embodiment of the present application;

[0063] Figure 4 The structure diagram of the on-chip integrated encoding aperture cone beam computed tomography system provided for the second embodiment of the present application;

[0064] Figure 5 The partial result diagram of the on-chip integrated encoding aperture cone beam computed tomography method provided for the third embodiment of the present application, in which the X-ray image is used as the training set, the two-dimensional encoding image is reconstructed into a two-dimensional high-resolution image by the SCI algorithm, and then the three-dimensional voxel image is reconstructed by the FDK algorithm, wherein (a1) is the original X-ray image of the 118th layer, (a2) is the low-quality X-ray image of the 118th layer, (a3) is the reconstructed X-ray image of the 118th layer, wherein (b1) is the X-ray image of the 201st layer, (b2) is the low-quality X-ray image of the 201st layer, (b3) is the reconstructed X-ray image of the 201st layer, wherein (c1) is the original X-ray image of the 297th layer, (c2) is the low-quality X-ray image of the 297th layer, and (c3) is the reconstructed X-ray image of the 297th layer;

[0065] Figure 6The results of the natural image training model used for X-ray image reconstruction in the on-chip integrated coded aperture cone beam computed tomography method provided by the third embodiment of the present application, wherein (a1) is the original X-ray image of the 118th layer, (a2) is the low-quality X-ray image of the 118th layer, (a3) is the reconstructed X-ray image obtained by the 118th layer, wherein (b1) is the X-ray image of the 201st layer, (b2) is the low-quality X-ray image of the 201st layer, (b3) is the reconstructed X-ray image obtained by the 201st layer, wherein (c1) is the original X-ray image of the 297th layer, (c2) is the low-quality X-ray image of the 297th layer, and (c3) is the reconstructed X-ray image obtained by the 297th layer;

[0066] Figure 7 The flowchart of on-chip integration of the coded mask plate on the X-ray image sensor in the on-chip integrated coded aperture cone beam computed tomography method provided by the third embodiment of the present application, wherein a is a schematic diagram of the coded mask plate close to the scintillator surface of the X-ray image sensor, wherein b is a schematic diagram of the coded mask plate close to the photodiode image sensor surface of the X-ray image sensor, and wherein c is a schematic diagram of the coded mask plate close to the photodiode image sensor surface of the X-ray image sensor without scintillator. DETAILED DESCRIPTION

[0067] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should fall within the scope of protection of the present application.

[0068] Embodiment 1, refer to Figures 1-3 For an embodiment of the present application, an on-chip integrated coded aperture cone beam computed tomography method is provided, comprising:

[0069] S1, according to the area array scale of the X-ray image sensor carried by the cone beam computed tomography device, the size of a single pixel and the scale and size of the high-resolution reconstruction voxel expected to be obtained, the scale and size of the coded pixel array of the coded mask plate are determined.

[0070] Specifically, first, the size of the encoding pixels and the size of the single encoding pixel in the encoding pixel array of the encoding mask plate are determined according to the size of the area array of the X-ray image sensor, the size of the single pixel and the size and the size of the high-resolution reconstruction voxel expected to be obtained. If the size of the area array of the X-ray image sensor is M1×N1pixels, the size of the single sensor pixel is m×n, the encoding mask plate contains M2×N2pixels, and the size of the single encoding mask plate pixel is a×b. The magnification of the expected high-resolution image is z1and z2in the x-axis direction and the y-axis direction, respectively, which are natural numbers greater than or equal to 2. Wherein, the x-axis direction is perpendicular to the scanning direction of the X-ray image sensor, and the y-axis direction is parallel to the scanning direction of the X-ray image sensor. Then the size of the area array of the X-ray image sensor and the encoding mask plate should satisfy M2=z1M1, N2=z2N1; the size of the single sensor pixel and the size of the single encoding mask plate pixel should also satisfy m=z1a, n=z2b. Each pixel in the X-ray image sensor corresponds to z1×z2pixels in the encoding mask plate.

[0071] S2, the encoding pixels of the encoding pixel array are randomly coded as "0" and "1" to obtain the encoding mask plate.

[0072] The pattern of the encoding mask plate depends on the encoding matrix composed of closed elements and non-closed elements, and there are only closed elements and non-closed elements in the M2×N2pixels of the encoding mask plate, which are denoted as "0" and "1" respectively. The encoding mask plate is integrated on the X-ray image sensor. Wherein, the imaging model of the encoding mask plate determines the construction of the encoding matrix.

[0073] Referring to Figure 2 , a structural block diagram of a high-resolution cone beam computed tomography method integrated with an encoding aperture on a chip and a design diagram of an encoding mask plate are provided.

[0074] Specifically, when z1=z2=3, the schematic diagram of the encoding mask plate obtained after the X-ray image sensor pixel array is encoded is shown in the figure. The square surrounded by the black solid line in the figure represents an X-ray image sensor pixel, and the 9 black dotted line frame small squares surrounded in each X-ray image sensor pixel are 9 encoding pixels. The black square represents the encoding pixel with the encoding value of "0", and the white square represents the encoding pixel with the encoding value of "1". The size of the X-ray image sensor is 153.6mm×76.8mm, the size of the single pixel is 150um×150um, and the resolution of the X-ray image sensor is 1024×512. The size of the currently designed encoding mask plate should also be 153.6mm×76.8mm, the size of the single pixel is 50um×50um, and the currently obtained image resolution is 3072×1536.

[0075] S3, retrofitting the cone-beam computed tomography device. According to the design of the coded mask plate obtained in S2, combined with the principle of X-ray imaging, the X-ray image sensor and the coded mask plate in the cone-beam computed tomography device are designed and manufactured on-chip.

[0076] It should be further explained that if the X-ray image sensor carried by the computed tomography imaging device is an X-ray image sensor containing a scintillator, there are two retrofitting methods, including:

[0077] The coded mask plate is tightly attached to the surface of the scintillator of the X-ray image sensor, and the coded mask is tightly attached to the surface of the scintillator in the form of on-chip integration, and the distance between the coded mask and the scintillator is 0, where "0" in the current coded mask plate is an X-ray non-penetrable part, and "1" is an X-ray penetrable part.

[0078] The coded mask plate is tightly attached to the surface of the photoelectric image sensor of the X-ray image sensor, and the coded mask is tightly attached to the surface of the photoelectric image sensor in the form of on-chip integration, and the distance between the coded mask and the photoelectric image sensor is 0, where "0" in the current coded mask plate is a light emitted by the scintillator after absorbing X-rays, which cannot penetrate, and "1" is a light emitted by the scintillator after absorbing X-rays, which can penetrate.

[0079] If the X-ray image sensor carried by the computed tomography imaging device is a scintillator-free X-ray photon counting detector, the coded mask plate is tightly attached to the surface of the photoelectric image sensor of the scintillator-free X-ray image sensor, and the coded mask plate is tightly attached to the surface of the photoelectric image sensor in the form of on-chip integration, and the distance between the coded mask and the photoelectric image sensor is 0, where "0" in the current coded mask plate is an X-ray non-penetrable part, and "1" is an X-ray penetrable part.

[0080] S4, constructing an on-chip integrated coded aperture X-ray image sensor imaging model.

[0081] Specifically, each pixel in the two-dimensional coded image corresponds to a z1x z2 pixel in the two-dimensional space, which is mapped to a three-dimensional vector 1x1x z1z2 in the three-dimensional space; and the expected high-resolution image which can be expressed as

[0082] Specifically, for the coded mask plate designed in S3, the specific method of mapping 2x2 pixels in the two-dimensional space corresponding to each pixel in the two-dimensional coded image to the three-dimensional space is: Figure 3

[0083] ​For the X-ray two-dimensional code image collected, the data corresponding to the code pixels in the corresponding positions in the 2*2 code pixels corresponding to each X-ray image sensor pixel are sequentially taken to form a two-dimensional vector, and four two-dimensional vectors are obtained; then the four two-dimensional vectors are stacked according to the order of extraction to form a three-dimensional vector, and a three-dimensional code image is obtained.

[0084] Each pixel of the X-ray image sensor corresponds to one pixel in the z1* z2 pixels in the code mask plate, and the measurement value L(x, y) of the (x, y) pixel in the low-resolution image with a resolution of X*Y collected by the X-ray image sensor in the working mode can be represented as:

[0085]

[0086] wherein F(x i ,y j ) is the value of the (x i ,y i ) pixel position on the code mask plate, and the value is "0" or "1", and S1(x i ,y j ) is the image pixel value obtained by scanning the X-ray image sensor at the desired high resolution.

[0087] The code mask plate imaging model can be represented as:

[0088] y=Fx

[0089] wherein, is the vectorization representation of the image obtained by scanning the X-ray image sensor, is the vectorization representation of the desired high-resolution image , and is the code perception matrix of the code mask plate.

[0090] The code perception matrix is specifically represented as:

[0091]

[0092] wherein the entire code perception matrix covers all the pixels on the X-ray image sensor surface array, that is, the entire code mask plate is composed of the code perception matrix .

[0093] Further, the on-chip integrated coded-aperture X-ray image sensor imaging model designed according to the application is a standard Snapshot-Compressive-Imaging problem, has complete mathematical convergence proof, and can use Snapshot-Compressive-Imaging open source algorithm for model training.

[0094] In terms of model training, it is difficult to obtain a large amount of data for X-ray images compared to natural images, it is more difficult to obtain expected high-resolution X-ray images, and X-ray images and natural images have different imaging principles and image features, and generally cannot be simply transferred and referenced. The on-chip integrated coded-aperture cone-beam computed tomography method described in the patent can use natural images in addition to X-ray images for model training and applied to X-ray image reconstruction during model training, and the reconstruction can also obtain good results, solving the problem of difficulty in obtaining sufficient expected high-resolution X-ray images during model training.

[0095] S5, constructing an on-chip integrated coded-aperture high-resolution cone-beam computed tomography model.

[0096] Specifically, when the modified computed tomography device is in the working mode, the X-ray source emits X-rays to penetrate different parts of the object at different angles to produce different degrees of absorption attenuation, forming a two-dimensional projection image on the X-ray image sensor. At the same time, the X-ray image sensor and the X-ray source rotate around the object to obtain a set of two-dimensional projection images. The current obtained image is an X-ray two-dimensional coded image modulated by the coded mask. When storing the image, all images obtained by one scan are saved as a set of two-dimensional coded images to be reconstructed.

[0097] S6, reconstructing a set of two-dimensional high-resolution projection images from a set of two-dimensional coded images obtained by the computed tomography device in the working mode by using the Snapshot-Compressive-Imaging open source method.

[0098] When reconstructing the two-dimensional coded image, first map the two-dimensional coded image to a three-dimensional space, arrange the pixels of each X-ray image sensor in order to form a 1x1xM1N1 three-dimensional coded image, and then use the three-dimensional coded image as the input of the reconstruction model. The expression of the reconstruction algorithm is:

[0099]

[0100] wherein, represents the reconstruction result, y is a three-dimensional matrix representing the input of the reconstruction, ||·|| represents the norm, X1 and X2 are positive integers.

[0101] The result after model reconstruction is a three-dimensional high-resolution image, at this time, the three-dimensional image needs to be mapped to a two-dimensional space to obtain a M2X N2 two-dimensional high-resolution image.

[0102] S7, a set of two-dimensional high-resolution projection images obtained by reconstruction are reconstructed into high-resolution three-dimensional voxel images by the Feldkamp-Davis-Kress open source method.

[0103] Through the FDK open source algorithm reconstruction process, a set of two-dimensional high-resolution images are taken as input, and three steps of projection data pre-weighting, one-dimensional filtering and back projection are performed on each two-dimensional high-resolution projection image to obtain a high-resolution three-dimensional voxel image.

[0104] Weighted filtering of two-dimensional projection data:

[0105]

[0106] Where g(a) is a one-dimensional filter, R is the radius of the orbit, β is the source position, also known as the projection angle, λ is the fan angle, k is the cone angle, and p(β,a,b) is the projection data collected at the detector array coordinates (a,b);

[0107] The weighted projection data is then back projected and reconstructed:

[0108]

[0109] U(x,y,β)=R+xcosβ+ysinβ

[0110] Where f FDK (x,y,z) is the back projected data, and U(x,y,β) is the coordinate position of a point in the two-dimensional projection plane in three-dimensional space.

[0111] It is further explained that the reconstruction includes reconstructing according to the on-chip integrated coded aperture X-ray image sensor imaging model by the Snapshot-Compressive-Imaging open source method; the training data required in the reconstruction process is independent of the scene and modality of tomographic imaging applications, and natural images, X-ray images, infrared images, etc. can be selected as the data set for cross-modality or hybrid-modality model training, and a good and stable training model can be obtained.

[0112] ​Further, another object of the present application is to construct an on-chip integrated coded aperture X-ray image sensor imaging model according to the X-ray transmission equation, which can use natural images as training sets in addition to X-ray images during model training, and apply it to X-ray image reconstruction, also achieving good reconstruction results. Solving the problem that X-ray images are difficult to obtain a large amount of data compared to natural images, it is difficult to obtain a large number of expected high-resolution X-ray images, and the imaging principles and image features of X-ray images and natural images are different, which generally cannot be simply migrated and referenced.

[0113] The present application modulates the two-dimensional projection image scanned by the X-ray image sensor through the on-chip integrated coded mask, and expands the image into a three-dimensional coded image, then reconstructs it through the coded perception matrix, to obtain a high-resolution image. The two-dimensional high-resolution image after reconstruction is reconstructed through the mature cone-beam computed tomography inversion algorithm to generate a high-resolution three-dimensional image, solving the problem that the spatial resolution of the X-ray image sensor is difficult to further improve under the existing process level. The implementation case does not need to add an additional optical modulation system in front of the sensor, and changes the resolution improvement method from complex and expensive X-ray image sensor research and development to a more economical and simple way of integrating the coded mask on the X-ray image sensor in an on-chip integrated manner, which can improve the resolution of the cone-beam computed tomography technology under the existing process level.

[0114] Further, it is further explained that:

[0115] The imaging mechanisms of X-ray images and natural images are different.

[0116] X-ray images are recorded by detectors after X-rays penetrate objects. X-rays are absorbed when passing through objects, and different materials have different absorption degrees of X-rays, resulting in different X-ray intensities recorded on the detector, thereby forming an image. This attenuation mechanism enables X-ray images to reflect the internal structure and density differences of objects. Natural images are formed by cameras capturing visible light reflected by objects. The camera sensor records the intensity and color information of the reflected light to generate an image, relying on the reflection, refraction and scattering of the object surface. The reflection characteristics of different objects and surfaces determine the brightness, color and details of the image. This reflection mechanism makes natural images mainly display the surface characteristics of objects.

[0117] Secondly, the difference in image features. The light imaging of natural images has a relatively wide spectrum due to scattering, so we approximately consider that the noise distribution of natural images is uniform, generally Gaussian distribution; in X-ray imaging, the spectrum becomes relatively single, so the noise is approximately Poisson distribution.

[0118] X-ray images are single-channel gray-scale images with less color information, mainly reflecting the absorption difference of different density materials. And there is no concept of depth of field. Natural images are usually color images, containing rich color information. Color information is formed by the mixture of different wavelengths of light reflected by objects.

[0119] Finally, in terms of feedback information, all the information contained in an X-ray image in the medical field has potential value, while a natural image may only have a part of ROI useful. Human tissues have high similarity, and a slight change may represent diseased tissue.

[0120] Different modalities of images reflect different information, such as CT reflecting bleeding and clear bones, and MRI looking better at soft tissues. In addition, even the same modality, different imaging parameters will bring great differences, and the image quality difference is also huge. The above characteristics are not available in natural images.

[0121] Embodiment 2

[0122] Reference Figure 2 For an embodiment of the present application, a system for integrating an encoding aperture cone beam computed tomography method on a chip is provided, comprising: an encoding mask plate generation module, an integrated design module, a data acquisition system, and an image reconstruction module.

[0123] The encoding mask plate generation module is to design and optimize the encoding pixel array of the encoding mask template. The size and layout of the encoding pixel are determined according to the required imaging resolution and the characteristics of the X-ray image sensor, and random "0" and "1" encoding is performed.

[0124] The integrated design module is to physically integrate the encoding mask template with the X-ray image sensor, so as to accurately align the mask template and the sensor.

[0125] The data acquisition system is to control the scanning process of the imaging device and acquire the encoded two-dimensional projection image.

[0126] The image reconstruction module is to use open source or customized algorithms to reconstruct the encoded two-dimensional projection image into a high-resolution two-dimensional and three-dimensional image.

[0127] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the technical solutions that essentially contribute to the prior art or the parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0128] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a list of executable instructions for implementing logic functions, which can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus or device, such as a computer-based system, a system including a processor or other system that can fetch the instructions from the instruction execution system, apparatus or device and execute the instructions, or in conjunction with these instructions execution systems, apparatus or devices. For the purpose of this specification, the "computer-readable medium" can be any device that can contain, store, communicate, propagate or transport programs for use by or in connection with an instruction execution system, apparatus or device, or in conjunction with these instruction execution systems, apparatus or devices.

[0129] More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection having one or more wires (electrical devices), a portable computer diskette (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium can even be paper or other suitable medium on which the program can be printed, because the program can be electronically obtained, for example, by optical scanning of the paper or other medium, followed by editing, interpreting or otherwise processing, if necessary, in other suitable ways, to be electronically obtained and then stored in the computer memory.

[0130] It should be understood that various aspects of the application can be implemented in hardware, software, firmware or a combination of them. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any of the following technologies, known in the art, or their combinations can be used: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0131] Embodiment 3, refer to Figures 5-7 Another embodiment of the present application is to verify the beneficial effects of the present application, through economic benefit calculation and simulation experiment for scientific demonstration. In this embodiment, the existing traditional method and the method of this embodiment are experimented respectively.

[0132] Refer to Figure 5 , which is part of the results of a simulation experiment of a set of low-resolution X-ray images reconstructed by an on-chip integrated coded aperture high-resolution cone beam computed tomography method. In the figure, (a1), (a2) and (a3) represent the slice images at the 118th layer, where (a1) is the original X-ray image, (a2) is the low-quality X-ray image, and (a3) is the X-ray image reconstructed after the method described in the present application; (b1), (b2) and (b3) in the figure represent the slice images at the 201st layer, where (b1) is the X-ray image, (b2) is the low-quality X-ray image, and (b3) is the X-ray image reconstructed after the method described in the present application; (c1), (c2) and (c3) in the figure represent the slice images at the 297th layer, where (c1) is the original X-ray image, (c2) is the low-quality X-ray image, and (c3) is the X-ray image reconstructed after the method described in the present application.

[0133] It is particularly pointed out that there are characteristic differences between X-ray images and natural images, which cannot be simply migrated and applied in general cases, but based on the method described in the present application, in addition to training with X-ray images, natural images can also be used for training and applied to the reconstruction of X-ray images, refer to Figure 6Fig. 1 is a result image of training a model by using natural images and for X-ray image reconstruction, (a1), (a2) and (a3) in the figure represent slice images at the 118th layer, wherein (a1) is an original X-ray image, (a2) is a low-quality X-ray image, and (a3) is an X-ray image reconstructed after the method of the present application; (b1), (b2) and (b3) in the figure represent slice images at the 201st layer, wherein (b1) is an X-ray image, (b2) is a low-quality X-ray image, and (b3) is an X-ray image reconstructed after the method of the present application; (c1), (c2) and (c3) in the figure represent slice images at the 297th layer, wherein (c1) is an original X-ray image, (c2) is a low-quality X-ray image, and (c3) is an X-ray image reconstructed after the method of the present application.

[0134] It can be found by comparison that the image quality is improved to some extent after reconstruction, and the reconstruction effect of the natural image and the X-ray image training set is not much different. The on-chip integrated coded aperture high-resolution cone beam computed tomography method provided by the present application can effectively improve the image resolution and has good robustness.

[0135] Reference Figure 7 Three on-chip integrated methods involved in the on-chip integrated coded aperture high-resolution cone beam computed tomography method are provided, including: wherein a and b are for X-ray image sensors containing scintillators, and for X-ray photon counting detectors without scintillators.

[0136] Figure 7 (a) is to tightly attach the coding mask plate to the surface of the scintillator of the X-ray image sensor. The coding mask plate is composed of materials that can effectively block X-rays, including but not limited to high-density lead film and other materials. The coding mask plate is tightly attached to the surface of the scintillator in an on-chip integrated manner, and the distance between the coding mask plate and the scintillator is 0. In the current coding mask plate, "0" represents an X-ray non-penetrable part, and "1" represents an X-ray penetrable part.

[0137] Figure 7 (b) is to tightly attach the coding mask plate to the surface of the photoelectric image sensor of the X-ray image sensor. The coding mask plate is tightly attached to the surface of the photoelectric image sensor in an on-chip integrated manner, and the distance between the coding mask plate and the photoelectric image sensor is 0. In the current coding mask plate, "0" represents a light non-penetrable part emitted by the scintillator after absorbing X-rays, and "1" represents a light penetrable part emitted by the scintillator after absorbing X-rays.

[0138] Figure 7(c) is to close the coded mask plate to the photoelectric image sensor surface of the non-flicker X-ray image sensor, the coded mask plate is composed of materials that can effectively isolate X-rays, including but not limited to high-density lead film and other materials, the coded mask plate is closely attached to the photoelectric image sensor surface in a way of on-chip integration, the distance between the coded mask plate and the photoelectric image sensor is 0, "0" in the current coded mask plate is an X-ray non-penetrable part, and "1" is an X-ray penetrable part. In the embodiment, the coded mask plate is integrated on the X-ray image sensor in a way of on-chip integration, so as to realize coding of the two-dimensional projection image collected by the X-ray image sensor.

[0139] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, and all of them should be covered in the scope of the claims of the present application.

Claims

1. An integrated on-chip coded aperture cone beam computed tomography method, characterized in that, The application relates to a cone-beam computed tomography device, and comprises the following steps: According to the cone-beam computed tomography device, the size of the coded pixels in the coded pixel array of the coded mask plate and the size of a single coded pixel are determined; Randomly coding the coded pixels of the coded pixel array as "0" and "1" to output the coded mask plate; According to the X-ray imaging principle, the X-ray image sensor of the cone-beam computed tomography device is integrated on the chip with the coded mask plate to be designed and manufactured; According to the X-ray transmission equation, an imaging model of the integrated-on-chip coded aperture X-ray image sensor is constructed; According to the cone-beam computed tomography principle, an integrated-on-chip coded aperture high-resolution cone-beam computed tomography imaging model is constructed; A group of two-dimensional coded images obtained by the computed tomography device in the working mode are reconstructed into a group of two-dimensional high-resolution projection images through the Snapshot-Compressive-Imaging open source method; The reconstructed group of two-dimensional high-resolution projection images are reconstructed into a high-resolution three-dimensional voxel image through the Feldkamp-Davis-Kress open source method; The constructing the on-chip integrated coded-aperture X-ray image sensor imaging model is mapping the z1 x z2 pixels in the two-dimensional space corresponding to each pixel in the two-dimensional coded image into a three-dimensional space, denoted as a three-dimensional vector 1 x 1 x z1 z2, and the expected high-resolution image is represented as For each collected X-ray two-dimensional coded image, a two-dimensional vector is obtained by sequentially taking the data of a coded pixel corresponding to the corresponding position from z1z2 coded pixels corresponding to each X-ray image sensor pixel, and then stacking the z1z2 two-dimensional vectors to form a three-dimensional vector, thereby obtaining a three-dimensional coded image; Each pixel of the X-ray image sensor corresponds to z1x z2 pixels of the coded mask plate, and the measurement value L(x, y) of the (x, y) pixel in the low-resolution image with the resolution of XxY collected by the X-ray image sensor in the working mode is expressed as: where F(x i ,y j ) is the value of the (x i ,y i )-th pixel position on the encoding mask plate, taking the value of "0" or "1", and S1(x i ,y j ) is the image pixel value obtained by scanning the X-ray image sensor at the desired high resolution; The imaging model of the coded mask plate is expressed as: y=F x wherein, is an image obtained by scanning with an X-ray image sensor is a vectorized representation of is a desired high resolution image is a vectorized representation of is an encoding perception matrix encoding a mask plate; The coded perception matrix is expressed as: Wherein, the whole coding perception matrix The coverage is all the pixels on the X-ray image sensor surface array, that is, the whole coding mask plate is composed of coding perception matrix .

2. The integrated on-chip coded-aperture cone-beam computed tomography method of claim 1, wherein: The size of the coded pixels in the coded pixel array of the coded mask plate and the size of a single coded pixel are determined according to the surface array size of the X-ray image sensor, the size of a single pixel and the size and dimension of the high-resolution reconstructed voxel expected to be obtained. If the surface array size of the X-ray image sensor is M1xN1 pixels, the size of a single sensor pixel is m x n, the coded mask plate contains M2xN2 pixels, the size of a single coded mask plate pixel is a x b, the magnification of the expected high-resolution image in the x-axis direction and the y-axis direction is z1 and z2 respectively, the value is a natural number greater than or equal to 2, the x-axis direction is perpendicular to the scanning direction of the X-ray image sensor, and the y-axis direction is parallel to the scanning direction of the X-ray image sensor; The surface array size of the X-ray image sensor and the coded mask plate should satisfy M2=z1M1 and N2=z2N1, the size of a single sensor pixel and the size of a single coded mask plate pixel should satisfy m=z1a and n=z2b, and each pixel of the X-ray image sensor corresponds to z1x z2 pixels of the coded mask plate.

3. The integrated on-chip coded-aperture cone-beam computed tomography method of claim 2, wherein: The pattern of the coding mask plate depends on a coding matrix composed of closed elements and non-closed elements, and only the closed elements and the non-closed elements exist in the M2xN2 pixels of the coding mask plate, which are respectively denoted as "0" and "1"; The coding mask plate is integrated on the X-ray image sensor, and the imaging model of the coding mask plate determines the construction of the coding matrix.

4. The integrated on-chip coded-aperture cone-beam computed tomography method of claim 3, wherein: The modification of the cone-beam computed tomography device is made according to the structure of the X-ray image sensor carried by the computed tomography imaging device to be modified, and different modification methods are formulated; If the X-ray image sensor carried by the computed tomography imaging device is an X-ray image sensor containing a scintillator, there are two modification methods, including: The coding mask plate is closely attached to the surface of the scintillator of the X-ray image sensor, and the coding mask is closely attached to the surface of the scintillator in a way of on-chip integration, and the distance between the coding mask and the scintillator is 0, wherein "0" in the current coding mask plate is an X-ray non-penetrable part, and "1" is an X-ray penetrable part; The coding mask plate is closely attached to the surface of the photoelectric image sensor of the X-ray image sensor, and the coding mask is closely attached to the surface of the photoelectric image sensor in a way of on-chip integration, and the distance between the coding mask and the photoelectric image sensor is 0, wherein "0" in the current coding mask plate is a light non-penetrable part emitted by the scintillator after absorbing X-rays, and "1" is a light penetrable part emitted by the scintillator after absorbing X-rays; If the X-ray image sensor carried by the computed tomography imaging device is a scintillator-free X-ray photon counting detector, the coding mask plate is closely attached to the surface of the photoelectric image sensor of the scintillator-free X-ray image sensor, and the coding mask plate is closely attached to the surface of the photoelectric image sensor in a way of on-chip integration, and the distance between the coding mask plate and the photoelectric image sensor is 0, wherein "0" in the current coding mask plate is an X-ray non-penetrable part, and "1" is an X-ray penetrable part.

5. The integrated on-chip coded-aperture cone-beam computed tomography method of claim 4, wherein: The construction of the on-chip integrated coding aperture high-resolution cone-beam computed tomography imaging model includes that, when the modified computed tomography imaging device is in a working mode, the X-ray source emits X-rays to penetrate different parts of an object at different angles to produce different degrees of absorption attenuation, and forms a two-dimensional projection image on the X-ray image sensor, and the X-ray image sensor and the X-ray source rotate and scan around the object as the center to obtain a group of two-dimensional projection images, and the current obtained image is an X-ray two-dimensional coding image modulated by the coding mask plate, and all images obtained by one scanning are saved as a group of two-dimensional coding images to be reconstructed.

6. The on-chip integrated coded-aperture cone-beam computed tomography method of claim 5, wherein: The group of two-dimensional high-resolution projection images are obtained by using a Snapshot-Compressive-Imaging open source reconstruction algorithm model trained by using a natural image or an X-ray image to reconstruct a group of two-dimensional high-resolution projection images obtained by the computed tomography imaging device in the working mode. When the two-dimensional coding image is reconstructed, each pixel of the two-dimensional coding image is first mapped to a three-dimensional space, denoted as 1x1xz1z2, and then the three-dimensional coding image is taken as an input of a reconstruction model. The reconstruction is converted into a mathematical convergence problem, and the expression is: wherein, represents the reconstruction result, y is a three-dimensional matrix, represents the input of the reconstruction, || || represents a norm, X1 and X2 are positive integers; The reconstructed result is a three-dimensional high-resolution image of Mapping the three-dimensional high-resolution image to a two-dimensional space obtains an M2xN2 two-dimensional high-resolution image. The reconstruction includes reconstructing by a Snapshot-Compressive-Imaging open source method according to an on-chip integrated coded aperture X-ray image sensor imaging model; training data required in the reconstruction process is irrelevant to a scene and a mode of a tomographic imaging application, and different modal data sets of natural images, X-ray images and infrared images are selected for model training in a cross-modal or multi-modal manner, alone or in combination; In the reconstruction process by the FDK open source algorithm, a group of two-dimensional high-resolution images are taken as inputs, and projection data pre-weighting, one-dimensional filtering and back projection are performed on each two-dimensional high-resolution projection image, so as to obtain a high-resolution three-dimensional voxel image; The two-dimensional projection data is weighted and filtered, and the expression is as follows: Wherein, g(a) is a one-dimensional filter, R is an orbit radius, β is a source position, also referred to as a projection angle, λ is a fan angle, k is a cone angle, and p(β, a, b) is projection data collected at a detector array coordinate (a, b); The weighted projection data is back projected and reconstructed as follows: U(x, y, β) = R + xcosβ + ysinβ where f FDK (x,y,z) is the back-projection data, and U(x,y,β) is the coordinate position of a point in three-dimensional space on the two-dimensional projection plane.

7. A system for cone-beam computed tomography using the on-chip integrated encoding aperture method according to any one of claims 1 to 6, characterized in that: The coded mask plate generation module, the integrated design module, the data acquisition system and the image reconstruction module are included. The coded mask plate generation module is a coded pixel array designed and optimized for the coded mask plate, and the size and layout of the coded pixel are determined according to the required imaging resolution and the characteristics of the X-ray image sensor, and random "0" and "1" coding is performed. The integrated design module is to physically integrate the coded mask plate with the X-ray image sensor to achieve accurate alignment between the mask plate and the sensor. The data acquisition system is to control the scanning process of the imaging device and acquire the coded two-dimensional projection image. The image reconstruction module is to use open source or customized algorithms to reconstruct the coded two-dimensional projection image into a high-resolution two-dimensional and three-dimensional image.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to realize the steps of the on-chip integrated coded aperture cone beam computer tomography method in any one of claims 1 to 6.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the on-chip integrated coded aperture cone beam computer tomography method in any one of claims 1 to 6.

Images