CT (Computed Tomography) image auxiliary reading method in post-chemotherapy reexamination stage

Abstract

The invention relates to the technical field of image enhancement, in particular to a CT (Computed Tomography) image auxiliary reading method for a review stage after a chemotherapy operation, which comprises the following steps of: acquiring CT images before and after the chemotherapy operation; performing region division on the first to-be-enhanced image and the second to-be-enhanced image; determining the matching degree between each first region and each second region; performing region matching screening according to the matching degree between the first region and the second region; determining a tumor expression degree and a blood vessel expression degree corresponding to each region in each target matching pair; according to the difference between the tumor representation degrees and the difference between the blood vessel representation degrees corresponding to the two areas in each target matching pair, determining the real recovery representation degree under each target matching pair; and performing histogram equalization on the first to-be-enhanced image and the second to-be-enhanced image according to the real recovery performance under all the target matching pairs. According to the invention, the enhancement of the CT image is realized, and the enhancement effect of the CT image is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image enhancement, and in particular to a CT image-assisted reading method in a post-chemotherapy follow-up phase. Background Art

[0002] With the development of science and technology, the application of image enhancement technology is becoming more and more extensive. For example, it can be used to enhance CT images required for the follow-up stage after chemotherapy to assist doctors in reading the images. At present, when performing image enhancement, the commonly used method is: according to the grayscale histogram of the image, the image is histogram equalized to achieve image enhancement. Among them, histogram equalization is also called grayscale histogram equalization.

[0003] However, when histogram equalization is performed on a CT image based on its grayscale histogram to achieve image enhancement, the following technical problems often occur: Since grayscale histogram equalization often performs statistical image enhancement according to the grayscale value distribution of the image, when the histogram equalization is performed on the CT image directly based on the grayscale histogram of the CT image, detail information with fewer pixels may be lost, resulting in poor enhancement effect on the CT image. Summary of the invention

[0004] In order to solve the technical problem of poor enhancement effect of CT images, the present invention proposes a CT image-assisted reading method in the post-chemotherapy review stage.

[0005] In a first aspect, the present invention provides a method for assisting CT image reading in a post-chemotherapy review phase, the method comprising: Acquire CT images before and after chemotherapy as the first image to be enhanced and the second image to be enhanced, respectively; Performing region division on the first image to be enhanced to obtain a first region, and performing region division on the second image to be enhanced to obtain a second region; Determining a matching degree between each first region and each second region according to a distance and a grayscale distribution difference between each first region and each second region; Performing regional matching screening according to the matching degree between the first region and the second region to obtain a target matching pair representing changes before and after the operation; According to the grayscale distribution of the pre-extracted connected domain in each region in each target matching pair, the tumor expression corresponding to each region in each target matching pair is obtained; Determine the blood vessel expression corresponding to each region in each target matching pair according to the characteristics of the pre-extracted skeleton in the connected domain in each region in each target matching pair; According to the difference between the tumor expressions corresponding to the two regions in each target matching pair and the difference between the corresponding blood vessel expressions, the true restored expression under each target matching pair is determined; According to the true restoration performance under all target matching pairs, histogram equalization is performed on the first image to be enhanced and the second image to be enhanced.

[0006] In combination with the first aspect, in a possible implementation manner, determining the matching degree between each first region and each second region according to the distance and grayscale distribution difference between each first region and each second region includes: Determine any one of the first regions as a marking region, and determine any one of the second regions as a reference region; Determine the region in the second image to be enhanced that is at the same position as the marked region as a position representative region; Determine the distance between the center point of the reference area and the center point of the position representative area as the target distance between the marked area and the reference area; Determine the absolute value of the difference between the grayscale values ​​corresponding to the pixels at the same position in the reference area and in the position representative area as the target grayscale difference, and obtain a target grayscale difference set between the reference area and the position representative area; Determine a target grayscale distribution difference between the reference area and the position representative area according to a target grayscale difference set between the reference area and the position representative area, wherein the target grayscale difference in the target grayscale difference set is positively correlated with the target grayscale distribution difference; The matching degree between the reference area and the position representative area is determined according to the target grayscale distribution difference and the target distance between the reference area and the position representative area, wherein the target grayscale distribution difference and the target distance are negatively correlated with the matching degree.

[0007] In combination with the first aspect, in a possible implementation manner, determining the target grayscale distribution difference between the reference area and the position representative area according to the target grayscale difference set between the reference area and the position representative area includes: The mean value of all target grayscale differences in the target grayscale difference set between the reference region and the position representative region is determined as the target grayscale distribution difference between the reference region and the position representative region.

[0008] In combination with the first aspect above, in a possible implementation, performing region matching screening according to the matching degree between the first region and the second region to obtain a target matching pair representing changes before and after the operation includes: Determine any one of the first regions as a marked region, and select a second region having the same position as the marked region from all the second regions as a matching reference region corresponding to the marked region; The marked area and its corresponding matching reference area constitute a reference matching pair corresponding to the marked area, and the matching degree between the two areas in the reference matching pair is determined as a target matching index of the reference matching pair; A reference matching pair whose target matching index is less than a preset matching change threshold is selected from all reference matching pairs corresponding to the first region as a target matching pair.

[0009] In combination with the first aspect above, in a possible implementation, obtaining the tumor expression corresponding to each region in each target matching pair according to the grayscale distribution of the pre-extracted connected domain in each region in each target matching pair includes: Determine any region in any target matching pair as a candidate region, and perform connected domain extraction on the candidate region; Determine any connected domain in the candidate region as a marked connected domain, and determine the mean of the grayscale values ​​corresponding to all pixels in the marked connected domain as a grayscale representative factor corresponding to the marked connected domain; Determine the chain code corresponding to each pixel point on the edge of the marked connected domain by using the eight-chain code algorithm to obtain the chain code sequence corresponding to the marked connected domain; The difference between each two adjacent chain codes in the chain code sequence corresponding to the marked connected domain is determined as a target difference, so as to obtain a target difference sequence corresponding to the marked connected domain; Determine the variance of all target difference values ​​in the target difference value sequence corresponding to the marked connected domain as the edge clutter corresponding to the marked connected domain; Normalizing the product of the grayscale representative factor and the edge clutter corresponding to the marked connected domain to obtain a suspected tumor indicator corresponding to the marked connected domain; The average of the suspected tumor indicators corresponding to all connected domains in the candidate region is determined as the tumor expression degree corresponding to the candidate region.

[0010] In combination with the first aspect above, in a possible implementation, determining the blood vessel expression corresponding to each region in each target matching pair according to the characteristics of the pre-extracted skeleton in the connected domain in each region in each target matching pair includes: Determine any region in any target matching pair as a candidate region, and perform skeleton extraction on all connected domains in the candidate region; Taking the intersection points of all skeletons in the candidate area as segmentation points, segmenting all skeletons in the candidate area to obtain skeleton branches; Using the normals at the two endpoints of each skeleton branch as cutting lines, the connected domain of each skeleton branch is segmented to obtain the sub-region corresponding to each skeleton branch; All intersection points of the normal line of each pixel point on each skeleton branch and the edge of its corresponding sub-region form a set of intersection points corresponding to each pixel point on each skeleton branch; The average value of the distances between all intersection points in the set of intersection points corresponding to each pixel point on each skeleton branch is determined as the representative distance factor corresponding to each pixel point on each skeleton branch; Determine the extensibility factor corresponding to each skeleton branch according to the grayscale values ​​and representative distance factors corresponding to all pixels on each skeleton branch; Perform threshold segmentation on each region in each target matching pair to obtain the foreground region and background region corresponding to each region in each target matching pair; The vascular expression degree corresponding to each region in each target matching pair is determined according to the grayscale difference between the foreground region and the background region corresponding to each region in each target matching pair, and the extensibility factor corresponding to all skeleton branches in each region in each target matching pair.

[0011] In combination with the first aspect above, in a possible implementation, determining the extensibility factor corresponding to each skeleton branch according to the grayscale values ​​corresponding to all pixels on each skeleton branch and the representative distance factor includes: The product of the grayscale value corresponding to each pixel point on each skeleton branch and the representative distance factor is determined as the width grayscale factor corresponding to each pixel point on each skeleton branch; The difference between the width grayscale factors corresponding to each adjacent pixel point on each skeleton branch is determined as the width grayscale difference, and a width grayscale difference set corresponding to each skeleton branch is obtained; The absolute value of the accumulated values ​​of all the width grayscale differences in the width grayscale difference set corresponding to each skeleton branch is determined as the width grayscale representative difference corresponding to each skeleton branch; The variance of all width grayscale differences in the width grayscale difference set corresponding to each skeleton branch is determined as the width grayscale variation factor corresponding to each skeleton branch; According to the width grayscale representative difference and the width grayscale variation factor corresponding to each skeleton branch, the extensibility factor corresponding to each skeleton branch is determined, wherein the width grayscale representative difference is positively correlated with the extensibility factor, and the width grayscale variation factor is negatively correlated with the extensibility factor.

[0012] In combination with the first aspect above, in a possible implementation, the formula corresponding to the blood vessel expression corresponding to the region in the target matching pair is: ; ;in, It is Target matching pair The vascular expression corresponding to each region; is the sequence number of the target matching pair; It is The serial number of the target matching area; is the normalization function; It is Target matching pair The cumulative value of the extensibility factors corresponding to all skeleton branches in a region; Characterization Target matching pair Disorder in the region; It is an exponential function with a natural constant as base; It is the absolute value function; It is Target matching pair The mean of the gray values ​​corresponding to all pixels in the foreground area of ​​the region; It is Target matching pair The mean of the grayscale values ​​corresponding to all pixels in the background area of ​​a region; It is Target matching pair The number of skeleton branches in a region; It is Target matching pair The serial number of the skeleton branches in each region; It is Target matching pair In the region The variance of the chain codes corresponding to all pixels on the skeleton branches; the chain codes corresponding to the pixels are obtained through the eight-chain code algorithm.

[0013] In combination with the first aspect above, in a possible implementation, the formula corresponding to the true recovery performance under the target matching pair is: ; ; ;in, It is The true recovery performance under target matching pairs; is the sequence number of the target matching pair; is the normalization function; Indicates Tumor elimination under target matching pairs; Indicates Local recovery under target matching pairs; It is The tumor expression corresponding to the first region in the target matching pair; It is The tumor expression corresponding to the second region in the target matching pair; It is The vascular expression corresponding to the second region in the target matching pair; It is The vascular expression corresponding to the first region in the target matching pair.

[0014] In combination with the first aspect, in a possible implementation manner, performing histogram equalization on the first image to be enhanced and the second image to be enhanced according to the true restoration representation under all target matching pairs includes: Determine the true recovery performance degree of each target matching pair to which each first region belongs as a true recovery performance factor corresponding to each first region; Adjusting the grayscale histogram of the first image to be enhanced according to the true restoration performance factors corresponding to all the first regions to obtain a first target histogram; Determine the true recovery performance degree under the target matching pair to which each second region belongs as the true recovery performance factor corresponding to each second region; According to the true restoration performance factors corresponding to all the second regions, the grayscale histogram of the second image to be enhanced is adjusted to obtain a second target histogram; Histogram equalization is performed on the first image to be enhanced and the second image to be enhanced according to the first target histogram and the second target histogram.

[0015] In a second aspect, the present invention provides a CT image-assisted reading system for the post-chemotherapy review stage, the system comprising: A CT image acquisition module is used to acquire CT images before and after chemotherapy as the first image to be enhanced and the second image to be enhanced respectively; A region division module, used to perform region division on the first image to be enhanced to obtain a first region, and to perform region division on the second image to be enhanced to obtain a second region; A matching degree determination module, used to determine the matching degree between each first region and each second region according to the distance and grayscale distribution difference between each first region and each second region; A region matching screening module, used to perform region matching screening according to the matching degree between the first region and the second region, and obtain a target matching pair representing changes before and after the operation; A tumor expression determination module is used to obtain the tumor expression corresponding to each region in each target matching pair according to the gray distribution of the pre-extracted connected domain in each region in each target matching pair; A blood vessel expression determination module, used to determine the blood vessel expression corresponding to each region in each target matching pair according to the characteristics of the pre-extracted skeleton in the connected domain in each region in each target matching pair; A true restoration expression determination module, used to determine the true restoration expression of each target matching pair according to the difference between the tumor expressions corresponding to the two regions in each target matching pair and the difference between the corresponding blood vessel expressions; The histogram equalization module is used to perform histogram equalization on the first image to be enhanced and the second image to be enhanced according to the true restoration representation under all target matching pairs.

[0016] In a third aspect, a server is provided, comprising a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, so that the device executes the method in the first aspect or any possible implementation of the first aspect.

[0017] In a fourth aspect, a computer program product is provided, comprising: a computer program code, which, when executed on a computer, enables the computer to execute the method in the first aspect or any possible implementation of the first aspect.

[0018] In a fifth aspect, a computer-readable storage medium is provided, which stores a computer program code. When the computer program code runs on a computer, the computer executes the method in the above-mentioned first aspect or any possible implementation manner of the first aspect.

[0019] The present invention has the following beneficial effects: The present invention provides a CT image-assisted reading method for the post-chemotherapy review stage, which realizes the enhancement of CT images and solves the technical problem of poor enhancement effect of CT images, thereby improving the enhancement effect of CT images. Compared with directly performing histogram equalization on CT images according to the grayscale histogram of CT images, the present invention comprehensively considers multiple indicators related to the detailed information of disease recovery of CT images, such as the matching degree between regions in different CT images, tumor expression degree, blood vessel expression degree and true recovery expression degree, thereby improving the retention degree of detailed information and further improving the effect of CT image enhancement. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 This is a flow chart of a CT image-assisted reading method in the post-chemotherapy review stage of the present invention; Figure 2 is a schematic diagram of an area at the same position in a first image to be enhanced and a second image to be enhanced according to the present invention; Figure 3 A schematic diagram of the structure of a CT image-assisted reading system for the post-chemotherapy review stage of the present invention; Figure 4 The figure is a schematic diagram of the structure of a computer device of the present invention.

[0022] The reference numerals include: a first image to be enhanced 201 , a second image to be enhanced 202 , a marked area 203 and a square area 204 . DETAILED DESCRIPTION

[0023] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation methods, structures, features and effects of the technical solutions proposed by the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0024] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0025] refer to Figure 1 , shows the process of some embodiments of a CT image-assisted reading method in the post-chemotherapy review stage according to the present invention. The CT image-assisted reading method in the post-chemotherapy review stage includes the following steps: Step S1, obtaining CT images before and after chemotherapy as the first image to be enhanced and the second image to be enhanced, respectively.

[0026] Among them, chemotherapy, also often referred to as chemotherapeutic treatment, is a treatment strategy for killing tumor cells, which works by inhibiting tumor growth or eliminating tumors through systemic or partial use of chemical drugs. For example, chemotherapy can be used to treat lung cancer. CT images can be images collected by a CT (Computed Tomography) machine. In the embodiment of the present invention, the CT images before chemotherapy and the CT images after chemotherapy can be taken of the same patient parts.

[0027] It should be noted that CT technology is widely used in the medical and industrial fields, mainly for diagnostic imaging, guided treatment, surgical planning and navigation, etc. In the review stage of chemotherapy, CT scans can help doctors evaluate whether chemotherapy is effective. Specifically, they can be used to compare the size, morphological changes and density changes of tumors before and after treatment. During the doctor's reading process, since chemotherapy may cause some tumor areas to become smaller, it may be difficult for doctors to find them, which may lead to poor efficiency in doctor's reading. Therefore, the embodiment of the present invention enhances the clarity of the changed lesion area to a certain extent by performing image enhancement on the CT images before and after chemotherapy, thereby improving the image enhancement effect.

[0028] As an example, taking chemotherapy for the treatment of lung cancer as an example, a CT machine can be used to collect lung CT images of the patient before the chemotherapy surgery as the first image to be enhanced, and a CT machine can be used to collect lung CT images of the patient after the chemotherapy surgery as the second image to be enhanced.

[0029] Step S2, dividing the first image to be enhanced into regions to obtain a first region, and dividing the second image to be enhanced into regions to obtain a second region.

[0030] The first area and the second area may have the same shape and size.

[0031] As an example, the first image to be enhanced may be divided into equal parts, and each region obtained by the equal division is used as the first region; similarly, the second image to be enhanced may be divided into equal parts in the same manner, and each region obtained by the equal division is used as the second region. The size of the first region and the size of the second region may both be 21×21.

[0032] Step S3, determining the matching degree between each first region and each second region according to the distance and grayscale distribution difference between each first region and each second region.

[0033] As an example, this step may include the following steps: In the first step, any first region is determined as a marking region, and any second region is determined as a reference region.

[0034] In the second step, the region in the second image to be enhanced that is at the same position as the marked region is determined as a position representative region.

[0035] For example, Figure 2 As shown, the marked area 203 is any first area in the first image to be enhanced 201; the square area 204 can be an area in the second image to be enhanced 202 with the same position as the marked area 203, that is, a position representative area.

[0036] In the third step, the distance between the center point of the reference area and the center point of the position representative area is determined as the target distance between the marked area and the reference area.

[0037] In the fourth step, the absolute value of the difference between the grayscale values ​​corresponding to the pixel points at the same position in the reference area and the position representative area is determined as the target grayscale difference, and a target grayscale difference set between the reference area and the position representative area is obtained.

[0038] The fifth step is to determine the target grayscale distribution difference between the reference area and the position representative area according to the target grayscale difference set between the reference area and the position representative area.

[0039] Among them, the target grayscale difference in the target grayscale difference set may be positively correlated with the target grayscale distribution difference.

[0040] For example, the mean value of all target grayscale differences in the target grayscale difference set between the reference region and the position representative region may be determined as the target grayscale distribution difference between the reference region and the position representative region.

[0041] In the sixth step, the matching degree between the reference area and the position representative area is determined according to the target grayscale distribution difference and the target distance between the reference area and the position representative area.

[0042] Among them, the target grayscale distribution difference and the target distance can both be negatively correlated with the matching degree.

[0043] For example, the formula for determining the matching degree between the first region and the second region may be: ; ;in, It is The first area and the The matching degree between the second regions. Is the sequence number of the first region. Is the sequence number of the second region. is a normalization function. It is an exponential function with a natural constant as its base. It is The first area and the The target grayscale distribution difference between the second regions. It is The first area and the The target distance between the second area. It is the number of pixels in the first area or the number of pixels in the second area, and the number of pixels in the first area is the same as the number of pixels in the second area. is the serial number of the pixel point in the first area or the serial number of the pixel point in the second area, and the first area pixel point and the The pixels may be pixels at the same position in different regions. It is the absolute value function. It is The first area The gray value corresponding to each pixel. It is Second area The gray value corresponding to each pixel. is the target grayscale difference.

[0044] It should be noted that when The smaller the time, the more likely it is that The grayscale distribution in the first area is The more similar the grayscale distribution in the second area is, the more likely it is that the The first area and the The smaller the difference between the second areas. The smaller the time, the more likely it is that The first area and the The smaller the distance between the second areas, the smaller the The first area and the The more likely the second region is to represent the same part. The larger the The first area and the The more similar the second regions are, the more likely it is that the The first area and the The closer the match is between the second regions.

[0045] Step S4, performing region matching screening according to the matching degree between the first region and the second region, and obtaining a target matching pair representing changes before and after the operation.

[0046] Each target matching pair may include a first region and a second region. The target matching pair may represent a disease region that has changed before and after chemotherapy.

[0047] As an example, this step may include the following steps: In the first step, any first region is determined as a marked region, and a second region having the same position as the marked region is selected from all second regions as a matching reference region corresponding to the marked region.

[0048] It should be noted that, since the same part is photographed before and after the operation, the matching reference area corresponding to the marked area is often the area in the second image to be enhanced that has the same position as the marked area.

[0049] In the second step, the marked area and its corresponding matching reference area constitute a reference matching pair corresponding to the marked area, and the matching degree between the two areas in the reference matching pair is determined as the target matching index of the reference matching pair.

[0050] In the third step, a reference matching pair whose target matching index is less than a preset matching change threshold is selected from all reference matching pairs corresponding to the first region as a target matching pair.

[0051] The preset matching change threshold may be a preset threshold for screening disease areas. For example, the preset matching change threshold may be 0.3.

[0052] It should be noted that the tumor area with lesions often undergoes certain changes before and after chemotherapy, that is, the degree of lesions in the tumor area often decreases before and after chemotherapy. Therefore, there are often certain changes between the tumor areas before and after chemotherapy. The area in the target matching pair can often represent the area with lesions.

[0053] Step S5, obtaining the tumor expression corresponding to each region in each target matching pair according to the grayscale distribution of the pre-extracted connected domain in each region in each target matching pair.

[0054] As an example, this step may include the following steps: In the first step, any region in any target matching pair is determined as a candidate region, and a connected domain extraction algorithm is used to extract the connected domains of the candidate region to obtain multiple connected domains.

[0055] It should be noted that, for areas in the first image to be enhanced and the second image to be enhanced where there are no connected domains, and areas at the same position as areas where there are no connected domains, there is no need to subsequently calculate indicators of tumor manifestation and vascular manifestation in these areas, and there is no need to subsequently adjust the number of pixels in the grayscale histogram corresponding to the grayscale values ​​in these areas, and histogram equalization can be directly performed using the actual number of pixels in the grayscale histogram corresponding to the grayscale values ​​in these areas.

[0056] In the second step, any connected domain in the candidate region is determined as a marked connected domain, and the mean of the grayscale values ​​corresponding to all pixels in the marked connected domain is determined as the grayscale representative factor corresponding to the marked connected domain.

[0057] It should be noted that in actual situations, the tumor area in the CT image is often relatively bright. When the grayscale representative factor corresponding to the labeled connected domain is larger, it often means that the labeled connected domain is relatively brighter, which often means that the labeled connected domain is more likely to be a tumor connected domain.

[0058] The third step is to determine the chain code corresponding to each pixel on the edge of the marked connected domain through the eight-chain code algorithm, and obtain the chain code sequence corresponding to the marked connected domain.

[0059] The chain code sequence corresponding to the marked connected domain may represent the chain code situation of the edge of the marked connected domain. The chain code sequence corresponding to the marked connected domain may include: chain codes corresponding to all pixel points on the edge of the marked connected domain.

[0060] In the fourth step, the difference between each two adjacent chain codes in the chain code sequence corresponding to the above-mentioned marked connected domain is determined as the target difference, and the target difference sequence corresponding to the above-mentioned marked connected domain is obtained.

[0061] Among them, each two adjacent chain codes in the chain code sequence corresponding to the marked connected domain can represent the chain code corresponding to each two adjacent pixel points on the edge of the marked connected domain. The difference between the two adjacent chain codes can be equal to the previous chain code minus the next chain code of the two chain codes. The target difference sequence corresponding to the marked connected domain may include: the difference between all two adjacent chain codes in the chain code sequence corresponding to the marked connected domain.

[0062] In the fifth step, the variance of all target difference values ​​in the target difference value sequence corresponding to the above-mentioned marked connected domain is determined as the edge clutter corresponding to the above-mentioned marked connected domain.

[0063] It should be noted that in actual situations, the shape of the tumor region is often irregular, so the edge of the tumor region is often messy. Therefore, when the edge clutter corresponding to the labeled connected domain is greater, it often means that the edge of the labeled connected domain is more messy, and it often means that the labeled connected domain is more likely to be a tumor connected domain.

[0064] In the sixth step, the product of the grayscale representative factor and the edge clutter corresponding to the above-mentioned marked connected domain is normalized to obtain the suspected tumor indicator corresponding to the above-mentioned marked connected domain.

[0065] It should be noted that when the suspected tumor index corresponding to the marked connected domain is larger, it often means that the marked connected domain is more likely to be a tumor connected domain.

[0066] In the seventh step, the mean value of the suspected tumor indicators corresponding to all connected domains in the candidate region is determined as the tumor expression degree corresponding to the candidate region.

[0067] It should be noted that when the tumor expression corresponding to the candidate region is greater, it often means that the candidate region is more likely to be a real tumor region.

[0068] Step S6, determining the blood vessel expression corresponding to each region in each target matching pair according to the characteristics of the pre-extracted skeleton in the connected domain in each region in each target matching pair.

[0069] As an example, this step may include the following steps: In the first step, any region in any target matching pair is determined as a candidate region, and the skeleton of all connected domains in the above candidate region is extracted through the OpenCV skeleton extraction algorithm to obtain multiple skeletons.

[0070] In the second step, all skeletons in the candidate area are segmented using the intersection points of all skeletons in the candidate area as segmentation points to obtain skeleton branches.

[0071] The intersection point may be an intersection point of at least two lines.

[0072] The third step is to use the normals at the two endpoints of each skeleton branch as the cutting line to segment the connected domain to which each skeleton branch belongs, and obtain the sub-region corresponding to each skeleton branch.

[0073] The fourth step is to use all the intersection points of the normal of each pixel point on each skeleton branch and the edge of its corresponding sub-region to form a set of intersection points corresponding to each pixel point on each skeleton branch.

[0074] It should be noted that if the sub-region corresponding to the skeleton branch represents a blood vessel, the number of intersection points in the intersection point set corresponding to the pixel points on the skeleton branch is often 2.

[0075] In the fifth step, the mean value of the distances between all intersection points in the set of intersection points corresponding to each pixel point on each skeleton branch is determined as the representative distance factor corresponding to each pixel point on each skeleton branch.

[0076] It should be noted that if a pixel is a blood vessel pixel, the representative distance factor corresponding to the pixel can represent the blood vessel width at the pixel. If the number of intersections in the intersection set corresponding to a pixel is 1, the representative distance factor corresponding to the pixel can be set to 0.

[0077] In the sixth step, according to the grayscale values ​​and representative distance factors corresponding to all pixels on each skeleton branch, determining the extensibility factor corresponding to each skeleton branch may include the following sub-steps: In the first sub-step, the product of the grayscale value corresponding to each pixel point on each skeleton branch and the representative distance factor is determined as the width grayscale factor corresponding to each pixel point on each skeleton branch.

[0078] In the second sub-step, the difference between the width grayscale factors corresponding to each adjacent pixel point on each skeleton branch is determined as the width grayscale difference, and a width grayscale difference set corresponding to each skeleton branch is obtained.

[0079] In the third sub-step, the absolute value of the accumulated values ​​of all the width grayscale differences in the width grayscale difference set corresponding to each skeleton branch is determined as the width grayscale representative difference corresponding to each skeleton branch.

[0080] In the fourth sub-step, the variance of all width grayscale differences in the width grayscale difference set corresponding to each skeleton branch is determined as the width grayscale variation factor corresponding to each skeleton branch.

[0081] In the fifth sub-step, the extensibility factor corresponding to each skeleton branch is determined according to the width grayscale representative difference and the width grayscale variation factor corresponding to each skeleton branch.

[0082] The width grayscale representative difference may be positively correlated with the extensibility factor, and the width grayscale variation factor may be negatively correlated with the extensibility factor.

[0083] For example, the formula for determining the extensibility factor corresponding to the skeleton branch can be: ; ;in, It is The extensibility factor corresponding to each skeleton branch. is the ordinal number of the skeleton branch. is a normalization function. It is an exponential function with a natural constant as its base. It is The width grayscale change factor corresponding to the skeleton branch, that is, The variance of all width grayscale differences in the width grayscale difference set corresponding to the skeleton branch. It is The number of pixels on each skeleton branch. It is The grayscale width corresponding to each skeleton branch represents the difference. It is The sequence number of the pixel points on the skeleton branch. It is On the skeleton branch The pixel point corresponds to the representative distance factor. It is On the skeleton branch The gray value corresponding to each pixel. It is On the skeleton branch The pixel point corresponds to the representative distance factor. It is On the skeleton branch The gray value corresponding to each pixel. It is the absolute value function. is the width grayscale difference. and is the width grayscale factor.

[0084] It should be noted that in actual situations, blood vessels often appear as one or more tubular or tree-like structures. Along the extension direction of the blood vessels, the diameter of the blood vessels tends to become narrower and the grayscale value tends to become smaller. For example, taking the pulmonary blood vessels as an example, along the extension direction of the pulmonary blood vessels, the width of the pulmonary blood vessels tends to become narrower, and the grayscale value of the pulmonary blood vessel pixels tends to become smaller. The more the extracted skeleton branches meet these characteristics, the stronger the extensibility of the extracted skeleton branches is, and the more likely the skeleton branches are to represent blood vessels. The larger the The more likely the sub-region corresponding to the skeleton branch is to meet the characteristics of grayscale and width decreasing in sequence, it often indicates that the The more skeleton branches the sub-region corresponds to, the more likely it is a vascular branch. The smaller the time, the more likely it is that The more likely the sub-regions corresponding to the skeleton branches are to conform to the grayscale and width decreasing in sequence, the more uniform the degree is, which often indicates that the The more skeleton branches the sub-region corresponds to, the more likely it is a vascular branch. The larger the The more skeleton branches a sub-region corresponds to, the more likely it is a vascular branch.

[0085] In the seventh step, threshold segmentation is performed on each region in each target matching pair to obtain the foreground region and background region corresponding to each region in each target matching pair.

[0086] In the eighth step, the vascular expression corresponding to each region in each target matching pair is determined based on the grayscale difference between the foreground region and the background region corresponding to each region in each target matching pair, and the extensibility factor corresponding to all skeleton branches in each region in each target matching pair.

[0087] For example, the formula for determining the blood vessel expression corresponding to the region in the target matching pair may be: ; ;in, It is Target matching pair The vascular expression corresponding to each region. is the sequence number of the target matching pair. It is The sequence number of the target matching area. is a normalization function. It is Target matching pair The cumulative value of the extensibility factors corresponding to all skeleton branches in a region. Characterization Target matching pair Chaos in the area. It is an exponential function with a natural constant as its base. It is the absolute value function. It is Target matching pair The mean grayscale value of all pixels in the foreground area of ​​a region. It is Target matching pair The mean grayscale value of all pixels in the background area of ​​a region. It is Target matching pair The number of skeleton branches in a region. It is Target matching pair The sequence number of the skeleton branch in the region. It is Target matching pair In the region The variance of the chain codes corresponding to all pixels on the skeleton branches. The chain codes corresponding to the pixels can be obtained through the eight-chain code algorithm.

[0088] It should be noted that when The larger the Target matching pair The more likely the skeleton branches in the first region are to be vascular branches, and the Target matching pair The more skeleton branches there are in a region, the more Target matching pair The more likely the area is to conform to the density of blood vessels, the more likely it is that the first Target matching pair The more likely the area is a vascular area. The larger the Target matching pair The more irregular the edges of the skeleton branches in a region, the more likely it is that Target matching pair The more chaotic the area is, the more likely it is that Target matching pair The more likely the area is to conform to the density of blood vessels. The smaller the time, the more likely it is that Target matching pair The worse the contrast of the segmentation of the first region, the lower the clarity of the segmentation. Target matching pair The more likely the area is to conform to the density of blood vessels. The larger the Target matching pair The more likely the region is to represent the true vascular region.

[0089] Step S7, determining the true restored expression of each target matching pair according to the difference between the tumor expressions corresponding to the two regions in each target matching pair and the difference between the corresponding blood vessel expressions.

[0090] As an example, the formula for determining the true recovery performance under the target matching pair can be: ; ; ;in, It is The true recovery performance under target matching pairs. is the sequence number of the target matching pair. is a normalization function. Indicates Tumor elimination under target matching pairs. Indicates Local recovery for target matching pairs. It is The tumor expression corresponding to the first region in the target matching pair. It is The tumor representation corresponding to the second region in the target matching pair. It is The vascular expression corresponding to the second region in the target matching pair. It is The vascular expression corresponding to the first region in the target matching pair.

[0091] It should be noted that before and after chemotherapy, the tumor expression level in the lesion area tends to decrease, and the vascular expression level tends to increase. The larger the The more the changes in tumor performance before and after chemotherapy in the target matching pair are consistent with the changes in the real lesion, the better the The more target matches there are, the more likely they are to represent the details that need to be preserved. The larger the The more the changes in vascular performance before and after chemotherapy in the target matching pair are consistent with the vascular changes of the actual lesion, the better the The more target matches there are, the more likely they are to represent the details that need to be preserved. The larger the The more likely a target match is to represent detailed information that needs to be retained, the more likely it is that the The number of pixels in its grayscale histogram that matches a target.

[0092] Step S8: performing histogram equalization on the first image to be enhanced and the second image to be enhanced according to the true restoration representation under all target matching pairs.

[0093] As an example, this step may include the following steps: In the first step, the real recovery performance degree under the target matching pair to which each first region belongs is determined as the real recovery performance factor corresponding to each first region.

[0094] In the second step, the grayscale histogram of the first image to be enhanced is adjusted according to the true restoration performance factors corresponding to all the first regions to obtain a first target histogram.

[0095] Among them, the horizontal and vertical axes of the grayscale histogram are the grayscale value and the number of pixels equal to the grayscale value.

[0096] For example, each grayscale value in all the first regions is determined as a candidate grayscale value, and based on the true restoration performance factor corresponding to all the first regions to which each candidate grayscale value belongs, the number of pixels in the grayscale histogram of the first image to be enhanced that is equal to the candidate grayscale value is increased to obtain the target number of pixels corresponding to each candidate grayscale value, and the number of pixels in the grayscale histogram of the first image to be enhanced for each candidate grayscale value is updated to its corresponding target number of pixels, and the grayscale histogram finally updated at this time is used as the first target histogram.

[0097] For example, determining the number of target pixels corresponding to each candidate grayscale value may include the following sub-steps: In the first sub-step, any candidate grayscale value is determined as the marked grayscale value, each first area to which the marked grayscale value belongs is determined as a temporary area, and the number of pixel points in each temporary area whose corresponding grayscale value is equal to the marked grayscale value is determined as the initial marked pixel number corresponding to each temporary area.

[0098] In the second sub-step, the product of the number of initial marked pixels corresponding to each temporary area and the true recovery performance factor is determined as the marked pixel increment corresponding to each temporary area.

[0099] The third sub-step is to round the accumulated values ​​of the marked pixel increments corresponding to all temporary areas to obtain the target pixel number corresponding to the marked grayscale value.

[0100] It should be noted that the same grayscale value may exist in different first regions, so a candidate grayscale value may belong to multiple first regions.

[0101] In the third step, the real recovery performance degree under the target matching pair to which each second region belongs is determined as the real recovery performance factor corresponding to each second region.

[0102] In the fourth step, the grayscale histogram of the second image to be enhanced is adjusted according to the true restoration performance factors corresponding to all the second regions to obtain a second target histogram.

[0103] It should be noted that the method for obtaining the second target histogram may be the same as the method for obtaining the first target histogram, which will not be described in detail herein.

[0104] In the fifth step, according to the first target histogram and the second target histogram, histogram equalization is performed on the first image to be enhanced and the second image to be enhanced respectively.

[0105] For example, the grayscale histogram of the first image to be enhanced is updated to the first target histogram, and the first image to be enhanced is histogram equalized according to the updated grayscale histogram. The grayscale histogram of the second image to be enhanced is updated to the second target histogram, and the second image to be enhanced is histogram equalized according to the updated grayscale histogram.

[0106] refer to Figure 3 Based on the same inventive concept as the above method embodiment, the present invention provides a CT image assisted reading system for the post-chemotherapy review stage, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the above computer program is executed by the processor, the steps of implementing a CT image assisted reading method for the post-chemotherapy review stage may specifically include: The CT image acquisition module 301 is used to acquire CT images before and after chemotherapy as the first image to be enhanced and the second image to be enhanced respectively; A region division module 302 is used to perform region division on the first image to be enhanced to obtain a first region, and to perform region division on the second image to be enhanced to obtain a second region; A matching degree determination module 303, used to determine the matching degree between each first region and each second region according to the distance and grayscale distribution difference between each first region and each second region; A region matching screening module 304 is used to perform region matching screening according to the matching degree between the first region and the second region to obtain a target matching pair representing changes before and after the operation; The tumor expression determination module 305 is used to obtain the tumor expression corresponding to each region in each target matching pair according to the gray distribution of the pre-extracted connected domain in each region in each target matching pair; A blood vessel expression determination module 306, configured to determine the blood vessel expression corresponding to each region in each target matching pair according to the characteristics of the pre-extracted skeleton in the connected domain in each region in each target matching pair; A true restoration expression determination module 307, used to determine the true restoration expression of each target matching pair according to the difference between the tumor expressions corresponding to the two regions in each target matching pair and the difference between the corresponding blood vessel expressions; The histogram equalization module 308 is used to perform histogram equalization on the first image to be enhanced and the second image to be enhanced according to the true restoration representation under all target matching pairs.

[0107] Figure 4 is a schematic diagram of the structure of a computer device provided by an embodiment of the present invention. Figure 4As shown, the computer device 400 includes: a memory 401, a processor 402, and a computer program 403 stored in the memory 401 and running on the processor 402, wherein when the processor 402 executes the computer program 403, the computer device can execute any one of the CT image-assisted reading methods for the post-chemotherapy review stage introduced above.

[0108] Based on the same inventive concept as the above method embodiment, the present invention provides a server, including a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, so that the device executes any of the above CT image-assisted reading methods in the post-chemotherapy review stage.

[0109] Based on the same inventive concept as the above-mentioned method embodiment, the present invention provides a computer program product, which includes: computer program code, when the computer program code is run on a computer, the computer executes any one of the above-mentioned CT image-assisted reading methods in the post-chemotherapy review stage.

[0110] Based on the same inventive concept as the above-mentioned method embodiment, the present invention provides a computer-readable storage medium, which stores a computer program code. When the computer program code runs on a computer, the computer executes any one of the above-mentioned CT image-assisted reading methods in the post-chemotherapy review stage.

[0111] In summary, compared with directly performing histogram equalization on CT images based on the grayscale histogram of the CT images, the present invention comprehensively considers multiple indicators related to the detailed information of disease recovery in CT images, such as the matching degree between regions in different CT images, tumor expression, vascular expression and true recovery expression, thereby improving the degree of retention of detailed information and further improving the effect of CT image enhancement.

[0112] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments can still be modified, or some of the technical features can be replaced by equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and should all be included in the protection scope of the present invention.

Claims

1. A CT image-assisted reading method for the post-chemotherapy review stage, characterized in that: The following steps are involved: Acquire CT images before and after chemotherapy as the first image to be enhanced and the second image to be enhanced, respectively; Performing region division on the first image to be enhanced to obtain a first region, and performing region division on the second image to be enhanced to obtain a second region; Determining a matching degree between each first region and each second region according to a distance and a grayscale distribution difference between each first region and each second region; Performing regional matching screening according to the matching degree between the first region and the second region to obtain a target matching pair representing changes before and after the operation; According to the grayscale distribution of the pre-extracted connected domain in each region in each target matching pair, the tumor expression corresponding to each region in each target matching pair is obtained; Determine the blood vessel expression corresponding to each region in each target matching pair according to the characteristics of the pre-extracted skeleton in the connected domain in each region in each target matching pair; According to the difference between the tumor expressions corresponding to the two regions in each target matching pair and the difference between the corresponding blood vessel expressions, the true restored expression under each target matching pair is determined; According to the true restoration performance under all target matching pairs, histogram equalization is performed on the first image to be enhanced and the second image to be enhanced.

2. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 1, characterized in that: The determining the matching degree between each first region and each second region according to the distance and grayscale distribution difference between each first region and each second region includes: Determine any one of the first regions as a marking region, and determine any one of the second regions as a reference region; Determine the region in the second image to be enhanced that is at the same position as the marked region as a position representative region; Determine the distance between the center point of the reference area and the center point of the position representative area as the target distance between the marked area and the reference area; Determine the absolute value of the difference between the grayscale values ​​corresponding to the pixels at the same position in the reference area and in the position representative area as the target grayscale difference, and obtain a target grayscale difference set between the reference area and the position representative area; Determine a target grayscale distribution difference between the reference area and the position representative area according to a target grayscale difference set between the reference area and the position representative area, wherein the target grayscale difference in the target grayscale difference set is positively correlated with the target grayscale distribution difference; The matching degree between the reference area and the position representative area is determined according to the target grayscale distribution difference and the target distance between the reference area and the position representative area, wherein the target grayscale distribution difference and the target distance are negatively correlated with the matching degree.

3. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 2, characterized in that: The determining the target grayscale distribution difference between the reference area and the position representative area according to the target grayscale difference set between the reference area and the position representative area comprises: The mean value of all target grayscale differences in the target grayscale difference set between the reference region and the position representative region is determined as the target grayscale distribution difference between the reference region and the position representative region.

4. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 1, characterized in that: The region matching screening is performed according to the matching degree between the first region and the second region to obtain a target matching pair representing changes before and after the operation, including: Determine any one of the first regions as a marked region, and select a second region having the same position as the marked region from all the second regions as a matching reference region corresponding to the marked region; The marked area and its corresponding matching reference area constitute a reference matching pair corresponding to the marked area, and the matching degree between the two areas in the reference matching pair is determined as a target matching index of the reference matching pair; A reference matching pair whose target matching index is less than a preset matching change threshold is selected from all reference matching pairs corresponding to the first region as a target matching pair.

5. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 1, characterized in that: The step of obtaining the tumor expression corresponding to each region in each target matching pair according to the grayscale distribution of the pre-extracted connected domain in each region in each target matching pair includes: Determine any region in any target matching pair as a candidate region, and perform connected domain extraction on the candidate region; Determine any connected domain in the candidate region as a marked connected domain, and determine the mean of the grayscale values ​​corresponding to all pixels in the marked connected domain as a grayscale representative factor corresponding to the marked connected domain; Determine the chain code corresponding to each pixel point on the edge of the marked connected domain by using the eight-chain code algorithm to obtain the chain code sequence corresponding to the marked connected domain; The difference between each two adjacent chain codes in the chain code sequence corresponding to the marked connected domain is determined as a target difference, so as to obtain a target difference sequence corresponding to the marked connected domain; Determine the variance of all target difference values ​​in the target difference value sequence corresponding to the marked connected domain as the edge clutter corresponding to the marked connected domain; Normalizing the product of the grayscale representative factor and the edge clutter corresponding to the marked connected domain to obtain a suspected tumor indicator corresponding to the marked connected domain; The average of the suspected tumor indicators corresponding to all connected domains in the candidate region is determined as the tumor expression degree corresponding to the candidate region.

6. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 1, characterized in that: Determining the blood vessel expression corresponding to each region in each target matching pair according to the characteristics of the pre-extracted skeleton in the connected domain in each region in each target matching pair includes: Determine any region in any target matching pair as a candidate region, and perform skeleton extraction on all connected domains in the candidate region; Taking the intersection points of all skeletons in the candidate area as segmentation points, segmenting all skeletons in the candidate area to obtain skeleton branches; Using the normals at the two endpoints of each skeleton branch as cutting lines, the connected domain of each skeleton branch is segmented to obtain the sub-region corresponding to each skeleton branch; All intersection points of the normal line of each pixel point on each skeleton branch and the edge of its corresponding sub-region form a set of intersection points corresponding to each pixel point on each skeleton branch; The average value of the distances between all intersection points in the set of intersection points corresponding to each pixel point on each skeleton branch is determined as the representative distance factor corresponding to each pixel point on each skeleton branch; Determine the extensibility factor corresponding to each skeleton branch according to the grayscale values ​​and representative distance factors corresponding to all pixels on each skeleton branch; Perform threshold segmentation on each region in each target matching pair to obtain the foreground region and background region corresponding to each region in each target matching pair; The vascular expression degree corresponding to each region in each target matching pair is determined according to the grayscale difference between the foreground region and the background region corresponding to each region in each target matching pair, and the extensibility factor corresponding to all skeleton branches in each region in each target matching pair.

7. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 6, characterized in that: Determining the extensibility factor corresponding to each skeleton branch according to the grayscale values ​​corresponding to all pixels on each skeleton branch and the representative distance factor includes: The product of the grayscale value corresponding to each pixel point on each skeleton branch and the representative distance factor is determined as the width grayscale factor corresponding to each pixel point on each skeleton branch; The difference between the width grayscale factors corresponding to each adjacent pixel point on each skeleton branch is determined as the width grayscale difference, and a width grayscale difference set corresponding to each skeleton branch is obtained; The absolute value of the accumulated values ​​of all the width grayscale differences in the width grayscale difference set corresponding to each skeleton branch is determined as the width grayscale representative difference corresponding to each skeleton branch; The variance of all width grayscale differences in the width grayscale difference set corresponding to each skeleton branch is determined as the width grayscale variation factor corresponding to each skeleton branch; According to the width grayscale representative difference and the width grayscale variation factor corresponding to each skeleton branch, the extensibility factor corresponding to each skeleton branch is determined, wherein the width grayscale representative difference is positively correlated with the extensibility factor, and the width grayscale variation factor is negatively correlated with the extensibility factor.

8. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 6, characterized in that: The formula for the vascular expression corresponding to the region in the target matching pair is: ； ;in, It is Target matching pair The vascular expression corresponding to each region; is the sequence number of the target matching pair; It is The serial number of the target matching area; is the normalization function; It is Target matching pair The cumulative value of the extensibility factors corresponding to all skeleton branches in a region; Characterization Target matching pair Disorder in the area; It is an exponential function with a natural constant as base; It is the absolute value function; It is Target matching pair The mean of the gray values ​​corresponding to all pixels in the foreground area of ​​the region; It is Target matching pair The mean of the grayscale values ​​corresponding to all pixels in the background area of ​​a region; It is Target matching pair The number of skeleton branches in a region; It is Target matching pair The serial number of the skeleton branches in each region; It is Target matching pair In the region The variance of the chain codes corresponding to all pixels on the skeleton branches; the chain codes corresponding to the pixels are obtained through the eight-chain code algorithm.

9. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 1, characterized in that: The formula corresponding to the true recovery performance under the target matching pair is: ； ； ;in, It is The actual recovery performance under target matching pairs; is the sequence number of the target matching pair; is the normalization function; Indicates Tumor elimination under target matching pairs; Indicates Local recovery under target matching pairs; It is The tumor expression corresponding to the first region in the target matching pair; It is The tumor expression corresponding to the second region in the target matching pair; It is The vascular expression corresponding to the second region in the target matching pair; It is The vascular expression corresponding to the first region in the target matching pair.

10. The CT image-assisted reading method for the post-chemotherapy review stage according to claim 1, characterized in that: The performing histogram equalization on the first image to be enhanced and the second image to be enhanced according to the true restoration performance under all target matching pairs includes: Determine the true recovery performance degree of each target matching pair to which each first region belongs as a true recovery performance factor corresponding to each first region; Adjusting the grayscale histogram of the first image to be enhanced according to the true restoration performance factors corresponding to all the first regions to obtain a first target histogram; Determine the true recovery performance degree under the target matching pair to which each second region belongs as the true recovery performance factor corresponding to each second region; According to the true restoration performance factors corresponding to all the second regions, the grayscale histogram of the second image to be enhanced is adjusted to obtain a second target histogram; Histogram equalization is performed on the first image to be enhanced and the second image to be enhanced according to the first target histogram and the second target histogram.

Images