WO2022000192A1 - 一种ct图像的构建方法、ct设备以及存储介质 - Google Patents

一种ct图像的构建方法、ct设备以及存储介质 Download PDF

Info

Publication number
WO2022000192A1
WO2022000192A1 PCT/CN2020/098942 CN2020098942W WO2022000192A1 WO 2022000192 A1 WO2022000192 A1 WO 2022000192A1 CN 2020098942 W CN2020098942 W CN 2020098942W WO 2022000192 A1 WO2022000192 A1 WO 2022000192A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
pixel
reconstructed
total variation
projection data
Prior art date
Application number
PCT/CN2020/098942
Other languages
English (en)
French (fr)
Inventor
胡战利
梁栋
付晶
杨永峰
郑海荣
刘新
Original Assignee
深圳先进技术研究院
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳先进技术研究院 filed Critical 深圳先进技术研究院
Priority to PCT/CN2020/098942 priority Critical patent/WO2022000192A1/zh
Publication of WO2022000192A1 publication Critical patent/WO2022000192A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—2D [Two Dimensional] image generation

Definitions

  • the present application relates to the technical field of computed tomography of medical images, and in particular, to a method, device and storage medium for constructing CT images.
  • X-ray CT scanning has been widely used in clinical medical imaging diagnosis, but excessive X-ray radiation dose during CT scanning may cause cancer risk.
  • how to minimize the dose of X-rays has become one of the key technologies in the field of medical CT imaging.
  • the easiest way to use is to reduce the tube current and scan time during the CT scan.
  • the projection data will contain a lot of noise, and the quality of the reconstructed image based on the traditional filtered back-projection method is seriously degraded, which is difficult to meet the needs of clinical diagnosis.
  • the iterative reconstruction method based on statistical model can achieve high-quality reconstruction of low-dose CT images by constructing an image reconstruction model based on the noise of the acquired projection data and the imaging system; the analytical reconstruction method based on projection data filtering The noise of the projection data and the imaging system perform data filtering modeling, and then use the analytical reconstruction method to achieve fast and high-quality low-dose CT image reconstruction.
  • the present application mainly provides a method for constructing a CT image, so as to improve the reconstruction quality of the CT image.
  • a technical solution adopted in the present application is to provide a method for constructing a CT image.
  • the method includes: acquiring projection data collected during scanning by CT equipment; using the projection data to construct an image reconstruction model, wherein the image reconstruction model is a total variation model based on a pixel neighborhood block, and the pixel neighborhood block is centered on one pixel input the preset image data into the image reconstruction model and perform iterative operations to obtain the reconstructed CT image output by the image reconstruction model.
  • the preset image data is input into the image reconstruction model and an iterative operation is performed to obtain a reconstructed CT image output by the image reconstruction model, including: reconstructing the projection data by using an analytical reconstruction algorithm to obtain an initial CT image, and converting the initial CT image As preset image data; the initial CT image is input into the image reconstruction model and iterative operation is performed to obtain the reconstructed CT image output by the image reconstruction model.
  • reconstructing the projection data by using an analytical reconstruction algorithm to obtain an initial CT image includes: reconstructing the projection data by using a filtered back-projection algorithm to obtain an initial CT image.
  • the total variation model based on pixel neighborhood block is:
  • f is the projection data
  • ⁇ is the CT image to be reconstructed
  • i, j are the number of pixels of the CT image to be reconstructed and the number of detection elements of the CT equipment detector, respectively
  • D(f, H ⁇ ) is the data fidelity term
  • TV represents the total variation regularization term
  • ⁇ represents the hyperparameter that balances the fidelity and regularization terms.
  • the total variation regularization term is:
  • s and t represent the index of the attenuation coefficient position in the CT image
  • ⁇ is a constant to maintain the differentiability with the image intensity
  • ⁇ s, t, l represent the position in the s-th row and the t-th column of the reconstructed CT image is the pixel value of the l-th pixel of the pixel neighborhood block of the center
  • ⁇ s-1,t,l represents the pixel located in the s-1th row and the t-th column of the reconstructed CT image, as the center pixel
  • the pixel value of the l-th pixel of the neighborhood block, ⁇ s,t-1,l represents the pixel located in the s-th row and the t-1-th column of the reconstructed CT image, as the center of the pixel.
  • the lth neighborhood block The pixel value of the pixel, N s,t represents the total number of pixels contained in the pixel neighborhood block.
  • the total variation model is a penalty weighted least squares total variation algorithm, wherein the penalty weighted least squares total variation algorithm introduces a total variation regular term based on pixel neighborhood blocks as a penalty term.
  • the total variation model based on pixel neighborhood block is:
  • ⁇ represents a diagonal matrix, and represents the variance of the projected data at detector channel i.
  • the variance of the projection data at detector channel i is obtained according to the following formula:
  • I 0 represents the X-ray incident photon intensity
  • I i represents the variance of the electronic noise of the system
  • I i represents the mean of the projected data at detector channel i.
  • the system matrix is a system matrix based on area weighting obtained from the CT equipment system, or a system matrix based on voxel weighting.
  • the iterative operation is any one of a gradient descent algorithm, a conjugate gradient descent algorithm, or an over-relaxation iterative algorithm.
  • the CT equipment includes: an internal bus, and a memory and a processor connected through the internal bus; the memory is used for storing computer programs; The steps of executing the computer program to realize the above-mentioned CT image construction method.
  • another technical solution adopted in the present application is to provide a computer-readable storage medium on which a computer program is stored, and when the program is executed by a processor, the above-mentioned CT image construction method is implemented.
  • the CT image construction method provided by the present application constructs a total variation model based on a pixel neighborhood block, wherein the pixel neighborhood block is a pixel centered on a pixel area. That is, the CT image reconstruction model introduces a total variation regular term based on pixel neighborhood blocks, and the total variation regular term is different from the traditional total variation regular term.
  • the total variation regular term provided in this embodiment uses the same
  • the pixel-related pixel neighborhood block calculates the gradient of the pixel in the two-dimensional image, which can eliminate the staircase effect, preserve the image details, and improve the resolution of the reconstructed CT image.
  • FIG. 1 is a schematic flowchart of an embodiment of a CT image construction method provided by the present application.
  • FIG. 2 is a schematic flowchart of an embodiment of step S30 in FIG. 1;
  • Figure 3(a) to Figure 3(e) describe the reconstruction effect comparison between the method of the present application and other methods
  • FIG. 4 is a schematic structural diagram of an embodiment of a CT device provided by the present application.
  • FIG. 5 is a schematic structural diagram of an embodiment of a storage medium provided by the present application.
  • first”, “second” and “third” in the embodiments of the present application are only used for description purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as “first”, “second”, “third” may expressly or implicitly include at least one of that feature.
  • "a plurality of” means at least two, such as two, three, etc., unless otherwise expressly and specifically defined.
  • the terms “comprising” and “having” and any variations thereof are intended to cover non-exclusive inclusion.
  • a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but optionally also includes unlisted steps or units, or optionally also includes For other steps or units inherent to these processes, methods, products or devices.
  • CT scans a layer of a certain thickness of a certain part of the human body with an X-ray beam.
  • the detector receives the X-ray that passes through the layer.
  • the measured signal is converted into digital information after analog-to-digital conversion, and then processed by the computer.
  • These data and information are stored in a magneto-optical disc or tape drive, and then converted into an analog signal after digital-to-analog conversion. After a certain transformation by the computer, it is output to the display device to display the image. Its density and resolution are high, and the X-ray plane can be directly displayed.
  • CT imaging is essentially the imaging of the attenuation coefficient of human tissue, and the physical principle of imaging is to obtain the equation system for solving the attenuation coefficient through the scanning of the CT scanning mechanism. Solve the equation system to obtain the attenuation coefficient value of each voxel at a certain body level of the human body, then convert the attenuation coefficient value into a CT value, and finally convert the CT value into a grayscale image that can be visually recognized.
  • CT image reconstruction is based on the determination of the absorption coefficient of X-rays in the human body using certain physical techniques, and the two-dimensional distribution matrix of the absorption coefficient value on a certain section of the human body is solved by using a certain mathematical method.
  • the two-dimensional distribution matrix of coefficients transforms the gray distribution on the image surface to achieve the purpose of reconstructing the volumetric image.
  • Each small unit is arranged and numbered according to the order in the scanning process to form an ordered array, and these ordered arrays form an image matrix on the image plane.
  • CT image reconstruction algorithms mainly include analytical methods and iterative methods.
  • the analytical method is the most commonly used convolution back-projection algorithm.
  • the advantage of the analytical method is that the reconstruction speed is fast, and the disadvantage is that the anti-noise performance is poor, but the completeness of the data is required.
  • the basic idea of the iterative method is to establish a set of algebraic equations of unknown vectors from the measured projection data, and to entangle the unknown image vectors through the set of equations.
  • Statistical iterative reconstruction algorithm is based on optimization theory. From the viewpoint of randomness of projection measurement process, image reconstruction is regarded as a parameter estimation problem, and a reasonable objective function is designed to find the parameter vector that makes the objective function reach the optimal value.
  • the prior information related to the image acquired in advance is introduced into the objective function of low-dose CT image reconstruction as a regularization term, which makes the solution process more stable, restores tissue structure information and suppresses noise better. Images can also be reconstructed for incomplete data.
  • Classical statistical iterative algorithms include maximum likelihood estimation algorithm, least squares algorithm, and Bayesian-based maximum a posteriori algorithm, etc.
  • the projection data contains noise, serious noise interference and stripe artifacts appear in the reconstructed image obtained by the above method, and some details are covered by noise, so a satisfactory reconstructed image cannot be obtained.
  • the inventor of the present application has found through long-term research that the main reason for the instability of the traditional non-quadratic penalty term is that the roughness of the image is calculated based on the intensity difference between adjacent pixels. When the image is noisy, differences in pixel intensities are unreliable in distinguishing between true edges and noise fluctuations.
  • the inventor of the present application proposes a total variational reconstruction method based on pixel neighborhood blocks, which uses pixel neighborhood blocks instead of individual pixels when measuring image roughness. Since it compares the similarity between different pixel neighborhood blocks, the pixel neighborhood block-based total variation reconstruction is more robust than the pixel-based total variation reconstruction method.
  • FIG. 1 is a schematic flowchart of an embodiment of a CT image construction method provided by the present application. The method includes the following steps:
  • S10 Acquire projection data acquired during scanning by the CT equipment.
  • the X-ray CT imaging equipment is used to acquire the line integral projection data of the CT medical image of the patient's low-dose radiation under the terrestrial milliampere-second scanning protocol, and the corresponding correction parameters and system matrix are obtained at the same time, wherein the radiation dose of the low-dose radiation is the standard 1/7 to 1/20 of the dose ray, which can be flexibly set by the user.
  • the line integral projection data refers to the projection data after logarithmic transformation
  • the obtained correction parameters refer to the X-ray incident photon intensity I 0 , the system electronic noise Variance etc., usually these values can be read or obtained directly from the test equipment.
  • the siddon algorithm is used to calculate the contribution of the jth tissue block in the object to be measured relative to the ith projection data f i , denoted as h ij , thereby obtaining the system matrix H.
  • the projection data in this embodiment may be the projection data obtained by the CT device scanning the patient's head, the projection data obtained by scanning the patient's whole body, or the projection data obtained by scanning the patient's vertebrae and other structural details, which are not specifically limited here.
  • preprocessing the scanned projection data and taking a negative logarithm to obtain the processed projection data Preprocessing the scanned projection data and taking a negative logarithm to obtain the processed projection data. Including noise processing of projection data, normalization processing of scan data, etc.
  • the image reconstruction model is a total variation model based on a pixel neighborhood block
  • the pixel neighborhood block is a pixel area centered on a pixel.
  • the pixel neighborhood block is a square pixel area with a pixel as the center and a set pixel unit as the pixel radius.
  • the total variation model based on pixel neighborhood blocks constructed by the present application is as follows:
  • f is the projection data
  • ⁇ is the CT image to be reconstructed
  • H ⁇ H i,j
  • ⁇ is the system matrix
  • i, j are the number of pixels of the CT image to be reconstructed and the number of detection elements of the CT detector, respectively, by
  • the specific CT imaging system decision can be calculated using different methods, for example, system matrix calculation based on area weighting, or system matrix calculation based on voxel weighting.
  • D(f, H ⁇ ) is the data fidelity term
  • TV represents the total variational regularization term
  • ⁇ represents the hyperparameter that balances the fidelity and regularization terms.
  • the total variational regularization term is:
  • ⁇ is a constant to maintain the differentiability with the image intensity
  • ⁇ s, t, l represent the position in the s-th row and the t-th row of the reconstructed CT image
  • the pixel of the column is the pixel value of the l-th pixel of the pixel neighborhood block of the center
  • ⁇ s-1,t,l represents the pixel located in the s-1th row and the t-th column of the reconstructed CT image, as the center.
  • the pixel value of the lth pixel of the pixel neighborhood block, ⁇ s,t-1,l represents the pixel located in the sth row and the t-1th column of the reconstructed CT image as the center of the pixel neighborhood block
  • the pixel value of each pixel, N s,t represents the total number of pixels contained in the pixel neighborhood block.
  • the total variation operator can make the reconstructed image have high precision and anisotropic smoothing effect. Through the calculation of the gradient, it can not only protect the edge of the image, but also better maintain the original contrast sharpness of the image boundary. . And the total variation of the noise-contaminated image is obviously larger than the total variation of the non-noise-contaminated image, and the noise of the image can be limited by limiting the size of the total variation. However, its processing effect in the smooth area is relatively poor, and the step effect is prone to occur, and some details such as problem information are easily filtered out in the process of noise removal by total variation, which affects the reconstruction effect.
  • the total variation regularization term based on the pixel neighborhood block provided by this application replaces the traditional total variation regularization term:
  • the pixel neighborhood block associated with the pixel is used to calculate the gradient of the pixel in the two-dimensional image. This replaces the instability caused by calculating the gradient of the pixel of the two-dimensional image based on the pixel itself based on the traditional total variation regular term.
  • the fidelity term of the pixel neighborhood block-based total variation model provided by the present application is a weighted least squares algorithm, that is, adding the pixel based total variation algorithm provided by the present application to the penalty weighted least squares total variation algorithm
  • the total variation regularization term of the neighborhood block is used as a penalty term to construct a penalized weighted square total variation reconstruction model.
  • ⁇ represents a diagonal matrix
  • ⁇ represents the variance of the projected data at detector channel i.
  • the variance of the projected data at detector channel i You can use the formula:
  • I 0 represents the X-ray incident photon intensity
  • I i represents the variance of the electronic noise of the system
  • the variance of the projected data at detector channel i It can also be obtained by other methods such as local neighborhood variance estimation, which is not specifically limited here.
  • S30 Input the preset image data into the image reconstruction model and perform an iterative operation to obtain a reconstructed CT image output by the image reconstruction model.
  • the above-mentioned iterative operation is any one of a gradient descent algorithm, a conjugate gradient descent algorithm or an over-relaxation iterative algorithm.
  • the iterative process it is judged whether the n-th iterative reconstructed image ⁇ n satisfies the iteration termination condition, and if so, the image data obtained at the n-th time is used as the final reconstructed image ⁇ * . If not, the CT image is updated based on the n-th iterative reconstructed image ⁇ n to obtain the updated iterative reconstructed image ⁇ n+1 , and the iteration is repeated to obtain the final reconstructed image ⁇ * .
  • the iteration termination condition is that the relative mean square error of the reconstruction results of two adjacent iterations is less than the set threshold k, that is, k is a positive real number.
  • the threshold value k set in the iteration termination condition is 0.001, and the value of the threshold value is set according to actual requirements, and no specific limitation is made here.
  • the CT image construction method constructs a total variation model based on a pixel neighborhood block, wherein the pixel neighborhood block is a pixel area centered on a pixel. That is, the CT reconstruction model introduces a total variation regular term based on pixel neighborhood blocks, and the total variation regular term is different from the traditional total variation regular term.
  • the associated pixel neighborhood block calculates the gradient of the pixel of the two-dimensional image, which can eliminate the staircase effect, preserve the image details, and improve the reconstructed image resolution.
  • step S30 may be performed through the following two steps:
  • S31 Reconstructing the projection data using an analytical reconstruction algorithm to obtain an initial CT image, and using the initial CT image as preset image data.
  • the projection data is reconstructed using a filtered back-projection method to obtain an initial CT image.
  • S32 Input the initial CT image into the image reconstruction model and perform an iterative operation to obtain a reconstructed CT image output by the image reconstruction model.
  • the analytical reconstruction algorithm of filtered back projection is a typical analytical reconstruction algorithm, and its advantages are that the reconstruction speed is fast, and the reconstruction effect is good when the projection data is complete.
  • the present application firstly uses the filtered back-projection reconstruction algorithm to pre-reconstruct the projection data obtained by scanning to speed up the reconstruction speed, and then uses the The iterative reconstruction algorithm updates the initial CT image to increase the details of the reconstructed CT image and improve the resolution of the reconstructed CT image.
  • FIG. 3(a) to FIG. 3(e) describe the reconstruction effect comparison between the method of the present application and other methods.
  • Fig. 3(a) is a reference image
  • Fig. 3(a) is a CT medical image of the CT equipment in standard dose radiation.
  • Figure 3(b) is a reconstructed image obtained by using the filtered back-projection method for low-dose data.
  • Figure 3(c) is the iteratively reconstructed image obtained by the penalized weighted least squares method for the low-dose image.
  • Figure 3(d) is an iteratively reconstructed image obtained by using the penalized weighted least squares total variation method for the low-dose image, and the total variation penalty term used in this method is the traditional total variation operator.
  • FIG. 4 is a schematic structural diagram of an embodiment of a CT device provided by the present application.
  • the CT device 400 includes an internal bus 401 , a memory 402 and a processor 403 connected through the internal bus 401 .
  • the memory 402 is used for storing computer programs.
  • the processor 403 is configured to execute a computer program to implement the steps of the CT image construction method provided by the present application.
  • the processor 403 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.
  • Memory 402 is used for executable instructions.
  • Memory 402 may include high-speed RAM memory, or may include non-volatile memory, such as at least one disk memory.
  • Memory 402 may also be a memory array.
  • the storage 402 may also be partitioned, and the blocks may be combined into virtual volumes according to certain rules.
  • the instructions stored in the memory 402 are executable by the processor 403 to enable the processor 403 to perform the method of CT image construction in any of the above-described method embodiments.
  • FIG. 5 is a schematic structural diagram of an embodiment of a computer-readable storage medium provided by the present application.
  • the computer-readable storage medium 500 stores a computer program 501, and when the computer program 501 is executed by the processor, implements the steps of the CT image construction method provided by the present application.
  • Computer storage medium 500 can be any available medium or data storage device that can be accessed by a computer, including but not limited to magnetic storage (eg, floppy disk, hard disk, magnetic tape, magneto-optical disk (MO), etc.), optical storage (eg, CD, DVD, BD, etc.) , HVD, etc.), as well as semiconductor memory (eg, ROM, EPROM, EEPROM, non-volatile memory 110 (NANDFLASH), solid state disk (SSD)), and the like.
  • magnetic storage eg, floppy disk, hard disk, magnetic tape, magneto-optical disk (MO), etc.
  • optical storage eg, CD, DVD, BD, etc.
  • HVD etc.
  • semiconductor memory eg, ROM, EPROM, EEPROM, non-volatile memory 110 (NANDFLASH), solid state disk (SSD)
  • the CT image construction method provided by this embodiment constructs a total variation model based on pixel neighborhood blocks. That is, the CT reconstruction model introduces a total variation regular term based on pixel neighborhood blocks, and the total variation regular term is different from the traditional total variation regular term.
  • the associated pixel neighborhood block calculates the gradient of the pixel of the two-dimensional image, which can eliminate the staircase effect, preserve the image details, and improve the resolution of the reconstructed CT image.
  • the present application first uses the filtered back-projection reconstruction algorithm to pre-reconstruct the projection data obtained by scanning to speed up the reconstruction speed, and then uses the The iterative reconstruction algorithm updates the initial CT image to increase the details of the final CT image, making the resulting CT image clearer.
  • the CT image construction method provided by the present application can not only speed up the reconstruction speed, but also improve the reconstruction quality.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Apparatus For Radiation Diagnosis (AREA)

Abstract

本申请提供一种CT图像构建方法、CT设备及计算机可读存储介质。该方法包括：获取CT设备进行扫描时采集的投影数据；利用投影数据构建图像重建模型，其中，图像重建模型是基于像素邻域块的全变分模型，像素邻域块为以一个像素为中心的像素区域；将预设图像数据输入至图像重建模型并进行迭代运算，以得到图像重建模型输出的重建CT图像。本申请提供的CT图像构建方法通过构建基于像素邻域块的全变分模型，并利用迭代算法求解该重建模型，使得到的重建CT图像分辨率更高，更加清晰。

Description

一种CT图像的构建方法、CT设备以及存储介质 【技术领域】
本申请涉及医学影像的计算机断层成像技术领域，特别是涉及一种CT图像的构建方法、设备以及存储介质。
【背景技术】
X射线CT扫描已经广泛应用于临床医学影像诊断，但是CT扫描过程中过高的X射线辐射剂量会存在致癌风险。为了降低对使用者的损害,如何最大限度地降低X射线使用剂量已经成为医学CT成像领域研究的关键技术之一。
为了降低X射线辐射剂量，一般地，使用的最简便的途径就是降低CT扫描过程中的管电流和扫描时间。然而，由于降低了管电流和扫描时间，就会使得投影数据中含有大量的噪声，基于传统的滤波反投影方法重建的图像质量存在严重的退化现象，难以满足临床诊断需要。
为了在保证图像质量的前提下大幅降低X射线辐射剂量，诸多基于降低管电流和扫描时间的低剂量CT图像重建方法相继提出，例如基于统计模型的迭代重建方法和基于投影数据滤波的解析重建方法。其中，基于统计模型的迭代重建方法，通过对采集的投影数据的噪声以及成像系统进行图像重建模型构建，可以实现低剂量CT图像优质重建；基于投影数据滤波的解析重建方法，是通过对采集的投影数据的噪声以及成像系统进行数据滤波建模，再通过解析重建方法实现快速且优质的低剂量CT图像重建。
【发明内容】
本申请主要提供一种CT图像的构建方法，以提高CT图像重建质量。
为解决上述技术问题，本申请采用的一个技术方案是：提供一种CT图像的构建方法。该方法包括：获取CT设备进行扫描时采集的投影数据；利用投影数据构建图像重建模型，其中，图像重建模型是基于像素 邻域块的全变分模型，像素邻域块为以一个像素为中心的像素区域；将预设图像数据输入至图像重建模型并进行迭代运算，以得到图像重建模型输出的重建CT图像。
其中，将预设图像数据输入至图像重建模型并进行迭代运算，以得到图像重建模型输出的重建CT图像，包括：利用解析重建算法对投影数据进行重建，获得初始CT图像，并将初始CT图像作为预设图像数据；将初始CT图像输入至图像重建模型并进行迭代运算，以得到图像重建模型输出的重建CT图像。
其中，利用解析重建算法对投影数据进行重建，获得初始CT图像，包括：利用滤波反投影算法对投影数据进行重建，获得初始CT图像。
其中，基于像素邻域块的全变分模型为：
Figure PCTCN2020098942-appb-000001
其中，f为投影数据，μ为待重建的CT图像，H＝{H i,j}为系统矩阵，i,j分别为待重建CT图像的像素个数和CT设备探测器的探测元个数，D(f,Hμ)为数据保真项，|μ| TV表示全变分正则化项，β表示平衡保真度和正则化项的超参数。
其中，全变分正则化项为：
Figure PCTCN2020098942-appb-000002
其中s和t表示CT图像中衰减系数位置的指标，δ为一常数，用以保持与图像强度的可微性，μ s,t,l表示以位于重建CT图像的第s行以及第t列的像素，为中心的像素邻域块的第l个像素的像素值，μ s-1,t,l表示以位于重建CT图像的第s-1行以及第t列的像素，为中心的像素邻域块的第l个像素的像素值，μ s,t-1,l表示以位于重建CT图像的第s行以及第t-1列的像素，为中心的像素邻域块的第l个像素的像素值，N s,t表示像素邻域块包含的像素总数。
其中，全变分模型为惩罚加权最小二乘全变分算法，其中，惩罚加权最小二乘全变分算法引入基于像素邻域块的全变分正则项为惩罚项。
其中，基于像素邻域块的全变分模型为：
Figure PCTCN2020098942-appb-000003
其中，Σ表示对角矩阵，且
Figure PCTCN2020098942-appb-000004
表示探测器信道i处投影数据的方差。
其中，探测器信道i处投影数据的方差根据以下公式获取：
Figure PCTCN2020098942-appb-000005
其中，I 0表示X射线入射光子强度，
Figure PCTCN2020098942-appb-000006
表示系统电子噪声的方差，
Figure PCTCN2020098942-appb-000007
表示探测器信道i处投影数据的均值。
其中，系统矩阵是从CT设备系统得到的基于面积加权的系统矩阵，或者基于体素加权的系统矩阵。
其中，迭代运算为梯度下降算法、共轭梯度下降算法或超松弛迭代算法中的任意一种。
为解决上述技术问题，本申请采用的又一个技术方案是，提供一种CT设备，该CT设备包括：内部总线，以及通过内部总线连接的存储器和处理器；存储器用于存储计算机程序；处理器用于执行计算机程序以实现上述CT图像构建方法的步骤。
为解决上述技术问题，本申请采用的另一个技术方案是，提供一种计算机可读存储介质，其上存储有计算机程序，该程序被处理器执行时实现上述CT图像构建方法。
本申请的有益效果是：区别于现有技术的情况，本申请提供的CT图像构建方法通过构建基于像素邻域块的全变分模型，其中，像素邻域块为以一个像素为中心的像素区域。也即，该CT图像重建模型引入基于像素邻域块的全变分正则项，且该全变分正则项不同于传统的全变分正则项，本实施例提供的全变分正则项使用与像素相关联的像素邻域块计算二维图像的该像素的梯度，能够消除阶梯效应，保留图像细节，提高重建CT图像的分辨率。
【附图说明】
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图，其中：
图1是本申请提供的CT图像构建方法一实施例的流程示意图；
图2是图1中步骤S30一实施例的流程示意；
图3(a)至图3(e)描述了本申请方法与其他方法的重建效果对比；
图4为本申请提供的CT设备一实施例的结构示意图；
图5为本申请提供的存储介质一实施例的结构示意图。
【具体实施方式】
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅是本申请的一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请实施例中的术语“第一”、“第二”、“第三”仅用于描述目的，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量。由此，限定有“第一”、“第二”、“第三”的特征可以明示或者隐含地包括至少一个该特征。本申请的描述中，“多个”的含义是至少两个，例如两个，三个等，除非另有明确具体的限定。此外，术语“包括”和“具有”以及它们任何变形，意图在于覆盖不排他的包含。例如包含了一系列步骤或单元的过程、方法、系统、产品或设备没有限定于已列出的步骤或单元，而是可选地还包括没有列出的步骤或单元，或可选地还包括对于这些过程、方法、产品或设备固有的其它步骤或单元。
在本文中提及“实施例”意味着，结合实施例描述的特定特征、结构或特性可以包含在本申请的至少一个实施例中。在说明书中的各个位 置出现该短语并不一定均是指相同的实施例，也不是与其它实施例互斥的独立的或备选的实施例。本领域技术人员显式地和隐式地理解的是，本文所描述的实施例可以与其它实施例相结合。
CT是用X线束对人体的某一部分一定厚度的层面进行扫描，由探测器接收透过该层面的X线，所测得的信号经过模数转换，转变为数字信息后由计算机进行处理，从而得到该层面的各个单位容积的X线吸收值即CT值，并排列成数字矩阵。这些数据信息科存储在磁光盘或磁带机中，经过数模转换后再形成模拟信号，经过计算机的一定变换后输出至显示设备上显示出图像，其密度分辨率高，可直接显示X线平面无法显示的器官和病变，它在发现病变、确定病变的位置、大小、数目方面非常敏感而可靠。CT成像本质上是人体组织的衰减系数成像，成像物理原理为通过CT扫描机构扫描获取求解衰减系数的方程组。解方程组获得人体某一体层面各个体素的衰减系数值，再将衰减系数值转换为CT值，最后将CT值变换为能视觉识别的灰度图像。CT图像重建是运用一定的物理技术测定X线在人体内的吸收系数为基础，采用一定的数学方法经计算机求解出吸收系数值在人体某剖面上的二维分布矩阵，再用电子技术把吸收系数二维分布矩阵转变诶图像面上的灰度分布，实现重建体层图像目的。每个小单元体按照扫描过程中的顺序进行排列和编号，形成一个有序得到数组，这些有序的数组在图像平面上形成图像矩阵。目前，CT图像重建算法主要包括解析法和迭代法。解析法以卷积反投影算法最为常用，解析法的优点是重建速度快，缺点是抗噪声性能较差，但是对数据的完备性要求较高。迭代法的基本思想是由测量的投影数据建立一组未知向量的代数方程式，通过方程组纠结未知图像向量。统计迭代重建算法以优化理论为基础，从投影测量过程的随机性观点出发，把图像重建看成是一个参数估计问题，通过设计合理的目标函数寻求使目标函数达到最优值的参数向量。统计迭代重建中，将提前获取的与图像相关的先验信息引入低剂量CT图像重建的目标函数中，作为正则化项，使得求解过程更加稳定，更好地恢复组织结构信息以及抑制噪声。对不完全数据也可重建图像，经典的统计迭代算法有最大似然估计算 法、最小二乘算法、基于Bayesian的最大后验算法等等。
线性代数中，将不适定问题转化为适定问题称为正则化。因此可以通过引入待建图像的先验知识来作为正则化项，使高度病态的不完备投影数据获得稳定而准确的重建。其中最常见的是全变分(Total Variation，TV)正则化迭代方法。2006年已经证明一个信号如果是可稀疏表示的，则可以利用全变分最小化作为正则化项，从少量测量数据中精确重建该信号。全变分最小化的稀疏角度CT图像重建方法每次迭代都由凸集投影和梯度下降两项组成。在此基础上，很多文献提出了一些改进的全变分正则化迭代CT图像重建方法。但是当投影数据含噪声时，利用上述方法得到的重建图像中出现了较为严重的噪声干扰和条形伪影，且部分细节被噪声掩盖，不能得到满意的重建图像。本申请发明人经过长期研究发现，传统的非二次惩罚项不稳定的主要原因是图像的粗糙度是基于相邻像素之间的强度差来计算的。当图像有噪声时，像素强度的差异在区分真实边缘和噪声波动时是不可靠的。
而现有的非局部正则化方法要么需要预先知道的参考图像来构造权重函数，要么涉及到非凸优化。在CT图像重建中，好的参考图像在重建前是不能直接得到的。由于μ的存在，非凸优化问题常导致图像估计不稳定。
为了克服这一问题，本申请发明人提出了一种基于像素邻域块的全变分重建方法，该方法在测量图像粗糙度时使用像素邻域块而不是单个像素。由于它比较了不同像素邻域块之间的相似性，因此基于像素邻域块的全变分重建比基于像素的全变分重建方法更具有鲁棒性。
参阅图1，图1是本申请提供的CT图像构建方法一实施例的流程示意图。该方法包括以下步骤：
S10：获取CT设备进行扫描时采集的投影数据。
利用X线CT成像设备在地毫安秒扫描协议下采集病人低剂量射线的CT医学图像的线积分投影数据，并同时获取相应的校正参数以及系统矩阵，其中，低剂量射线的射线剂量为标准剂量射线的1/7至1/20，可以由用户灵活设定，其中线积分投影数据是指对数变换后的投影数 据，获取的校正参数是指X线入射光子强度I 0、系统电子噪声的方差
Figure PCTCN2020098942-appb-000008
等，通常这些数值可以直接从测试设备读取或得到。利用siddon算法计算待测体中第j个组织块相对于第i个投影数据f i的贡献，记为h ij，从而得到系统矩阵H。本实施例中的投影数据可以是CT设备对病人头部扫描得到投影数据、对病人全身扫描得到的投影数据或者是对病人脊椎骨等结构细节扫描得到的投影数据，在此不做具体限制。
可选地，对扫描的投影数据进行预处理和取负对数得到处理后的投影数据。包括对投影数据进行噪声处理、扫描数据标准化处理等。
S20：利用投影数据构建图像重建模型，其中，图像重建模型是基于像素邻域块的全变分模型，像素邻域块为以一个像素为中心的像素区域。可选地，像素邻域块为以一个像素为中心，以设定像素单位为像素半径的正方形像素区域。
具体地，本申请构建的基于像素邻域块的全变分模型如下：
Figure PCTCN2020098942-appb-000009
其中，f为投影数据，μ为待重建的CT图像，H＝{H i,j}为系统矩阵，i,j分别为待重建CT图像的像素个数和CT探测器探测元个数，由具体的CT成像系统决定，可以采用不同的方法计算，比如，基于面积加权的系统矩阵计算，或者基于体素加权的系统矩阵计算。
D(f,Hμ)为数据保真项，|μ| TV表示全变分正则化项，β表示平衡保真度和正则化项的超参数。经本申请发明人反复试验，发现本申请提供的图像构建方法在平衡保真度和正则化项的超参数取β＝0.02时能够取得很好的重建效果。
更具体地，全变分正则化项为：
Figure PCTCN2020098942-appb-000010
其中s和t表示重建CT图像中衰减系数位置的指标，δ为一常数，用以保持与图像强度的可微性，μ s,t,l表示以位于重建CT图像的第s行以及第t列的像素，为中心的像素邻域块的第l个像素的像素值，μ s-1,t,l表示 以位于重建CT图像的第s-1行以及第t列的像素，为中心的像素邻域块的第l个像素的像素值，μ s,t-1,l表示以位于重建CT图像的第s行以及第t-1列的像素，为中心的像素邻域块的第l个像素的像素值，N s,t表示像素邻域块包含的像素总数。
全变分算子能使重建后的图像有较高的精度，具有各项异性的平滑作用，通过梯度的计算不但能够保护图像的边缘，而且可以较好地保持图像边界原有的对比度锐度。并且受噪声污染的图形的全变分要明显比没有经过噪声污染的图像的全变分大，通过限制全变分的大小就可以限制图像的噪声。但是，其在平滑区域的处理效果相对较差，容易出现阶梯效应，并且全变分在去除噪声的过程中一些细节如问题信息容易别过滤掉，影响重建效果。
本申请提供的基于像素邻域块的全变分正则化项代替传统全变分正则化项：
Figure PCTCN2020098942-appb-000011
也即，在对二维图像中的像素点进行梯度运算时，使用与像素相关联的像素邻域块计算二维图像的该像素的梯度。以此代替传统全变分正则项基于像素本身计算二维图像的该像素的梯度所带来的不稳定性。
可选地，本申请提供的基于像素邻域块的全变分模型的保真项为加权最小二乘算法，也就是说，在惩罚加权最小二乘全变分算法加入本申请提供的基于像素邻域块的全变分正则化项作为惩罚项，以构建惩罚加权二乘全变分重建模型。
也即，惩罚加权二乘全变分重建模型为:
Figure PCTCN2020098942-appb-000012
其中，Σ表示对角矩阵，且
Figure PCTCN2020098942-appb-000013
表示探测器信道i处投影数据的方差。可选地，探测器信道i处投影数据的方差
Figure PCTCN2020098942-appb-000014
可以通过公式：
Figure PCTCN2020098942-appb-000015
估计得到，
其中，I 0表示X射线入射光子强度，
Figure PCTCN2020098942-appb-000016
表示系统电子噪声的方差，
Figure PCTCN2020098942-appb-000017
表示探测器信道i处投影数据的均值。当然探测器信道i处投影数据的方差
Figure PCTCN2020098942-appb-000018
也可以通过局部邻域方差估计等其他方式得到，在此不做具体限制。
S30：将预设图像数据输入至图像重建模型并进行迭代运算，以得到图像重建模型输出的重建CT图像。
可选地，上述迭代运算为梯度下降算法、共轭梯度下降算法或超松弛迭代算法中的任意一种。
具体地，在迭代过程中，判断第n次迭代重建图像μ n是否满足迭代终止条件，若满足，将第n次得到的图像数据作为最终重建图像μ *。若不满足，则基于第n次迭代重建图像μ n更新CT图像，得到更新迭代重建图像μ n+1，重复迭代以得到最终重建图像μ *。其中，迭代终止条件为相邻两次迭代重建结果的相对均方误差小于设定的阈值k，即
Figure PCTCN2020098942-appb-000019
k为一正实数。比如，在一具体实施例中，迭代终止条件中设定的阈值k为0.001，阈值的取值根据实际需求设定，在此不做具体限制。
本实施例提供的CT图像构建方法通过构建基于像素邻域块的全变分模型，其中，像素邻域块为以一个像素为中心的像素区域。也即，该CT重建模型引入基于像素邻域块的全变分正则项，且该全变分正则项不同于传统的全变分正则项，本实施例提供的全变分正则项使用与像素相关联的像素邻域块计算二维图像的该像素的梯度，能够消除阶梯效应，保留图像细节，提高重建图像分辨率。
参阅图2，在一实施方式中，在步骤S30可以通过以下两个步骤进行：
S31：利用解析重建算法对投影数据进行重建，获得初始CT图像，并将初始CT图像作为预设图像数据。可选地，利用滤波反投影方法对投影数据进行重建，获得初始CT图像。
S32：将初始CT图像输入至图像重建模型并进行迭代运算，以得到 图像重建模型输出的重建CT图像。
滤波反投影的解析重建算法是典型的解析重建算法，其优点是重建速度快，在投影数据完备的情况下重建效果好。本申请为了弥补上述迭代重建算法对CT图像进行重建可能造成的迭代时间长等问题，首先利用滤波反投影重建算法对扫描得到的投影数据进行预重建，以加快重建速度，再利用本申请提供的迭代重建算法更新初始CT图像，以增加重建CT图像细节，提高重建CT图像的分辨率。
参阅图3(a)至图3(e)，图3(a)至图3(e)描述了本申请方法与其他方法的重建效果对比。图3(a)为参考图像，图3(a)是CT设备在标准剂量射线CT医学图像。图3(b)为低剂量数据采用滤波反投影方法得到的重建图像。图3(c)为低剂量图像采用惩罚加权最小二乘方法得到的迭代重建图像。图3(d)为低剂量图像采用惩罚加权最小二乘全变分方法得到的迭代重建图像，且该方法用到的全变分惩罚项为传统全变分算子。图3(e)为采用本申请方法即采用基于像素邻域块的全变分模型得到的迭代重建图像。其中，本方法采用参数为：β＝0.02，δ＝0.001。可以对比看出本申请方法保持了良好的边缘并且分辨率更高更接近参考图像图3(a)。
参阅图4，图4是本申请提供的CT设备一实施例的结构示意图，CT设备400包括：内部总线401，以及通过内部总线401连接的存储器402、处理器403。其中存储器402用于存储计算机程序。处理器403用于执行计算机程序以实现本申请提供的CT图像构建方法的步骤。
处理器403可能是一个中央处理器CPU，或者是专用集成电路ASIC(Application Specific Integrated Circuit)，或者是被配置成实施本申请实施例的一个或多个集成电路。存储器402用于可执行的指令。存储器402可能包含高速RAM存储器，也可能包括非易失性存储器(non-volatile memory)，例如至少一个磁盘存储器。存储器402也可以是存储器阵列。存储器402还可能被分块，并且块可按一定的规则组合成虚拟卷。存储器402存储的指令可被处理器403执行，以使处理器403能够执行上述任意方法实施例中的CT图像构建的方法。
参阅图5，图5是本申请提供的计算机可读存储介质一实施例的结构示意图。该计算机可读存储介质500上存储有计算机程序501，计算机程序501被处理器执行时实现本申请提供的CT图像构建方法的步骤。计算机存储介质500可以是计算机能够存取的任何可用介质或数据存储设备，包括但不限于磁性存储器(例如软盘、硬盘、磁带、磁光盘(MO)等)、光学存储器(例如CD、DVD、BD、HVD等)、以及半导体存储器(例如ROM、EPROM、EEPROM、非易失性存储器110(NANDFLASH)、固态硬盘(SSD))等。
本实施例提供的CT图像构建方法通过构建基于像素邻域块的全变分模型。也即，该CT重建模型引入基于像素邻域块的全变分正则项，且该全变分正则项不同于传统的全变分正则项，本实施例提供的全变分正则项使用与像素相关联的像素邻域块计算二维图像的该像素的梯度，能够消除阶梯效应，保留图像细节，提高重建CT图像的分辨率。
另外本申请为了弥补上述迭代重建算法对CT图像进行重建可能存在的迭代时间长等问题，首先利用滤波反投影重建算法对扫描得到的投影数据进行预重建，以加快重建速度，再利用本申请提供的迭代重建算法更新初始CT图像，以增加最终CT图像细节，使得到的CT图像更清晰。综上，本申请提供的CT图像构建方法既能加快重建速度，又能提高重建质量。
以上，仅为本申请中的具体实施方式，但本申请的保护范围并不局限于此，任何熟悉该技术的人在本申请所揭露的技术范围内，可理解想到的变换或替换，都应涵盖在本申请的包含范围之内，因此，本申请的保护范围应该以权利要求书的保护范围为准。

Claims (12)

  1. 一种CT图像的构建方法，其特征在于，所述方法包括：
    获取CT设备进行扫描时采集的投影数据；
    利用所述投影数据构建图像重建模型，其中，所述图像重建模型是基于像素邻域块的全变分模型，所述像素邻域块为以一个像素为中心的像素区域；
    将预设图像数据输入至所述图像重建模型并进行迭代运算，以得到所述图像重建模型输出的重建CT图像。
  2. 根据权利要求1所述的方法，其特征在于，所述将预设图像数据输入至所述图像重建模型并进行迭代运算，以得到所述图像重建模型输出的重建CT图像，包括：
    利用解析重建算法对所述投影数据进行重建，获得初始CT图像，并将所述初始CT图像作为所述预设图像数据；
    将所述初始CT图像输入至所述图像重建模型并进行迭代运算，以得到所述图像重建模型输出的重建CT图像。
  3. 根据权利要求2所述的方法，其特征在于，所述利用解析重建算法对所述投影数据进行重建，获得初始CT图像，包括：
    利用滤波反投影算法对所述投影数据进行重建，获得初始CT图像。
  4. 根据权利要求1或2所述的方法，其特征在于，所述基于像素邻域块的全变分模型为：
    Figure PCTCN2020098942-appb-100001
    其中，f为所述投影数据，μ为待重建的所述重建CT图像，H＝{H i,j}为系统矩阵，i,j分别为待重建所述重建CT图像的像素个数和所述CT设备探测器的探测元个数，D(f,Hμ)为数据保真项，|μ| TV表示全变分正则化项，β表示平衡保真度和正则化项的超参数。
  5. 根据权利要求4所述的方法，其特征在于，所述全变分正则化项为：
    Figure PCTCN2020098942-appb-100002
    其中s和t表示所述重建CT图像中衰减系数位置的指标，δ为一常数，用以保持与图像强度的可微性，μ s,t,l表示以位于所述重建CT图像的第s行以及第t列的像素，为中心的像素邻域块的第l个像素的像素值，μ s-1,t,l表示以位于所述重建CT图像的第s-1行以及第t列的像素，为中心的像素邻域块的第l个像素的像素值，μ s,t-1,l表示以位于所述重建CT图像的第s行以及第t-1列的像素，为中心的像素邻域块的第l个像素的像素值，N s,t表示所述像素邻域块包含的像素总数。
  6. 根据权利要求5所述的方法，其特征在于，所述基于像素邻域块的全变分模型为惩罚加权最小二乘全变分算法，其中，所述惩罚加权最小二乘全变分算法引入所述基于像素邻域块的全变分正则项为惩罚项。
  7. 根据权利要求6所述的方法，其特征在于，所述基于像素邻域块的全变分模型为：
    Figure PCTCN2020098942-appb-100003
    其中，Σ表示对角矩阵，且
    Figure PCTCN2020098942-appb-100004
    表示探测器信道i处的所述投影数据的方差。
  8. 根据权利要求7所述的方法，其特征在于，所述探测器信道i处的所述投影数据的方差根据以下公式获取：
    Figure PCTCN2020098942-appb-100005
    其中，I 0表示X射线入射光子强度，
    Figure PCTCN2020098942-appb-100006
    表示系统电子噪声的方差，
    Figure PCTCN2020098942-appb-100007
    表示所述探测器信道i处投影数据的均值。
  9. 根据权利要求4所述的方法，其特征在于，所述系统矩阵是从所述CT设备系统得到的基于面积加权的系统矩阵，或者基于体素加权的系统矩阵。
  10. 根据权利要求1所述的方法，其特征在于，所述迭代运算为梯 度下降算法、共轭梯度下降算法或超松弛迭代算法中的任意一种。
  11. 一种CT设备，其特征在于，所述CT设备包括：
    内部总线，以及通过内部总线连接的存储器和处理器；
    所述存储器用于存储计算机程序；
    所述处理器用于执行所述计算机程序以实现如权利要求1-10任意一项所述的CT图像构建方法的步骤。
  12. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质上存储有计算机程序，所述计算机程序被处理器执行时实现如权利要求1-10任意一项所述的CT图像构建方法的步骤。
PCT/CN2020/098942 2020-06-29 2020-06-29 一种ct图像的构建方法、ct设备以及存储介质 WO2022000192A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/098942 WO2022000192A1 (zh) 2020-06-29 2020-06-29 一种ct图像的构建方法、ct设备以及存储介质

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2020/098942 WO2022000192A1 (zh) 2020-06-29 2020-06-29 一种ct图像的构建方法、ct设备以及存储介质

Publications (1)

Publication Number Publication Date
WO2022000192A1 true WO2022000192A1 (zh) 2022-01-06

Family

ID=79315041

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/098942 WO2022000192A1 (zh) 2020-06-29 2020-06-29 一种ct图像的构建方法、ct设备以及存储介质

Country Status (1)

Country Link
WO (1) WO2022000192A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116071450A (zh) * 2023-02-07 2023-05-05 深圳扬奇医芯智能科技有限公司 鲁棒的低剂量ct成像算法、装置
CN116664707A (zh) * 2022-02-17 2023-08-29 北京朗视仪器股份有限公司 一种ct图像迭代重建方法、装置及计算机设备
CN116977473A (zh) * 2023-09-21 2023-10-31 北京理工大学 一种基于投影域和图像域的稀疏角ct重建方法及装置
CN117078662A (zh) * 2023-10-11 2023-11-17 杭州睿影科技有限公司 一种叠片式电池的检测方法、装置、图像处理设备及介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060062443A1 (en) * 2004-09-23 2006-03-23 General Electric Company System and method for reconstruction of cone beam tomographic projections with missing data
US20080240335A1 (en) * 2007-03-30 2008-10-02 General Electric Company Iterative reconstruction of tomographic image data method and system
CN106127825A (zh) * 2016-06-15 2016-11-16 赣南师范学院 一种基于广义惩罚加权最小二乘的x射线ct图像重建方法
CN108460723A (zh) * 2018-02-05 2018-08-28 西安电子科技大学 基于邻域相似性的双边全变分图像超分辨率重建方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20060062443A1 (en) * 2004-09-23 2006-03-23 General Electric Company System and method for reconstruction of cone beam tomographic projections with missing data
US20080240335A1 (en) * 2007-03-30 2008-10-02 General Electric Company Iterative reconstruction of tomographic image data method and system
CN106127825A (zh) * 2016-06-15 2016-11-16 赣南师范学院 一种基于广义惩罚加权最小二乘的x射线ct图像重建方法
CN108460723A (zh) * 2018-02-05 2018-08-28 西安电子科技大学 基于邻域相似性的双边全变分图像超分辨率重建方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
QI ZEYAO, WANG YUAN-JUN: "Total Variation Regularization Sparse-view Reconstruction Algorithm Combined with Neigh- borhood Information", JOURNAL OF CHINESE COMPUTER SYSTEMS, GAI-KAN BIANJIBU , SHENYANG, CN, vol. 40, no. 8, 31 August 2019 (2019-08-31), CN , pages 1745 - 1749, XP055883636, ISSN: 1000-1220 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116664707A (zh) * 2022-02-17 2023-08-29 北京朗视仪器股份有限公司 一种ct图像迭代重建方法、装置及计算机设备
CN116664707B (zh) * 2022-02-17 2024-02-13 北京朗视仪器股份有限公司 一种ct图像迭代重建方法、装置及计算机设备
CN116071450A (zh) * 2023-02-07 2023-05-05 深圳扬奇医芯智能科技有限公司 鲁棒的低剂量ct成像算法、装置
CN116071450B (zh) * 2023-02-07 2024-02-13 深圳扬奇医芯智能科技有限公司 鲁棒的低剂量ct成像算法、装置
CN116977473A (zh) * 2023-09-21 2023-10-31 北京理工大学 一种基于投影域和图像域的稀疏角ct重建方法及装置
CN116977473B (zh) * 2023-09-21 2024-01-26 北京理工大学 一种基于投影域和图像域的稀疏角ct重建方法及装置
CN117078662A (zh) * 2023-10-11 2023-11-17 杭州睿影科技有限公司 一种叠片式电池的检测方法、装置、图像处理设备及介质

Similar Documents

Publication Publication Date Title
WO2022000192A1 (zh) 一种ct图像的构建方法、ct设备以及存储介质
JP2020168352A (ja) 医用装置及びプログラム
US20210209771A1 (en) Systems and methods for determining quality metrics of an image or images based on an edge gradient profile and characterizing regions of interest in an image or images
JP2019516460A (ja) 空間とスペクトル情報に基づく複数エネルギーのｃｔ画像におけるノイズ制御のためのシステムと方法
US20030076988A1 (en) Noise treatment of low-dose computed tomography projections and images
US11276208B2 (en) System, method and computer-accessible medium for ultralow dose computed tomography image reconstruction
US20110286651A1 (en) Projection-Space Denoising with Bilateral Filtering in Computed Tomography
JP5590548B2 (ja) Ｘ線ｃｔ画像処理方法，ｘ線ｃｔプログラムおよび該プログラムが搭載されたｘ線ｃｔ装置
WO2014172938A1 (zh) Ct图像重建方法
Zeng et al. Cerebral perfusion computed tomography deconvolution via structure tensor total variation regularization
Niu et al. Iterative reconstruction for sparse-view x-ray CT using alpha-divergence constrained total generalized variation minimization
US20220375038A1 (en) Systems and methods for computed tomography image denoising with a bias-reducing loss function
Xu et al. Statistical iterative reconstruction to improve image quality for digital breast tomosynthesis
CN112529977B (zh) 一种pet图像重建的方法和系统
Kim et al. An effective post-filtering framework for 3-D PET image denoising based on noise and sensitivity characteristics
JP7187131B2 (ja) 画像生成装置、ｘ線コンピュータ断層撮影装置及び画像生成方法
JP7362460B2 (ja) 医用画像処理装置、方法及び記憶媒体
EP3920144A1 (en) Methods and systems for multi-material decomposition
CN111968192B (zh) 一种ct图像的构建方法、ct设备以及存储介质
CN115439423A (zh) 一种基于ct图像的识别方法、装置、设备及存储介质
Ghane et al. Quantitative analysis of image quality in low-dose computed tomography imaging for COVID-19 patients
CN105488824B (zh) 一种重建pet图像的方法和装置
Arya et al. Regularization based modified SART iterative method for CT image reconstruction
Yang et al. Sparse angle CT reconstruction with weighted dictionary learning algorithm based on adaptive group-sparsity regularization
CN111968192A (zh) 一种ct图像的构建方法、ct设备以及存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20943208

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20943208

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 20943208

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 14.08.2023)

122 Ep: pct application non-entry in european phase

Ref document number: 20943208

Country of ref document: EP

Kind code of ref document: A1