WO2018103015A1 - 一种环形伪影修正的方法及装置 - Google Patents

一种环形伪影修正的方法及装置 Download PDF

Info

Publication number
WO2018103015A1
WO2018103015A1 PCT/CN2016/108894 CN2016108894W WO2018103015A1 WO 2018103015 A1 WO2018103015 A1 WO 2018103015A1 CN 2016108894 W CN2016108894 W CN 2016108894W WO 2018103015 A1 WO2018103015 A1 WO 2018103015A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
ring
artifact
iteration
unit
Prior art date
Application number
PCT/CN2016/108894
Other languages
English (en)
French (fr)
Inventor
梁晓坤
谢耀钦
张志诚
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/CN2016/108894 priority Critical patent/WO2018103015A1/zh
Publication of WO2018103015A1 publication Critical patent/WO2018103015A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20004—Adaptive image processing
    • G06T2207/20012—Locally adaptive

Definitions

  • the present invention relates to medical image processing techniques, and more particularly to a method and apparatus for ring pseudo-correction.
  • CBCT cone beam computed tomography
  • Artifacts are important factors that restrict the quality of CBCT images and thus affect clinical diagnosis.
  • ring artifacts are the most serious.
  • the ring artifact appears on the CBCT image as a series of concentric rings of a certain width centered on the center of the CBCT image and the gray scale is distinguished from the surrounding pixels.
  • the existence of ring artifacts brings a lot of troubles to subsequent CBCT image processing and clinical diagnosis. Therefore, how to correct ring artifacts in CBCT images to reduce or eliminate ring artifacts in CBCT images is very important.
  • Pre-processing techniques based on projection fields can utilize sinograms because in a sinogram, the ring artifacts of the image appear as straight lines, making it easier to obtain information about the ring artifacts.
  • a technique for removing ring artifacts from a sinogram was proposed.
  • a simple low-pass filter is used in the sinogram to eliminate discontinuous signals due to ring artifacts, but this operation disturbs the high frequency signal of the original image and affects image quality.
  • Raven.C represented a ring-shaped artifact as a parallel straight line on a sinogram, and only had a sharp change in the horizontal direction. Fourier transform the sinogram, and vertically in the frequency domain.
  • Polar coordinate transformation of images is one of the popular image domain-based post-processing techniques. Specifically, the CT reconstructed image is converted from Cartesian coordinates to polar coordinates, and after processing, converted to Cartesian coordinates.
  • Jan Sijbers proposed a ring artifact correction method based on morphological operators. In the space coordinate system, the morphological operator is used to separate the region of interest in the CBCT from the background image, and then the separated image is converted into polar coordinates. After the image processing, a artifact template is obtained, and finally the corrected image is obtained by doing the difference. But this method is too dependent on the choice of parameters.
  • Chen.Y proposed a ring artifact correction method based on independent component analysis, but this method affects the details of the image.
  • the existing image domain-based post-processing technology when identifying and correcting artifacts, easily destroys the original image details of the image, causing the corrected image to deviate from the objective image, affecting the doctor's judgment on the objective image;
  • Many post-processing techniques reduce the resolution of the image in the process of correcting the image ring artifacts using polar coordinates, which is not suitable for clinical use.
  • a method for providing ring artifact correction including:
  • Coordinate conversion step converting the original image from Cartesian coordinates to polar coordinates, and as an input image in the processing step, thereby performing a processing step;
  • Processing step performing processing for eliminating image details and ring artifacts on the input image, and obtaining a template image for removing ring shape and image detail;
  • Subtracting step subtracting the template image from the original image in the polar coordinates to obtain a residual image
  • Extracting step extracting the residual image to obtain a ring-shaped artifact image
  • Judgment step judging whether the current iteration condition is reached
  • Iterative step when it is judged to continue the iteration, the current template image is compensated according to the current residual image and the current ring-shaped artifact image to obtain a compensated image, and the compensated image is used as an input image in the processing step. And performing processing steps;
  • Correction step When it is judged to stop the iteration, the original image is corrected using the current ring artifact image to obtain the ring artifact corrected image.
  • an apparatus for providing ring artifact correction including:
  • a coordinate conversion unit for converting the original image from Cartesian coordinates to polar coordinates, and inputting to the processing unit as an input image in the processing unit;
  • a processing unit configured to perform processing for eliminating image details and ring artifacts on the input image, to obtain a template image that removes ring artifacts and image details;
  • a subtracting unit configured to subtract the template image from the original image in the polar coordinates to obtain a residual image
  • An extracting unit configured to extract the residual image to obtain a ring-shaped artifact image
  • a determining unit configured to determine whether the current iteration condition is reached
  • An iteration unit configured to compensate the current template image according to the current residual image and the current ring artifact image to obtain a compensation image, and use the compensation image as a processing unit Input the image and input it to the processing unit for processing;
  • the correcting unit is configured to correct the original image by using the current ring artifact image to obtain the ring artifact corrected image when determining to stop the iteration.
  • the method and apparatus for ring artifact correction according to the above implementation will be based on the current residual map
  • the image and the ring artifact image compensate the current template image to obtain a compensated image as an object of each iteration, so that in the process of eliminating the ring artifact, the details of the original image are not damaged, and the resolution of the image is not reduced;
  • the extraction of ring artifacts is performed only in the image domain, so the present application is fully compatible with the flow of clinical use.
  • FIG. 1 is a schematic flow chart of a method for ring artifact correction according to an embodiment of the present application
  • FIG. 2 is a schematic diagram of conversion of image processing in a method of ring artifact correction according to an embodiment of the present application
  • FIG. 3(a) is a schematic structural diagram of a device for ring artifact correction according to an embodiment of the present application
  • FIG. 3(b) is a schematic structural diagram of a device for ring artifact correction according to another embodiment of the present application
  • FIG. 4 is a diagram showing experimental results performed by the method and apparatus for ring artifact correction implemented by the present application.
  • the cone beam CT (CBCT) image reconstructed by the FDK algorithm produces severe ring artifacts, which destroys the details of the image. Further, this Errors in the application of CBCT for radiotherapy placement and industrial material testing.
  • the present invention can effectively eliminate ring artifacts without destroying the original image information.
  • the present application discloses a method for ring artifact correction, which includes a coordinate conversion step S01, a processing step S03, a subtraction step S05, an extraction step S07, and a determination step S09.
  • the iterative step S11 and the correction step S13 are specifically described below.
  • Coordinate conversion step S01 The original image is converted from Cartesian coordinates to polar coordinates, and as an input image in the processing step S03, a processing step S03 is performed.
  • the coordinate transformation step S01 includes converting the original image from Cartesian coordinates to polar coordinates using cubic spline interpolation. Since the ring of the ring artifact appears as a strip in polar coordinates, the original image is converted from Cartesian coordinates to polar coordinates, which facilitates better extraction of artifact information later. It should be noted that the original image in the coordinate conversion step S01 may be the original CBCT image.
  • Processing step S03 performing processing for eliminating image details and ring artifacts on the input image to obtain a template image from which ring artifacts and image details are removed.
  • the original image in the polar coordinate in the coordinate conversion step S01 is input in the processing step S03.
  • the input in the processing step S03 is the compensation in the iterative step S11.
  • the processing step S03 includes performing a Total Total-Variance (RTV) smoothing process on the input image to eliminate image details and ring artifacts, thereby removing ring artifacts and image details. Template image.
  • RTV Total Total-Variance
  • the objective function of the RTV smoothing in step S03 is as follows:
  • I p is the input image
  • p is the pixel index of the image
  • S is the obtained template image
  • (S p -I p ) 2 is the smoothing term.
  • the pixel index q belongs to the window R(p)
  • the partial derivatives of the two directions, g p,q are the weighting functions of the spatial correlation degree.
  • ⁇ is the weighting of the image smoothing intensity
  • ⁇ is a very small positive number, mainly to avoid the occurrence of zero. Since the objective function is a non-convex function, the present application can solve this objective function by a quadratic penalty method.
  • Subtracting step S05 The template image obtained in the processing step S03 is subtracted from the original image in the polar coordinates in the coordinate conversion step S01 to obtain a residual image. Since the residual image is obtained by subtracting the template image that does not include ring artifacts and image details from the original image, the residual image is actually an image containing image details and ring artifacts.
  • Extraction step S07 Extracting the residual image to obtain a ring-shaped artifact image. Due to the inconsistent detector gain and the like, the strip artifacts can be estimated to be the same pixel value in polar coordinates, so the residual image can be extracted in the angular direction of the polar coordinates to obtain a circular artifact image. Therefore, in a preferred embodiment, the extracting step S07 includes extracting the residual image in the angular direction of the polar coordinates to obtain a circular artifact image. In a specific embodiment, the extracting step S07 includes performing a median extraction on the residual image in the angular direction of the polar coordinates to obtain a circular artifact image.
  • the extracting step S07 includes: pixels that are not too high and not too low in the residual image when the residual image is extracted in the angular direction of the polar coordinates Performing extraction, further, may perform median extraction; specifically, extracting pixel points of non-over-high values and non-low-value values in the residual image, and setting a pixel threshold, and setting too high pixel values and The pixel points of the pixel values that are too low are excluded, for example, the exclusion of the pixel whose value is lower than 10% and higher than 90%, and the intermediate 80% is taken as the extracted sample value.
  • the determining step S09 determining whether the current iteration condition is reached.
  • the determining step S09 includes determining whether to stop the iteration according to the current ring-shaped artifact image and the previous ring-shaped artifact image, which is an iterative stop condition; and/or determining whether to stop the iteration according to the current number of iterations, If the current number of iterations reaches the set number of iterations threshold, it is judged to stop the iteration.
  • the determining step S09 includes determining a second norm of the difference between the current shaped artifact image and the previous annular artifact image. When the two norm is less than a threshold, determining to stop the iteration, and vice versa, Then judge to continue the iteration. In an embodiment, the determining step S09 may use the following formula as a condition for determining the two norm:
  • the threshold value can be set to 0.002, less than 0.002 s d when the iteration is stopped, otherwise, continue iterating.
  • the determining step S09 is based on the current ring-shaped artifact image and the previous ring-shaped artifact image, it is determined whether to stop the iteration, so when the initial iteration is performed, the ring artifact is present. The image does not have the previous ring-shaped artifact image. At this time, in the determining step S09, it is determined that the iteration needs to be continued.
  • the previous ring-shaped artifact image that does not exist at the initial iteration may be set as a blank image, or may be The number of times of this iteration is 1, and it is directly judged to continue the iteration.
  • Iterative step S11 when iteratively judges to continue the iteration, the current template image is compensated according to the current residual image and the current ring artifact image to obtain a compensation image, and the compensation image is taken as the processing step S03.
  • the image is input, and processing step S03 is performed.
  • the iterative step S11 includes: when iteratively continues to iterate, subtracting the current ring artifact image from the current residual image to obtain a detail image; and compensating the detail image into the current template image, To get a compensated image.
  • the detail image is compensated to the current template image, and the detail image may be added to the current template image, and the added image is the compensation image.
  • the iterative step S11 is substantially to output the compensation image to the processing step S03, so that the compensation image is re-started from the processing step S03, and then the subtraction step S05, the extraction step S07, the determination step S09 are sequentially performed, and in the determination step S09, it is determined whether or not To continue the iteration, the loop is repeated until it is determined in the judgment step S09 that the iteration is stopped in a certain iteration, and then the correction step S13 is performed.
  • the iteration step S11 outputs the current compensation image to the processing step S03 every time to perform iteration, which ensures the effect of eliminating the ring artifact.
  • Correction step S13 When it is determined that the iteration is stopped, the original image is corrected using the current ring artifact image to obtain a ring artifact corrected image.
  • the correcting step S13 includes: when it is determined to stop the iteration, the current annular artifact image is converted from polar coordinates to Cartesian coordinates, and the original image in Cartesian coordinates is subtracted from the current circular artifact image in Cartesian coordinates. , get the image after the ring artifact correction.
  • the present application also discloses a device for ring artifact correction, which includes a coordinate conversion unit 01, a processing unit 03, a subtraction unit 05, and an extraction unit. 07.
  • the judging unit 09, the iterating unit 11, and the correcting unit 13 are specifically described below.
  • the coordinate conversion unit 01 is for converting the original image from Cartesian coordinates to polar coordinates, and as an input image in the processing unit 03, is input to the processing unit 03 for processing.
  • coordinate transformation unit 01 includes an interpolation unit 01a for converting the original image from Cartesian coordinates to polar coordinates using cubic spline interpolation. Since the ring of the ring artifact appears as a strip in polar coordinates, the original image is converted from Cartesian coordinates to polar coordinates, which facilitates better extraction of artifact information later. It should be noted that the original image processed in the coordinate conversion unit 01 may be an original CBCT image.
  • the processing unit 03 is configured to perform the process of eliminating image details and ring artifacts on the input image described above, and obtaining a template image that removes ring artifacts and image details. It should be noted that, at the initial iteration, the original image in the polar coordinate in the coordinate conversion unit 01 is input in the processing unit 03. In the subsequent iteration, the input in the processing unit 03 is the compensation in the iteration unit 11. image.
  • the processing unit 03 includes a correlation total variation unit 03a that performs a correlation total variation (RTV) smoothing process on the input image to eliminate image detail and ring shape. Artifacts, resulting in a template image that removes the ring and image detail.
  • the target function of the RTV smoothing in processing unit 03 is as follows:
  • I p is the input image
  • p is the pixel index of the image
  • S is the obtained template image
  • (S p -I p ) 2 is the smoothing term.
  • the pixel index q belongs to the window R(p)
  • the partial derivatives of the two directions, g p,q are the weighting functions of the spatial correlation degree.
  • ⁇ is the weighting of the image smoothing intensity
  • ⁇ is a very small positive number, mainly to avoid the occurrence of zero. Since the objective function is a non-convex function, the present application can solve this objective function by a quadratic penalty method.
  • the subtraction unit 05 is for subtracting the template image from the original image in polar coordinates to obtain a residual image. Since the residual image is obtained by subtracting the original image from a template image that does not include shape artifacts and image details, the residual image is actually an image containing image details and ring artifacts.
  • the extracting unit 07 is configured to extract the residual image to obtain a ring-shaped artifact image. Due to the inconsistent detector gain and the like, the strip artifacts can be estimated to be the same pixel value in polar coordinates, so the residual image can be extracted in the angular direction of the polar coordinates to obtain a circular artifact image.
  • the extracting unit 07 includes a direction extracting unit 07a for extracting the residual image in the angular direction of the polar coordinates to obtain a ring-shaped artifact image.
  • the direction extracting unit 07a includes a median extracting unit 07b for performing median extraction on the residual image in the angular direction of the polar coordinates to obtain a ring-shaped artifact image.
  • the extracting unit 07 includes a selection extracting unit 07c for extracting pixel points of non-over-high values and non-under-low values in the residual image to Obtaining a ring-shaped artifact image, preferably, the selection extracting unit 07c may extract the pixel points of the residual image that are not too high and not too low in the angular direction of the polar coordinate, and further, The selection extraction unit 07c may perform median extraction; when the selection extraction unit 07c is implemented, when extracting pixel points that are not too high and not too low in the residual image, a pixel threshold may be set. Pixel points with high pixel values and too low pixel values are excluded, for example, the exclusion of pixels whose values are below 10% and above 90%, and the remaining 80% as
  • the judging unit 09 judges whether or not the stop iteration condition is currently reached.
  • the determining unit 09 includes a comparing unit 09a and/or a number of times unit 09b.
  • the comparing unit 09a is configured to determine whether to stop the iteration according to the current ring-shaped artifact image and the previous ring-shaped artifact image;
  • the number of times unit 09b is used to determine whether to stop the iteration according to the current number of iterations. If the current number of iterations reaches the set number of iterations threshold, it is judged to stop the iteration.
  • the comparing unit 09a includes a two norm comparing unit 09c, and the two norm comparing unit 09c is configured to determine a two norm of the difference between the current shaped artifact image and the previous annular artifact image.
  • the two norm comparison unit 09c may use the following formula as a judgment condition:
  • the threshold value can be set to 0.002, less than 0.002 s d when the iteration is stopped, otherwise, continue iterating.
  • the comparison unit 09a determines whether to stop the iteration according to the current ring-shaped artifact image and the previous ring-shaped artifact image, when the initial iteration is performed, the current artifact image does not have the previous ring-shaped pseudo-pseudo. In the image unit, the comparison unit 09a also judges that the iteration needs to be continued.
  • the previous ring artifact image that does not exist at the initial iteration may be set as a blank image, or the number of times of the generation may be 1, Direct judgment continues to iterate.
  • the iterative unit 11 is configured to compensate the current template image according to the current residual image and the current ring artifact image to obtain a compensation image, and use the compensation image as the processing unit 03.
  • the input image is input to the processing unit 03 for processing.
  • the iteration unit 11 includes an iteration sub-unit 11a for subtracting the current ring artifact image from the current residual image to obtain a detail image when it is determined to continue the iteration;
  • the detail image is compensated into the current template image to obtain a compensated image.
  • the iteration sub-unit 11a compensates the detail image to the current template image, and may add the detail image to the current template image, and the added image is the compensation image.
  • the iteration unit 11 essentially outputs the compensation image to the processing unit 03, so that the compensation image is re-started from the processing unit 03, and then sequentially operated by the subtraction unit 05, the extraction unit 07, and the determination unit 09, and judged in the determination unit 09. Whether or not to continue the iteration, the loop is repeated until it is determined in the judgment unit 09 that the iteration is stopped in a certain iteration, and then the correction unit 13 is performed.
  • the iteration unit 11 outputs the current compensation image to the processing unit 03 every time for iteration, which ensures the effect of eliminating the ring artifact.
  • the correcting unit 13 is configured to correct the original image using the ring artifact image to obtain the ring artifact corrected image when it is determined to stop the iteration.
  • the correction unit 13 includes a correction sub-unit 13a for converting the current ring-shaped artifact image from polar coordinates to Cartesian coordinates when determining to stop the iteration, and Descartes The original image under coordinates is subtracted from the ring-shaped artifact image in Cartesian coordinates to obtain a ring-shaped artifact-corrected image.
  • FIG. 4 is an experimental result diagram of a method and apparatus for applying ring artifact correction according to the present application.
  • Columns 1 to 3 are CBCT images of different layers respectively; images of the first row are original CBCT images (ie, ring pseudo The image before the image correction); the image of the second line is the image corrected by the ring artifact.
  • the image corrected by the ring artifact not only eliminates the influence of the ring artifact on the image, but also Effectively preserves image detail and image Resolution;
  • the image of the third line is the ring-shaped artifact image extracted by the final iteration. It can be seen that the extracted ring-shaped artifact image is very good, so that the extracted ring-shaped artifact image is corrected in the original step in the correction step. The effect is naturally good too.
  • the method and device for ring artifact correction implements iterative ring artifact correction by using polar coordinate transformation, related total variation processing, median extraction and iterative correction, and specifically, using the relevant total variation to edge the image Smoothing of protection, obtaining details and artifact images, applied to ring artifact correction; using iterative correction method, continuously extracting ring artifacts to achieve the result of image ring artifact correction.
  • the present application does not damage the details of the original image and does not reduce the resolution of the image, and the processing flow is simple; and the present application only extracts the ring artifact in the image domain, so its Handle processes that are fully compatible with clinical use.
  • GPU acceleration can also be utilized to greatly reduce the time of ring artifact correction.

Landscapes

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

Abstract

本申请公开了一种环形伪影修正的方法及装置，利用极坐标转化、相关总变分处理、中值提取和迭代修正等实现迭代环形伪影修正，具体地，利用相关总变分对图像进行边缘保护的平滑，得到细节和伪影图像，应用于环形伪影修正；利用迭代修正的方法，不断提取环形伪影，达到图像环形伪影修正的结果。本申请在去除环形伪影过程中，做到了不损坏原始图像的细节和不降低图像的分辨率，并且处理流程简单等；另外本申请仅在图像域中对环形伪影进行提取，因此它的处理完全兼容临床使用的流程。为了更加切合临床应用。

Description

一种环形伪影修正的方法及装置 技术领域
本发明涉及医学图像处理技术，特别涉及一种环形伪影修正的方法及装置。
背景技术
在医学诊断中，常常需要通过对患处拍摄从而得到图像，再进行分析。其中CBCT(锥形束CT，Cone beam Computer Tomography)技术已经广泛应用于医学诊断的诸多领域。伪影是制约CBCT图像质量进而影响临床诊断的重要因素，其中又尤以环形伪影最为严重。产生环形伪影的原因有很多，例如探测器上的探测元响应不一致等。环形伪影在CBCT图像上表现为以CBCT图像中心为圆心且灰度区别于周围像素的一系列的具有一定宽度的同心圆环。环形伪影的存在，给后续的CBCT图像处理和临床诊断带来很多的困扰，因此如何对CBCT图像中的环形伪影进行修正，以减轻或消除CBCT图像中的环形伪影显得十分重要。
近几年许多学者提出了各种CBCT环形伪影修正方法。这些方法可以分为两大类：基于投影域的前处理技术和基于图像域的后处理技术。
基于投影域的前处理技术可以利用正弦图，因为在正弦图中，图像的环形伪影表现为直线，从而更容易获取环形伪影的信息。1978年就有人提出从正弦图中相除环形伪影的技术。在正弦图中利用一个简单的低通滤波器消除由于环形伪影引起的不连续信号，但是这个操作扰乱了原始图像的高频信号，影响图像质量。最近几年，又出现了一些方法来解决这个问题。1998年，Raven.C根据环形伪影在正弦图上表现为平行直线，而且仅在水平方向上有剧烈变化这一性质，对正弦图进行傅立叶变换，在频域中，竖直方向上成列的伪影表现为水平方向上位于图像中心的高频信号。因此，在这个区域与方向上对图像进行低通滤波就可以去除高频细节信息。最近，Cem Altunbasa通过识别探测器像素增益，获取像素增益校正数据集，以达到环形伪影修正的目的。
基于投影域的前处理技术不仅需要占用大量的计算机内存，而且还需花费大量的计算时间。因此，基于图像域的后处理技术的环形伪影修正方法成为了研究人员的首选方向。
对图像进行极坐标转化是流行的基于图像域的后处理技术之一。具体地，将CT重建图像由笛卡尔坐标转化为极坐标，通过处理后，再转化为笛卡尔坐标。2004年Jan Sijbers提出了一种基于形态学算子的环形伪影修正方法。该方法在空间坐标系下，利用形态学算子将CBCT中的感兴趣区域从背景图像中分离出来，然后将分离的图像转化为极坐标， 进行图像处理后，得到一个伪影模版，最后通过做差得到修正的图像。但是该方法太过依赖参数的选择。随后，在2009年Chen.Y提出了一个基于独立成分分析的环形伪影修正方法，但是该方法影响了图像的细节信息。现有的基于图像域的后处理技术，在对伪影进行识别与修正时，容易破坏图像原有的图像细节，导致修正后的图像与客观图像产生偏离，影响医生对客观图像的判断；此外，很多后处理技术在利用极坐标对图像环形伪影进行修正的过程中，降低了图像的分辨率，不适用于临床。
发明内容
根据第一方面，一种实施例中提供一种环形伪影修正的方法，包括：
坐标转换步骤：将原始图像由笛卡尔坐标转换为极坐标，并并作为处理步骤中的输入图像，进而执行处理步骤；
处理步骤：对输入的图像进行消除图像细节和环形伪影的处理，得到去除环形和图像细节的模版图像；
相减步骤：将所述极坐标下的原始图像减去所述模版图像，得到残差图像；
提取步骤：对所述残差图像进行提取，以得到环形伪影图像；
判断步骤：判断当前是否达到停止迭代条件；
迭代步骤：当判断继续迭代时，则根据本次残差图像和本次环形伪影图像来对本次的模版图像进行补偿，以得到补偿图像，并将该补偿图像作为处理步骤中的输入图像，进而执行处理步骤；
修正步骤：当判断停止迭代时，则使用本次的环形伪影图像对原始图像进行修正，以获取环形伪影修正后的图像。
根据第二方面，一种实施例中提供一种环形伪影修正的装置，包括：
坐标转换单元，用于将原始图像由笛卡尔坐标转换为极坐标，并作为处理单元中的输入图像，输入给处理单元进行处理；
处理单元，用于对所述输入图像进行消除图像细节和环形伪影的处理，得到去除环形伪影和图像细节的模版图像；
相减单元，用于将所述极坐标下的原始图像减去所述模版图像，得到残差图像；
提取单元，用于对所述残差图像进行提取，以得到环形伪影图像；
判断单元，用于判断当前是否达到停止迭代条件；
迭代单元，用于当判断继续迭代时，则根据本次残差图像和本次环形伪影图像来对本次的模版图像进行补偿，以得到补偿图像，并将该补偿图像作为处理单元中的输入图像，输入给处理单元进行处理；
修正单元，用于当判断停止迭代时，则使用本次的环形伪影图像对原始图像进行修正，以获取环形伪影修正后的图像。
依上述实施的环形伪影修正的方法及装置，由于将基于本次残差图 像和环形伪影图像对本次的模版图像进行补偿得到补偿图像，作为每次迭代的对象，从而在消除环形伪影的过程中，不损坏原始图像的细节，不降低图像的分辨率；并且对环形伪影进行提取是仅在图像域中进行，因此本申请完全兼容临床使用的流程。
附图说明
图1为本申请一种实施例的环形伪影修正的方法的流程示意图；
图2为本申请一种实施例的环形伪影修正的方法中图像处理的转换示意图；
图3(a)为本申请一种实施例的环形伪影修正的装置的结构示意图；图3(b)为本申请另一种实施例的环形伪影修正的装置的结构示意图；
图4为应用本申请实施的环形伪影修正的方法及装置进行的实验结果图。
具体实施方式
在探测器增益未标定良好与存在像素坏点的情况下，经过FDK算法重建的锥形束CT(CBCT)图像，会产生严重的环形伪影，这破坏了图像的细节信息，进一步地，这对应用CBCT进行放疗摆位和工业材料检测等操作带来误差。在不破坏原有图像信息的基础上，本申请能有效消除环形伪影。
下面通过具体实施方式结合附图对本申请作进一步详细说明。
请参照图1和图2，本申请公开了一种环形伪影修正的方法，该环形伪影修正的方法包括坐标转换步骤S01、处理步骤S03、相减步骤S05、提取步骤S07、判断步骤S09、迭代步骤S11和修正步骤S13，下面具体说明。
坐标转换步骤S01：将原始图像由笛卡尔坐标转换为极坐标，并作为处理步骤S03中的输入图像，进而执行处理步骤S03。在一具体实施例中，坐标转换步骤S01包括使用三次样条插值将将原始图像由笛卡尔坐标转换为极坐标。由于环形伪影的环形在极坐标下表现为条状，因此将原始图像由笛卡尔坐标转换为极坐标，这便于之后更好地提取出伪影信息。需要说明的是，坐标转换步骤S01中原始图像可以为原始CBCT图像。
处理步骤S03：对输入的图像进行消除图像细节和环形伪影的处理，得到去除环形伪影和图像细节的模版图像。需要说明的是，在初次迭代时，处理步骤S03中输入的是坐标转换步骤S01中的极坐标下的原始图像，在后续迭代过程中，处理步骤S03中输入的则是迭代步骤S11中的补偿图像。在一较优实施例中，处理步骤S03包括对所述输入图像进行相关总变分(RTV，Relative Total-Variance)平滑处理，以消除图像细节和环形伪影，得到去除环形伪影和图像细节的模版图像。在一较优实施 例中，处理步骤S03中的RTV平滑的目标函数如下：
Figure PCTCN2016108894-appb-000001
其中Ip为输入的图像，p为图像的像素索引，S为得到的模版图像，(Sp-Ip)2为平滑项，
Figure PCTCN2016108894-appb-000002
和
Figure PCTCN2016108894-appb-000003
为全变分窗口，像素索引q属于窗口R(p)，
Figure PCTCN2016108894-appb-000004
和
Figure PCTCN2016108894-appb-000005
分别为两个方向的偏导数，gp,q为空间关联度的加权函数。为了更好的捕捉图像的空间变化，利于结构和纹理的分离，本申请引入两个窗口
Figure PCTCN2016108894-appb-000006
和
Figure PCTCN2016108894-appb-000007
λ为图像平滑强度的加权，ε为一个极小的正数，主要是为了避免零的出现。由于目标函数是一个非凸函数，所以本申请可以采用二次惩罚的方法求解此目标函数。
相减步骤S05：将坐标转换步骤S01中的极坐标下的原始图像减去处理步骤S03中得到的模版图像，得到残差图像。由于残差图像是由原始图像减去不包括环形伪影和图像细节的模版图像得到，因此残差图像实际上就是包含图像细节和环形伪影的图像。
提取步骤S07：对残差图像进行提取，以得到环形伪影图像。由于受到探测器增益不一致等的影响，在极坐标下，条状伪影可以估计为一相同的像素值，因此可以在极坐标的角度方向上对残差图像进行提取，以得到环形伪影图像，所以在一较优实施例中，提取步骤S07包括对残差图像在极坐标的角度方向上进行提取，以得到环形伪影图像。在一具体实施例中，提取步骤S07包括对残差图像在极坐标的角度方向上进行中值提取，以得到环形伪影图像。另外，由于RTV平滑的误差和探测器坏点影响导致的坏点等原因，残差图像上可能表现出一些过高像素值和过低像素值的像素点，在进行提取的过程中，可以排除这些点，因此在一较优实施例中，提取步骤S07包括：在对残差图像在极坐标的角度方向上进行提取时，对残差图像中非过高值和非过低值的像素点进行提取，进一步地，可以是进行中值提取；具体地，对残差图像中非过高值和非过低值的像素点进行提取，可以设定一个像素阈值，将过高的像素值和过低的像素值的像素点排除，例如将像素点中其值低于10％和高于90％的排除，保留中间的80％作为提取的采样值。
判断步骤S09：判断当前是否达到停止迭代条件。在一较优实施例 中，判断步骤S09包括根据本次环形伪影图像与前一次环形伪影图像，判断是否停止迭代，这是一种迭代停止条件；和/或，根据当前的迭代次数，判断是否停止迭代，若当前的迭代次数达到了设定的迭代次数阈值，则判断停止迭代，反之，则判断继续迭代，这是另一种迭代停止条件；当判断步骤S09同时包括这两种迭代停止条件时，在一实施例中，可以是只要满足其中一个迭代停止条件，即判断停止迭代，当两个迭代停止条件都不满足时，则判断继续迭代。在一较优实施例中，判断步骤S09包括判断本次形伪影图像与前一次环形伪影图像之差的二范数，当该二范数小于一阈值时，则判断停止迭代，反之，则判断继续迭代。在一实施例中，判断步骤S09可以使用下述公式来作为判断二范数的条件：
Figure PCTCN2016108894-appb-000008
其中rk+1和rk分别为本次迭代的环形伪影图像和前一次迭代的环形伪影图像，r1为第一次迭代提取的环形伪影图像，||·||2为图像的二范数；在一实施例中，可以将阈值设置为0.002，当sd小于0.002时停止迭代，否则，继续迭代。
需要说明的是，在上述的一个实施例中，由于判断步骤S09是根据本次环形伪影图像与前一次环形伪影图像，判断是否停止迭代，因此当进行初次迭代时，本次环形伪影图像并不存在前一次环形伪影图像，这时在判断步骤S09也是判断需要继续迭代，具体实现时，可以将初次迭代时不存在的那个前一次环形伪影图像设置为空白图像，也可以根据该次迭代的次数为1，直接判断继续迭代。
迭代步骤S11：当判断继续迭代时，则根据本次残差图像和本次环形伪影图像来对本次的模版图像进行补偿，以得到补偿图像，并将该补偿图像作为处理步骤S03中的输入图像，进而执行处理步骤S03。在一具体实施例中，迭代步骤S11包括当判断继续迭代时，则将本次残差图像减去本次环形伪影图像，得到细节图像；将该细节图像补偿至本次的模版图像中，以得到补偿图像。在一实施例中，将细节图像补偿至本次的模版图像，可以是将细节图像与本次的模版图像相加，相加后得到的图像即为补偿图像。迭代步骤S11实质上是将补偿图像输出给处理步骤S03，使得补偿图像重新从处理步骤S03开始，再依次进行相减步骤S05、提取步骤S07、判断步骤S09，并在判断步骤S09中判断是否还要继续进行迭代，如此循环，直到某一次迭代中在判断步骤S09中判断为停止迭代，然后再进行修正步骤S13。迭代步骤S11每次都是将本次补偿图像输出给处理步骤S03，以进行迭代，这可以保证消除环形伪影的效果。
修正步骤S13：当判断停止迭代时，则使用本次环形伪影图像对原始图像进行修正，以获取环形伪影修正后的图像。在一具体实施例中， 修正步骤S13包括当判断停止迭代时，则将本次的环形伪影图像由极坐标转换为笛卡尔坐标，并将笛卡尔坐标下的原始图像减去笛卡尔坐标下的本次环形伪影图像，得到环形伪影修正后的图像。
请参照图3(a)和(b)，本申请还公开了一种环形伪影修正的装置，该环形伪影修正的装置包括坐标转换单元01、处理单元03、相减单元05、提取单元07、判断单元09、迭代单元11和修正单元13，下面具体说明。
坐标转换单元01用于将原始图像由笛卡尔坐标转换为极坐标,并作为处理单元03中的输入图像，输入给处理单元03进行处理。在一具体实施例中，坐标转换单元01包括插值单元01a，插值单元01a用于使用三次样条插值将将原始图像由笛卡尔坐标转换为极坐标。由于环形伪影的环形在极坐标下表现为条状，因此将原始图像由笛卡尔坐标转换为极坐标，这便于之后更好地提取出伪影信息。需要说明的是，坐标转换单元01中处理的原始图像可以为原始CBCT图像。
处理单元03用于对上述的输入图像进行消除图像细节和环形伪影的处理，得到去除环形伪影和图像细节的模版图像。需要说明的是，在初次迭代时，处理单元03中输入的是坐标转换单元01中的极坐标下的原始图像，在后续迭代过程中，处理单元03中输入的则是迭代单元11中的补偿图像。在一较优实施例中，处理单元03包括相关总变分单元03a，相关总变分单元03a对输入图像进行相关总变分(RTV，Relative Total-Variance)平滑处理，以消除图像细节和环形伪影，得到去除环形和图像细节的模版图像。在一较优实施例中，处理单元03中的RTV平滑的目标函数如下：
Figure PCTCN2016108894-appb-000009
其中Ip为输入的图像，p为图像的像素索引，S为得到的模版图像，(Sp-Ip)2为平滑项，
Figure PCTCN2016108894-appb-000010
和
Figure PCTCN2016108894-appb-000011
为全变分窗口，像素索引q属于窗口R(p)，
Figure PCTCN2016108894-appb-000012
和
Figure PCTCN2016108894-appb-000013
分别为两个方向的偏导数，gp,q为空间关联度的加权函数。为了更好的捕捉图像的空间变化，利于结构和纹理的分离，本申请引入两个窗口
Figure PCTCN2016108894-appb-000014
和
Figure PCTCN2016108894-appb-000015
λ为图像平滑强度的加权，ε为一个极小的正数，主要是为了避免零的出现。由于 目标函数是一个非凸函数，所以本申请可以采用二次惩罚的方法求解此目标函数。
相减单元05用于将极坐标下的原始图像减去模版图像，得到残差图像。由于残差图像是由原始图像减去不包括形伪影和图像细节的模版图像得到，因此残差图像实际上就是包含图像细节和环形伪影的图像。
提取单元07用于对残差图像进行提取，以得到环形伪影图像。由于受到探测器增益不一致等的影响，在极坐标下，条状伪影可以估计为一相同的像素值，因此可以在极坐标的角度方向上对残差图像进行提取，以得到环形伪影图像，所述在一较优实施例中，提取单元07包括方向提取单元07a，方向提取单元07a用于对残差图像在极坐标的角度方向上进行提取，以得到环形伪影图像。在一具体实施例中，方向提取单元07a包括中值提取单元07b，中值提取单元07b用于对残差图像在极坐标的角度方向上进行中值提取，以得到环形伪影图像。另外，由于RTV平滑的误差和探测器坏点影响导致的坏点等原因，残差图像上可能表现出一些过高像素值和过低像素值的像素点，在进行中值提取的过程中，可以排除这些点，因此在一较优实施例中，提取单元07包括选择提取单元07c，选择提取单元07c用于对残差图像中非过高值和非过低值的像素点进行提取，以得到环形伪影图像，较优地，选择提取单元07c可以是对残差图像在极坐标的角度方向上对残差图像中非过高值和非过低值的像素点进行提取，进一步地，选择提取单元07c可以是进行中值提取；选择提取单元07c在实现时，在对残差图像中非过高值和非过低值的像素点进行提取时，可以设定一个像素阈值，将过高的像素值和过低的像素值的像素点排除，例如将像素点中其值低于10％和高于90％的排除，保留中间的80％作为提取的采样值。
判断单元09判断当前是否达到停止迭代条件。在一较优实施例中，判断单元09包括比较单元09a和/或次数单元09b，具体地，比较单元09a用于根据本次环形伪影图像与前一次环形伪影图像，判断是否停止迭代；次数单元09b用于根据当前的迭代次数，判断是否停止迭代，若当前的迭代次数达到了设定的迭代次数阈值，则判断停止迭代，反之，则判断继续迭代；在一实施例中，当判断单元09同时包括比较单元09a和次数单元09b时，则可以是当比较单元09a和次数单元09b任一个判断当前要停止迭代，则判断单元09作出当前停止迭代的判断。在一较优实施例中，比较单元09a包括二范数比较单元09c，二范数比较单元09c用于判断本次形伪影图像与前一次环形伪影图像之差的二范数，当该二范数小于一阈值时，则判断停止迭代，反之，则判断继续迭代。在一实施例中，二范数比较单元09c可以使用下述公式来作为判断条件：
Figure PCTCN2016108894-appb-000016
其中rk+1和rk分别为本次迭代的环形伪影图像和前一次迭代的环形伪影图像，r1为第一次迭代提取的环形伪影图像，||·||2为图像的二范数；在一实施例中，可以将阈值设置为0.002，当sd小于0.002时停止迭代，否则，继续迭代。
需要说明的是，由于比较单元09a是根据本次环形伪影图像与前一次环形伪影图像，判断是否停止迭代，因此当进行初次迭代时，本次环形伪影图像并不存在前一次环形伪影图像，这时在比较单元09a也是判断需要继续迭代，具体实现时，可以将初次迭代时不存在的那个前一次环形伪影图像设置为空白图像，也可以根据该次迭代的次数为1，直接判断继续迭代。
迭代单元11用于当判断继续迭代时，则根据本次残差图像和本次环形伪影图像来对本次的模版图像进行补偿，以得到补偿图像，并将该补偿图像作为处理单元03中的输入图像，输入给处理单元03进行处理。在一具体实施例中，迭代单元11包括迭代子单元11a，迭代子单元11a用于当判断继续迭代时，则将本次残差图像减去本次环形伪影图像，得到细节图像；将该细节图像补偿至本次的模版图像中，以得到补偿图像。在一实施例中，迭代子单元11a将细节图像补偿至本次的模版图像，可以是将细节图像与本次的模版图像相加，相加后得到的图像即为补偿图像。迭代单元11实质上是将补偿图像输出给处理单元03，使得补偿图像重新从处理单元03开始，再依次由相减单元05、提取单元07和判断单元09进行工作，并在判断单元09中判断是否还要继续进行迭代，如此循环，直到某一次迭代中在判断单元09中判断为停止迭代，然后再进行修正单元13。迭代单元11每次都是将本次补偿图像输出给处理单元03，以进行迭代，这可以保证消除环形伪影的效果。
修正单元13用于当判断停止迭代时，则使用本次环形伪影图像对原始图像进行修正，以获取环形伪影修正后的图像。在一具体实施例中，修正单元13包括修正子单元13a，修正子单元13a用于当判断停止迭代时，则将本次的环形伪影图像由极坐标转换为笛卡尔坐标，并将笛卡尔坐标下的原始图像减去笛卡尔坐标下的本次环形伪影图像，得到环形伪影修正后的图像。
请参照图4，为应用本申请的环形伪影修正的方法及装置进行的实验结果图，第1至3列分别为不同层面的CBCT图像；第一行的图像为原始CBCT图像(即环形伪影修正前的图像)；第二行的图像为环形伪影修正后的图像，可以看到应用本申请后，环形伪影修正后的图像不仅被消除了环形伪影对图像造成的影响，还有效地保留了图像细节和图像 分辨率；第三行的图像为最终迭代提取的环形伪影图像，可以看到，提取的环形伪影图像很好，这样最终在修正步骤中将此提取的环形伪影图像对原始图像进行修正时，效果自然也很好。
本申请提供的环形伪影修正的方法及装置，利用极坐标转化、相关总变分处理、中值提取和迭代修正等实现迭代环形伪影修正，具体地，利用相关总变分对图像进行边缘保护的平滑，得到细节和伪影图像，应用于环形伪影修正；利用迭代修正的方法，不断提取环形伪影，达到图像环形伪影修正的结果。本申请在去除环形伪影过程中，做到了不损坏原始图像的细节和不降低图像的分辨率，并且处理流程简单等；另外本申请仅在图像域中对环形伪影进行提取，因此它的处理完全兼容临床使用的流程。为了更加切合临床应用，在一实施例中，还可以利用GPU加速大大降低环形伪影修正的时间。
以上应用了具体个例对本发明进行阐述，只是用于帮助理解本发明，并不用以限制本发明。对于本领域的一般技术人员，依据本发明的思想，可以对上述具体实施方式进行变化。

Claims (20)

  1. 一种环形伪影修正的方法，其特征在于，包括：
    坐标转换步骤：将原始图像由笛卡尔坐标转换为极坐标，并作为处理步骤中的输入图像，进而执行处理步骤；
    处理步骤：对所述输入图像进行消除图像细节和环形伪影的处理，得到去除环形伪影和图像细节的模版图像；
    相减步骤：将所述极坐标下的原始图像减去所述模版图像，得到残差图像；
    提取步骤：对所述残差图像进行提取，以得到环形伪影图像；
    判断步骤：判断当前是否达到停止迭代条件；
    迭代步骤：当判断继续迭代时，则根据本次残差图像和本次环形伪影图像来对本次的模版图像进行补偿，以得到补偿图像，并将该补偿图像作为处理步骤中的输入图像，进而执行处理步骤；
    修正步骤：当判断停止迭代时，则使用本次环形伪影图像对原始图像进行修正，以获取环形伪影修正后的图像。
  2. 如权利要求1所述的环形伪影修正的方法，其特征在于，所述坐标转换步骤包括：使用三次样条插值将原始图像由笛卡尔坐标转换为极坐标。
  3. 如权利要求1或2所述的环形伪影修正的方法，其特征在于，所述处理步骤包括：对所述输入图像进行相关总变分平滑处理，以消除图像细节和环形伪影，得到去除环形和图像细节的模版图像。
  4. 如权利标1所述的环形伪影修正的方法，其特征在于，所述提取步骤包括：对所述残差图像在极坐标的角度方向上进行提取，以得到环形伪影图像。
  5. 如权利要求4所述的环形伪影修正的方法，其特征在于，所述提取步骤包括：对所述残差图像在极坐标的角度方向上进行中值提取，以得到环形伪影图像。
  6. 如权利要求4或5所述的环形伪影修正的方法，其特征在于，提取步骤包括：对所述残差图像中非过高值和非过低值的像素点进行提取。
  7. 如权利标1所述的环形伪影修正的方法，其特征在于，所述判断步骤包括：
    根据本次环形伪影图像与前一次环形伪影图像，判断是否停止迭代；
    和/或，
    根据当前的迭代次数，判断是否停止迭代，若当前的迭代次数达到了设定的迭代次数阈值，则判断停止迭代，反之，则判断继续迭代。
  8. 如权利要求7所述的环形伪影修正的方法，其特征在于，所述判断步骤包括：判断本次形伪影图像与前一次环形伪影图像之差的二范数， 当该二范数小于一阈值时，则判断停止迭代。
  9. 如权利要求1所述的环形伪影修正的方法，其特征在于，所述迭代步骤包括：当判断继续迭代时，则将本次残差图像减去本次环形伪影图像，得到细节图像；将该细节图像补偿至本次的模版图像中，以得到补偿图像。
  10. 如权利要求1所述的环形伪影修正的方法，其特征在于，所述修正步骤包括：当判断停止迭代时，则将本次环形伪影图像由极坐标转换为笛卡尔坐标，并将笛卡尔坐标下的原始图像减去笛卡尔坐标下的本次环形伪影图像，得到环形伪影修正后的图像。
  11. 一种环形伪影修正的装置，其特征在于，包括：
    坐标转换单元，用于将原始图像由笛卡尔坐标转换为极坐标，并作为处理单元中的输入图像，输入给处理单元进行处理；
    处理单元，用于对所述输入图像进行消除图像细节和环形伪影的处理，得到去除环形伪影和图像细节的模版图像；
    相减单元，用于将所述极坐标下的原始图像减去所述模版图像，得到残差图像；
    提取单元，用于对所述残差图像进行提取，以得到环形伪影图像；
    判断单元，用于判断是否达到停止迭代条件；
    迭代单元，用于当判断继续迭代时，则根据本次残差图像和本次环形伪影图像来对本次的模版图像进行补偿，以得到补偿图像，并将该补偿图像作为处理单元中的输入图像，输入给处理单元进行处理；
    修正单元，用于当判断停止迭代时，则使用本次的环形伪影图像对原始图像进行修正，以获取环形伪影修正后的图像。
  12. 如权利要求11所述的环形伪影修正的装置，其特征在于，所述坐标转换单元包括插值单元，用于使用三次样条插值将将原始图像由笛卡尔坐标转换为极坐标。
  13. 如权利要求11或12所述的环形伪影修正的装置，其特征在于，所述处理单元包括相关总变分单元，用于对所述输入图像进行相关总变分平滑处理，以消除图像细节和环形伪影，得到去除环形和图像细节的模版图像。
  14. 如权利要求11所述的环形伪影修正的装置，其特征在于，所述提取单元包括方向提取单元，用于对所述残差图像在极坐标的角度方向上进行提取，以得到环形伪影图像。
  15. 如权利要求14所述的环形伪影修正的装置，其特征在于，所述方向提取单元包括中值提取单元，用于对所述残差图像在极坐标的角度方向上进行中值提取，以得到环形伪影图像。
  16. 如权利要求14或15所述的环形伪影修正的装置，其特征在于， 所述提取单元包括选择提取单元，用于对所述残差图像中非过高值和非过低值的像素点进行提取。
  17. 如权利标11所述的环形伪影修正的装置，其特征在于，所述判断单元包括：
    比较单元，用于根据本次环形伪影图像与前一次环形伪影图像，判断是否停止迭代；
    和/或，
    次数单元，用于根据当前的迭代次数，判断是否停止迭代，若当前的迭代次数达到了设定的迭代次数阈值，则判断停止迭代。
  18. 如权利要求17所述的环形伪影修正的装置，其特征在于，所述比较单元包括二范数比较单元，用于判断本次形伪影图像与前一次环形伪影图像之差的二范数，当该二范数小于一阈值时，则判断停止迭代。
  19. 如权利要求11所述的环形伪影修正的装置，其特征在于，所述迭代单元包括迭代子单元，用于当判断继续迭代时，将本次残差图像减去本次环形伪影图像，得到细节图像；将该细节图像补偿至本次的模版图像中，以得到补偿图像。
  20. 如权利要求11所述的环形伪影修正的装置，其特征在于，所述修正单元包括修正子单元，用于当判断停止迭代时，将本次的环形伪影图像由极坐标转换为笛卡尔坐标，并将笛卡尔坐标下的原始图像减去笛卡尔坐标下的本次环形伪影图像，得到环形伪影修正后的图像。
PCT/CN2016/108894 2016-12-07 2016-12-07 一种环形伪影修正的方法及装置 WO2018103015A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2016/108894 WO2018103015A1 (zh) 2016-12-07 2016-12-07 一种环形伪影修正的方法及装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2016/108894 WO2018103015A1 (zh) 2016-12-07 2016-12-07 一种环形伪影修正的方法及装置

Publications (1)

Publication Number Publication Date
WO2018103015A1 true WO2018103015A1 (zh) 2018-06-14

Family

ID=62490609

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/108894 WO2018103015A1 (zh) 2016-12-07 2016-12-07 一种环形伪影修正的方法及装置

Country Status (1)

Country Link
WO (1) WO2018103015A1 (zh)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109636872A (zh) * 2018-12-10 2019-04-16 合肥中科离子医学技术装备有限公司 基于滑动窗口差值和条带噪声检测的cbct环形伪影消除方法
CN110400358A (zh) * 2019-07-17 2019-11-01 郑州信大先进技术研究院 一种平板探测器环状伪影去除方法
CN110889878A (zh) * 2019-11-08 2020-03-17 湖北科技学院 结合图像平滑处理的ct图像环形伪影抑制方法
US11544821B2 (en) * 2019-04-30 2023-01-03 Shanghai United Imaging Healthcare Co., Ltd. Systems and methods for determining ring artifact
CN117078791A (zh) * 2023-10-13 2023-11-17 俐玛精密测量技术(苏州)有限公司 一种ct环形伪影校正方法、装置、电子设备及存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102521801A (zh) * 2011-11-23 2012-06-27 中国科学院深圳先进技术研究院 Ct图像的环形伪影及弧形伪影校正方法
US20120195483A1 (en) * 2011-01-27 2012-08-02 Siemens Aktiengesellschaft Method And Computer System For The Reduction Of Artifacts In Reconstructed CT Image Datasets
CN102800073A (zh) * 2012-06-28 2012-11-28 西北工业大学 一种锥束ct环形伪影的自动判别与校正方法
CN103593826A (zh) * 2013-10-22 2014-02-19 南方医科大学 图像环形伪影校正方法
CN105321155A (zh) * 2015-10-29 2016-02-10 北京理工大学 一种cbct图像环形伪影消除方法
CN105719245A (zh) * 2016-01-12 2016-06-29 南方医科大学 一种利用投影数据去除ct探测元故障造成的环状伪影的方法

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120195483A1 (en) * 2011-01-27 2012-08-02 Siemens Aktiengesellschaft Method And Computer System For The Reduction Of Artifacts In Reconstructed CT Image Datasets
CN102521801A (zh) * 2011-11-23 2012-06-27 中国科学院深圳先进技术研究院 Ct图像的环形伪影及弧形伪影校正方法
CN102800073A (zh) * 2012-06-28 2012-11-28 西北工业大学 一种锥束ct环形伪影的自动判别与校正方法
CN103593826A (zh) * 2013-10-22 2014-02-19 南方医科大学 图像环形伪影校正方法
CN105321155A (zh) * 2015-10-29 2016-02-10 北京理工大学 一种cbct图像环形伪影消除方法
CN105719245A (zh) * 2016-01-12 2016-06-29 南方医科大学 一种利用投影数据去除ct探测元故障造成的环状伪影的方法

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109636872A (zh) * 2018-12-10 2019-04-16 合肥中科离子医学技术装备有限公司 基于滑动窗口差值和条带噪声检测的cbct环形伪影消除方法
CN109636872B (zh) * 2018-12-10 2022-12-09 合肥中科离子医学技术装备有限公司 基于滑动窗口差值和条带噪声检测的cbct环形伪影消除方法
US11544821B2 (en) * 2019-04-30 2023-01-03 Shanghai United Imaging Healthcare Co., Ltd. Systems and methods for determining ring artifact
CN110400358A (zh) * 2019-07-17 2019-11-01 郑州信大先进技术研究院 一种平板探测器环状伪影去除方法
CN110400358B (zh) * 2019-07-17 2022-10-04 郑州信大先进技术研究院 一种平板探测器环状伪影去除方法
CN110889878A (zh) * 2019-11-08 2020-03-17 湖北科技学院 结合图像平滑处理的ct图像环形伪影抑制方法
CN117078791A (zh) * 2023-10-13 2023-11-17 俐玛精密测量技术(苏州)有限公司 一种ct环形伪影校正方法、装置、电子设备及存储介质
CN117078791B (zh) * 2023-10-13 2024-01-12 俐玛精密测量技术(苏州)有限公司 一种ct环形伪影校正方法、装置、电子设备及存储介质

Similar Documents

Publication Publication Date Title
WO2018103015A1 (zh) 一种环形伪影修正的方法及装置
JP5645399B2 (ja) Ｘ線画像処理装置、ｘ線画像処理方法、及びコンピュータプログラム
EP2293239B1 (en) Line noise reduction apparatus and method
CN104200442B (zh) 基于改进的canny边缘检测的非局部均值MRI图像去噪方法
JP3995854B2 (ja) 画像処理方法および装置並びに記録媒体
US9111174B2 (en) Machine learnng techniques for pectoral muscle equalization and segmentation in digital mammograms
CN106651981B (zh) 一种环形伪影修正的方法及装置
CN105321155A (zh) 一种cbct图像环形伪影消除方法
CN102737382B (zh) 一种前列腺超声图像自动精确分割方法
KR20130038794A (ko) 디지털 엑스레이 프레임들 시리즈에서의 잡음 감소의 방법
CN106530236B (zh) 一种医学图像处理方法及系统
US8494239B2 (en) Image processing device, method and program
WO2020062843A1 (zh) 药丸包衣图像处理方法、装置、计算机设备和存储介质
JP6071444B2 (ja) 画像処理装置及びその作動方法、プログラム
CN116630762B (zh) 一种基于深度学习的多模态医学图像融合方法
WO2019220417A1 (en) Method and system for automatically generating and analyzing fully quantitative pixel-wise myocardial blood flow and myocardial perfusion reserve maps to detect ischemic heart disease using cardiac perfusion magnetic resonance imaging
CN116342444B (zh) 一种双通道多模态图像融合方法及电子设备
WO2018068195A1 (zh) 一种基于图像梯度矢量流场的血管脊点提取方法及装置
US8989462B2 (en) Systems, methods and computer readable storage mediums storing instructions for applying multiscale bilateral filtering to magnetic resonance (RI) images
CN107292847B (zh) 一种数据降噪方法及系统
Desiani et al. A robust techniques of enhancement and segmentation blood vessels in retinal image using deep learning
Rani A novel mathematical morphology based edge detection method for medical images
CN108734674B (zh) 一种改进nas-rif的oct图像盲复原方法
CN110335287A (zh) 建筑图像数据的提取方法及装置
CN115222700A (zh) 一种多尺度retinex和FASTICA结合的红外热像处理方法

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: 16923467

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

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 07/10/2019)

122 Ep: pct application non-entry in european phase

Ref document number: 16923467

Country of ref document: EP

Kind code of ref document: A1