CN118297812B - Liver cancer pathological image enhancement method based on artificial intelligence - Google Patents
Liver cancer pathological image enhancement method based on artificial intelligence Download PDFInfo
- Publication number
- CN118297812B CN118297812B CN202410704759.1A CN202410704759A CN118297812B CN 118297812 B CN118297812 B CN 118297812B CN 202410704759 A CN202410704759 A CN 202410704759A CN 118297812 B CN118297812 B CN 118297812B
- Authority
- CN
- China
- Prior art keywords
- pixel point
- pixel
- liver cancer
- points
- neighborhood
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
- 201000007270 liver cancer Diseases 0.000 title claims abstract description 150
- 208000014018 liver neoplasm Diseases 0.000 title claims abstract description 150
- 238000000034 method Methods 0.000 title claims abstract description 61
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 16
- 230000001575 pathological effect Effects 0.000 title claims abstract description 15
- 210000004185 liver Anatomy 0.000 claims abstract description 47
- 238000001914 filtration Methods 0.000 claims abstract description 46
- 238000007781 pre-processing Methods 0.000 claims abstract description 3
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000002708 enhancing effect Effects 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 230000011218 segmentation Effects 0.000 claims description 4
- 206010028980 Neoplasm Diseases 0.000 claims description 3
- 201000011510 cancer Diseases 0.000 claims description 3
- 230000007170 pathology Effects 0.000 claims 2
- 230000000694 effects Effects 0.000 abstract description 6
- 230000007704 transition Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000002591 computed tomography Methods 0.000 description 2
- 238000010827 pathological analysis Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000000758 substrate Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/20—Image enhancement or restoration using local operators
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/10—Image enhancement or restoration using non-spatial domain filtering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30056—Liver; Hepatic
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The invention relates to the technical field of image processing, in particular to an artificial intelligence-based liver cancer pathological image enhancement method, which comprises the following steps: collecting and preprocessing CT images of liver parts; acquiring a liver region in a CT image; acquiring a first liver cancer edge factor of each pixel point according to the gray value of each pixel point in the liver region; obtaining local pixel points of each pixel point; acquiring a second liver cancer edge factor of the pixel point according to the pixel point, the gradient direction of the local pixel point of the pixel point and the first liver cancer edge factor; and obtaining the Laplacian filtering weight of the pixel point according to the second liver cancer edge factor of the pixel point to enhance the CT image. According to the invention, the Laplace filtering weight of each pixel point is obtained in a self-adaptive manner by analyzing the possibility that each pixel point is the liver cancer tissue edge, so that the occurrence of image distortion is avoided, and the image enhancement effect is improved.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an artificial intelligence-based liver cancer pathological image enhancement method.
Background
CT images of liver sites can be generally used for pathological analysis of liver cancer; however, in pathological analysis of liver cancer, a doctor usually marks liver cancer tissue manually, and it is time-consuming and cumbersome to mark liver cancer tissue manually, so that it is necessary to enhance CT images of liver parts of patients in order to make it more convenient and accurate for the doctor to obtain liver cancer tissue.
However, in general, when a CT image of a liver portion of a patient is enhanced by using a conventional laplace filter, a convolution kernel of the laplace filter for all pixels is the same by the conventional laplace filter, and thus, the enhanced image may be distorted, and the enhancement effect may be poor.
Disclosure of Invention
The invention provides an artificial intelligence-based liver cancer pathological image enhancement method, which aims to solve the existing problems: the enhancement effect of the traditional Laplace filtering on the CT image is poor.
The invention discloses an artificial intelligence-based liver cancer pathological image enhancement method, which adopts the following technical scheme:
The method comprises the following steps:
collecting and preprocessing CT images of liver parts;
Acquiring a liver region in a CT image; acquiring a first liver cancer edge factor of each pixel point in the liver region according to the difference of gray values of each pixel point and the neighborhood pixel points;
presetting a local range, and obtaining local pixel points of each pixel point; combining the pixel points and local pixel points of the pixel points to obtain a plurality of combinations of the pixel points; acquiring a second liver cancer edge factor of the pixel point according to the gradient direction of the pixel point and the first liver cancer edge factor of the pixel point in all combinations of the pixel points;
Obtaining the Laplacian filtering weight of the pixel point according to the second liver cancer edge factor of the pixel point; and acquiring a convolution kernel for carrying out Laplacian filtering on the pixel points according to the Laplacian filtering weights of the pixel points, and enhancing the CT image.
Preferably, the method for obtaining the first liver cancer edge factor of the pixel point according to the difference of gray values between each pixel point and the neighboring pixel points in the liver region includes the following specific steps:
according to the first Combining all point characteristic values on the neighborhood circle of each pixel point with the firstThe gray value of each pixel point is obtainedThe specific calculation formula of the first liver cancer edge factor of each pixel point is as follows:
In the method, in the process of the invention, Represent the firstA first liver cancer edge factor of each pixel point; Represent the first The number of the neighborhood circle points of each pixel point; Represent the first Gray values of the individual pixels; Represent the first Neighborhood circle of each pixel pointA point feature value; Representing a hyperbolic tangent function.
Preferably, the firstThe specific acquisition method of the characteristic values of all points on the neighborhood circle of each pixel point comprises the following steps:
According to First, theNeighborhood circle of each pixel pointGray values of a first reference pixel point and a second reference pixel point of each point are obtainedNeighborhood circle of each pixel pointThe specific calculation formula of the characteristic values of the points is as follows:
In the method, in the process of the invention, Represent the firstNeighborhood circle of each pixel pointA point feature value; Represent the first Neighborhood circle of each pixel pointGray values of first reference pixel points of the points; Represent the first Neighborhood circle of each pixel pointGray values of second reference pixel points of the points; Represent the first The pixel point and the firstNeighborhood circle of each pixel pointThe angle of the connection line of the points; Representing the remainder function.
Preferably, the saidFirst, theNeighborhood circle of each pixel pointThe specific acquisition method of the first reference pixel point and the second reference pixel point of each point comprises the following steps:
for the first Neighborhood circle of each pixel pointA plurality of points, wherein the horizontal rightward direction is 0 DEG, and the vertical upward direction is 90 DEG; connect the firstThe pixel point and the firstNeighborhood circle of each pixel pointA point is obtained to obtain the firstThe pixel point and the firstNeighborhood circle of each pixel pointThe connection of the points and the acquisition of the angle of the connection are recorded as；
If it isGreater thanAnd less than or equal toWill be the firstThe first neighborhood pixel point and the second neighborhood pixel point of each pixel point are respectively marked as the first neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstThe second neighborhood pixel point and the third neighborhood pixel point of each pixel point are marked as the second neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstThe third neighborhood pixel point and the fourth neighborhood pixel point of each pixel point are marked as the fourth neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstA fourth adjacent pixel point and a first adjacent pixel point of each pixel point are marked as the firstNeighborhood circle of each pixel pointThe first reference pixel point and the second reference pixel point of each point.
Preferably, the firstThe specific acquisition method of the neighborhood circle of each pixel point comprises the following steps:
will be the first All pixel points in the four-neighborhood range of each pixel point are marked as the first pixel pointNeighborhood pixel point of each pixel point and is to be positioned at the first pixel pointRight side, upper side, left side and lower side of each pixel pointNeighborhood pixel points of each pixel point are respectively marked as the first pixel pointThe first neighborhood pixel point, the second neighborhood pixel point, the third neighborhood pixel point and the fourth neighborhood pixel point of the pixel points; in the first placeThe center of each pixel point is used as the center of a circle, and a circle with any radius is made and marked as the firstNeighborhood circles of individual pixels.
Preferably, a local range is preset, and local pixel points of each pixel point are obtained; combining the pixel points and local pixel points of the pixel points to obtain a plurality of combinations of the pixel points, wherein the specific method comprises the following steps:
for the first A first pixel pointConstructing one pixel point as centerDividing the window by the first windowAll the pixel points except the pixel point are marked as the first pixel pointLocal pixel points of the pixel points;
will be the first Pixel dot and slaveSelecting local pixel points of the pixel pointsThe pixel point is taken as the firstA combination of pixel points to obtain the firstSeveral combinations of pixels.
Preferably, the method for obtaining the second liver cancer edge factor of the pixel point according to the gradient direction of the pixel point and the first liver cancer edge factor of the pixel point in all combinations of the pixel points comprises the following specific steps:
for the first The first pixel pointCombinations of (a) and (b)The first pixel pointThe pixel points in the combinations are paired pairwise to obtain the first pixel pointThe first pixel pointA plurality of pixel point pairs in the plurality of combinations;
for the first The first pixel pointIn the first combinationA pixel point pair is used for calculating the firstThe first pixel pointIn the first combinationCosine similarity between gradient directions of two pixel points in each pixel point pair is marked as the firstThe first pixel pointIn the first combinationSimilarity values for the pairs of pixels;
Will be for the first The first pixel pointThe average value of the similarity values of all pixel point pairs in each combination is recorded as the firstThe first pixel pointSimilarity values for each combination;
according to the first The first pixel pointCharacteristic values of each combinationThe first pixel pointThe first liver cancer edge factor of each pixel point in each combination is obtainedThe first pixel pointThe liver cancer edge coefficients of each combination;
maximum liver cancer edge coefficient The combination of the pixel points is marked as the firstCombining the most similar edges of the pixel points and combining the first pixel pointLiver cancer edge coefficient of the most similar edge combination of each pixel point is used as the firstAnd a second liver cancer edge factor of each pixel point.
Preferably, the acquiring a firstThe first pixel pointThe specific calculation method of the liver cancer edge coefficients comprises the following steps:
In the method, in the process of the invention, Represent the firstThe first pixel pointThe liver cancer edge coefficients of each combination; Represent the first The first pixel pointSimilarity values for each combination; representing a preset window side length; Represent the first The first pixel pointIn the first combinationA first liver cancer edge factor of each pixel point; Representing a maximum minimum normalization function.
Preferably, the laplace filter weight of the pixel point is obtained according to the second liver cancer edge factor of the pixel point; according to the Laplace filtering weight of the pixel point, a convolution kernel for carrying out Laplace filtering on the pixel point is obtained, and the specific method comprises the following steps:
for the first A pixel point is acquiredSecond liver cancer edge factors of all pixel points in the most similar edge combination of the pixel pointsThe average value of the second liver cancer edge factors of all the pixel points in the most similar edge combination of the pixel points is recorded as the firstThe liver cancer edge characteristic value of each pixel point is compared with the first curve tangent functionNormalizing the liver cancer edge characteristic value of each pixel point to obtain the first pixel pointThe Laplacian filtering weight of each pixel point is recorded as；
Presetting an initial Laplace convolution kernel; according to the firstThe Laplace filtering weight of each pixel point weights the initial Laplace convolution kernel to obtain the second pixel pointA convolution kernel of Laplace filtering is carried out on each pixel point;
according to the pair of Convolution kernel of Laplace filtering is carried out on each pixel point, and the first pixel point is subjected toAnd carrying out Laplacian filtering on each pixel point.
Preferably, the method for acquiring the liver region in the CT image includes the following specific steps:
and acquiring a liver region in the CT image by using a semantic segmentation algorithm.
The technical scheme of the invention has the beneficial effects that: according to the gray value of each pixel point in the liver region, the first liver cancer edge factor of the pixel point is obtained, the pixel points in liver cancer tissues are primarily quantized by analyzing the smoothness degree of gray transition between each pixel point and the neighborhood pixel point, and meanwhile, data support is provided for subsequently obtaining the second liver cancer edge factor of the pixel point; obtaining local pixel points of each pixel point; obtaining a second liver cancer edge factor of the pixel point according to the gradient directions of the pixel point and the local pixel point of the pixel point and the first liver cancer edge factor, and obtaining the characteristics of the liver cancer tissue edge of each pixel point by measuring the possibility that the pixel point is an edge pixel point of liver cancer tissue from the continuity, thereby providing support for the subsequent acquisition of the Laplace filtering weight of each pixel point; and according to the second liver cancer edge factors of the pixels, obtaining the Laplacian filtering weights of the pixels to strengthen the CT image, and giving the Laplacian filtering weights of the pixels with large second liver cancer edge factors and giving the Laplacian filtering weights of the pixels with small second liver cancer edge factors to sharpen the liver cancer tissue edges, so that the situation of image enhancement distortion is avoided, and the enhancement effect of liver cancer pathological images is improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of steps of a liver cancer pathological image enhancement method based on artificial intelligence;
Fig. 2 is a schematic diagram of a neighborhood circle and a neighborhood pixel of a pixel.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following detailed description is given below of a liver cancer pathological image enhancement method based on artificial intelligence according to the invention, which is provided by combining the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
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.
The invention provides a specific scheme of an artificial intelligence-based liver cancer pathological image enhancement method, which is specifically described below with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of steps of an artificial intelligence-based liver cancer pathological image enhancement method according to an embodiment of the present invention is shown, and the method includes the following steps:
Step S001: CT images of the liver region are acquired and preprocessed.
It should be noted that CT (Computed Tomography, electronic computer tomography) is a safe and reliable image inspection method, and a doctor can obtain liver cancer tissues in the liver of a patient while judging whether the patient has liver cancer or not through CT images of the liver part of the patient, so that the method is widely applied to the medical field; however, since the traditional method for acquiring liver cancer tissue through CT image is that doctor marks liver cancer tissue manually, but since the gray value of pixel point in liver cancer tissue is similar to the gray value of pixel point in normal tissue, it is time consuming and complicated to mark liver cancer tissue manually, so in order to make doctor acquire liver cancer tissue better, this embodiment provides a liver cancer pathological image enhancement method based on artificial intelligence, by analyzing the feature of liver cancer tissue in CT image, the edge of liver cancer tissue in CT image is enhanced, so doctor can acquire liver cancer tissue better, for this reason, CT image of liver region of patient needs to be acquired first.
Specifically, the CT scanner is used to collect CT images of the liver of the patient, and the detailed process of collecting CT by the CT scanner is known as the prior art, so that the description is omitted in this embodiment.
It should be further noted that noise is often associated with the process of acquiring CT images, and for this purpose, denoising the CT image of the liver region is also required.
Specifically, median filtering denoising treatment is carried out on the CT image of the liver part of the patient, and the CT image of the liver part of the patient after median filtering treatment is obtained and recorded as a CT image; since the specific process of the median filtering denoising is known as the prior art, the description is omitted in this embodiment.
Thus, a CT image is obtained.
Step S002: acquiring a liver region in a CT image; and acquiring a first liver cancer edge factor of each pixel point in the liver region according to the difference of the gray value between each pixel point and the neighborhood pixel points.
It should be noted that, in this embodiment, as an artificial intelligence-based liver cancer pathological image enhancement method, specifically, by analyzing features of a liver cancer edge, if laplace filtering is performed on an entire CT image, an enhanced image distortion enhancement effect may be poor, so that in this embodiment, by acquiring a laplace filtering weight of each pixel point in the CT image, by giving a laplace filtering weight that is large to a pixel point having a liver cancer edge feature, giving a laplace filtering weight that is small to a pixel point not having a liver cancer edge feature, an edge of a liver cancer tissue in the CT image is sharpened, and thus an edge of a liver cancer tissue in the CT image is enhanced, and finally, a doctor can better acquire a liver cancer tissue.
It should be further noted that, in addition to the liver region of the patient, the CT image also includes a portion other than the liver, and the liver cancer only exists in the liver region, so that the liver region in the CT image needs to be segmented, and the liver region in the CT image is analyzed separately, so as to obtain the liver cancer tissue more accurately.
Specifically, a semantic segmentation algorithm is utilized to segment the CT image, and a liver region in the CT image is obtained.
It should be noted that, in this embodiment, deepLabV neural network is used as the neural network model of the semantic segmentation algorithm, the cross entropy loss function is used as the loss function of the DeepLabV neural network, which is a well-known prior art, and in this embodiment, a specific training method of the neural network is not repeated, because liver cancer tissue exists in the liver region, after the liver region is obtained, all the pixels analyzed later are pixels in the liver region. However, the edge of liver cancer tissue in the liver region has irregular characteristics, so that the edge can be used as a basis for distinguishing liver cancer tissue from normal tissue in the liver region, and the first liver cancer edge factor of the pixel point in the liver region can be obtained.
Specifically, for the firstA pixel point of the first pixelAll pixel points in the four-neighborhood range of each pixel point are marked as the first pixel pointNeighborhood pixel point of each pixel point and is to be positioned at the first pixel pointRight side, upper side, left side and lower side of each pixel pointNeighborhood pixel points of each pixel point are respectively marked as the first pixel pointThe first neighborhood pixel point, the second neighborhood pixel point, the third neighborhood pixel point and the fourth neighborhood pixel point of the pixel points; in the first placeThe center of each pixel point is used as the center of a circle, and a circle with any radius is made and marked as the firstNeighborhood circles of the pixel points are shown in fig. 2;
for the first Neighborhood circle of each pixel pointA plurality of points, wherein the horizontal rightward direction is 0 DEG, and the vertical upward direction is 90 DEG; connect the firstThe pixel point and the firstNeighborhood circle of each pixel pointA point is obtained to obtain the firstThe pixel point and the firstNeighborhood circle of each pixel pointThe connection of the points and the acquisition of the angle of the connection are recorded as；
If it isGreater thanAnd less than or equal toWill be the firstThe first neighborhood pixel point and the second neighborhood pixel point of each pixel point are respectively marked as the first neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstThe second neighborhood pixel point and the third neighborhood pixel point of each pixel point are marked as the second neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstThe third neighborhood pixel point and the fourth neighborhood pixel point of each pixel point are marked as the fourth neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstA fourth adjacent pixel point and a first adjacent pixel point of each pixel point are marked as the firstNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point;
According to First, theNeighborhood circle of each pixel pointGray values of a first reference pixel point and a second reference pixel point of each point are obtainedNeighborhood circle of each pixel pointThe specific calculation formula of the characteristic values of the points is as follows:
In the method, in the process of the invention, Represent the firstNeighborhood circle of each pixel pointA point feature value; Represent the first Neighborhood circle of each pixel pointGray values of first reference pixel points of the points; Represent the first Neighborhood circle of each pixel pointGray values of second reference pixel points of the points; Represent the first The pixel point and the firstNeighborhood circle of each pixel pointThe angle of the connection line of the points; Representing the remainder function.
Further, according to the firstCombining all point characteristic values on the neighborhood circle of each pixel point with the firstThe gray value of each pixel point is obtainedThe specific calculation formula of the first liver cancer edge factor of each pixel point is as follows:
In the method, in the process of the invention, Represent the firstA first liver cancer edge factor of each pixel point; Represent the first The number of the neighborhood circle points of each pixel point; Represent the first Gray values of the individual pixels; Represent the first Neighborhood circle of each pixel pointA point feature value; a hyperbolic tangent function is shown, used in this example to perform the normalization operation.
It should be noted that the number of the substrates,Representing the firstThe bumpy level of the gray value transitions of the individual pixels,The larger the value of (2), the description of the (1)The less smooth the transition of the gray value of each pixel point, and when the thirdThe first pixel is positioned at the edge of liver cancer tissueThe transition of the gray value of each pixel point is not smooth; thus (2)The greater the value of (2)The greater the likelihood that a pixel is an edge pixel of liver cancer.
Thus, the first liver cancer edge factor of the pixel point is obtained.
Step S003: presetting a local range, and obtaining local pixel points of each pixel point; combining the pixel points and local pixel points of the pixel points to obtain a plurality of combinations of the pixel points; and acquiring a second liver cancer edge factor of the pixel point according to the gradient direction of the pixel point and the first liver cancer edge factor of the pixel point in all combinations of the pixel points.
It should be noted that, since the first liver cancer edge factor of all the pixels in the liver region obtained in step S002 represents the feature of the liver cancer tissue edge pixel point of the single pixel, in order to better enhance the edge of the liver cancer tissue in the CT image, further analysis is required for all the pixels in the liver region. Since the edge pixel points of other liver cancer tissues are necessarily present in the local range of the edge pixel points of the liver cancer tissues in the liver region, the second liver cancer edge factors of the pixel points can be obtained based on the edge pixel points.
Specifically, for the firstA first pixel pointConstructing one pixel point as centerDividing the window by the first windowAll the pixel points except the pixel point are marked as the first pixel pointLocal pixel points of the pixel points; the saidFor a predetermined window edge length,The specific value of (2) can be set by combining with the actual situation, the embodiment does not have hard requirement, in the embodiment, the method usesDescription is made;
will be the first Pixel dot and slaveSelecting local pixel points of the pixel pointsThe pixel point is taken as the firstA combination of pixel points to obtain the firstA plurality of combinations of pixel points;
for the first The first pixel pointCombinations of (a) and (b)The first pixel pointThe pixel points in the combinations are paired pairwise to obtain the first pixel pointThe first pixel pointA plurality of pixel point pairs in the plurality of combinations;
for the first The first pixel pointIn the first combinationA pixel point pair is used for calculating the firstThe first pixel pointIn the first combinationCosine similarity between gradient directions of two pixel points in each pixel point pair is marked as the firstThe first pixel pointIn the first combinationSimilarity values for the pairs of pixels; since the specific method for obtaining the gradient direction and cosine similarity of the pixel points is known as the prior art, the description is not repeated in this embodiment;
Will be for the first The first pixel pointThe average value of the similarity values of all pixel point pairs in each combination is recorded as the firstThe first pixel pointSimilarity values for each combination;
Further, according to the first The first pixel pointCharacteristic values of each combinationThe first pixel pointThe first liver cancer edge factor of each pixel point in each combination is obtainedThe first pixel pointThe specific calculation formula of the combined liver cancer edge coefficients is as follows:
In the method, in the process of the invention, Represent the firstThe first pixel pointThe liver cancer edge coefficients of each combination; Represent the first The first pixel pointSimilarity values for each combination; representing a preset window side length; Represent the first The first pixel pointIn the first combinationA first liver cancer edge factor of each pixel point; representing a maximum-minimum normalization function, the normalization object being all combinations of all pixel points ；
Maximum liver cancer edge coefficientThe combination of the pixel points is marked as the firstCombining the most similar edges of the pixel points and combining the first pixel pointLiver cancer edge coefficient of the most similar edge combination of each pixel point is used as the firstAnd a second liver cancer edge factor of each pixel point.
Note that, since the edge pixel points of other liver cancer tissues are necessarily present in the local area of the edge pixel points of the liver cancer tissues in the liver region, the following appliesThe pixel points are edge pixel points of liver cancer tissue, and the first pixel point isAt least one local pixel point of each pixel point existsEdge pixel points of liver cancer tissues; also, since the edge of liver cancer tissue is always uneven, the gradient direction of the edge pixel point of liver cancer tissue in local area is not uniform, soThe smaller the value, the firstThe first pixel pointThe more the pixel points in the combination have the characteristics of the edge pixel points of liver cancer tissues, and simultaneouslyRepresenting the firstThe first pixel pointIn the first combinationA first liver cancer edge factor of each pixel point, thusThe greater the value of (2)The first pixel pointThe more the pixel points in the combination are provided with the characteristics of the edge pixel points of liver cancer tissues; therefore, it isThe greater the value of (2)The first pixel pointThe more likely the pixel points in the combination are edge pixel points of liver cancer tissues; thus if the firstThe pixel points are edge pixel points of liver cancer tissue, the firstThe pixel points in the most similar edge combination of the pixel points are the edge pixel points most likely to be liver cancer tissues, and the continuity is measured based on the edge pixel pointsThe pixel points are the edge pixel point possibility of liver cancer tissues.
Thus, the second liver cancer edge factor of the pixel point is obtained.
Step S004: obtaining the Laplacian filtering weight of the pixel point according to the second liver cancer edge factor of the pixel point; and acquiring a convolution kernel for carrying out Laplacian filtering on the pixel points according to the Laplacian filtering weights of the pixel points, and enhancing the CT image.
It should be noted that, in the step S003, the second liver cancer edge factor of the pixel point is the firstThe probability that each pixel point is an edge pixel point of liver cancer tissue can be enhanced by giving larger Laplacian filtering weight to the edge pixel point which is more likely to be liver cancer tissue and giving smaller Laplacian filtering weight to the edge pixel point which is less likely to be liver cancer tissue.
Specifically, for the firstA pixel point is acquiredSecond liver cancer edge factors of all pixel points in the most similar edge combination of the pixel pointsThe average value of the second liver cancer edge factors of all the pixel points in the most similar edge combination of the pixel points is recorded as the firstThe liver cancer edge characteristic value of each pixel point is compared with the first curve tangent functionNormalizing the liver cancer edge characteristic value of each pixel point to obtain the first pixel pointThe Laplacian filtering weight of each pixel point is recorded as；
Then an initial Laplace convolution kernel is preset, and the initial Laplace convolution kernel can be set by combining with the actual situation to make the embodiment do not do hard requirements, in the embodiment, the initial Laplace convolution kernel is taken asDescription is made; according to the firstThe Laplace filtering weight of each pixel point weights the initial Laplace convolution kernel to obtain the second pixel pointConvolution kernel of laplace filtering is performed on each pixel point:
according to the pair of Convolution kernel of Laplace filtering is carried out on each pixel point, and the first pixel point is subjected toThe laplace filtering is performed on each pixel, and since the laplace filtering is a well-known prior art, a detailed description is omitted in this embodiment.
It should be noted that, for a pixel, if the pixel and other pixels in the local area of the pixel are more provided with the feature of the edge pixel of the liver cancer tissue, the value of the laplace convolution kernel center of the pixel should be increased to strengthen the liver cancer tissue edge detail of the pixel, and if the pixel and other pixels in the local area of the pixel are less provided with the feature of the edge pixel of the liver cancer tissue, the value of the laplace convolution kernel center of the pixel should be decreased to avoid the occurrence of the condition of enhancement distortion; through filtering all pixel points in the liver region, the edge of liver cancer tissues in the liver region is enhanced finally, so that doctors can obtain the liver cancer tissues better.
This embodiment is completed.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the invention, but any modifications, equivalent substitutions, improvements, etc. within the principles of the present invention should be included in the scope of the present invention.
Claims (4)
1. An artificial intelligence-based liver cancer pathological image enhancement method is characterized by comprising the following steps of:
collecting and preprocessing CT images of liver parts;
Acquiring a liver region in a CT image; acquiring a first liver cancer edge factor of each pixel point in the liver region according to the difference of gray values of each pixel point and the neighborhood pixel points;
presetting a local range, and obtaining local pixel points of each pixel point; combining the pixel points and local pixel points of the pixel points to obtain a plurality of combinations of the pixel points; acquiring a second liver cancer edge factor of the pixel point according to the gradient direction of the pixel point and the first liver cancer edge factor of the pixel point in all combinations of the pixel points;
Obtaining the Laplacian filtering weight of the pixel point according to the second liver cancer edge factor of the pixel point; acquiring a convolution kernel for carrying out Laplace filtering on the pixel points according to the Laplace filtering weights of the pixel points, and enhancing the CT image;
According to the difference of gray values of each pixel point and the neighborhood pixel points in the liver region, a first liver cancer edge factor of the pixel point is obtained, and the specific method comprises the following steps:
according to the first Combining all point characteristic values on the neighborhood circle of each pixel point with the firstThe gray value of each pixel point is obtainedThe specific calculation formula of the first liver cancer edge factor of each pixel point is as follows:
In the method, in the process of the invention, Represent the firstA first liver cancer edge factor of each pixel point; Represent the first The number of the neighborhood circle points of each pixel point; Represent the first Gray values of the individual pixels; Represent the first Neighborhood circle of each pixel pointA point feature value; Representing a hyperbolic tangent function;
Said first The specific acquisition method of the characteristic values of all points on the neighborhood circle of each pixel point comprises the following steps:
According to First, theNeighborhood circle of each pixel pointGray values of a first reference pixel point and a second reference pixel point of each point are obtainedNeighborhood circle of each pixel pointThe specific calculation formula of the characteristic values of the points is as follows:
In the method, in the process of the invention, Represent the firstNeighborhood circle of each pixel pointA point feature value; Represent the first Neighborhood circle of each pixel pointGray values of first reference pixel points of the points; Represent the first Neighborhood circle of each pixel pointGray values of second reference pixel points of the points; Represent the first The pixel point and the firstNeighborhood circle of each pixel pointThe angle of the connection line of the points; Representing a remainder function;
The said First, theNeighborhood circle of each pixel pointThe specific acquisition method of the first reference pixel point and the second reference pixel point of each point comprises the following steps:
for the first Neighborhood circle of each pixel pointA plurality of points, wherein the horizontal rightward direction is 0 DEG, and the vertical upward direction is 90 DEG; connect the firstThe pixel point and the firstNeighborhood circle of each pixel pointA point is obtained to obtain the firstThe pixel point and the firstNeighborhood circle of each pixel pointThe connection of the points and the acquisition of the angle of the connection are recorded as；
Will be the firstAll pixel points in the four-neighborhood range of each pixel point are marked as the first pixel pointNeighborhood pixel point of each pixel point and is to be positioned at the first pixel pointRight side, upper side, left side and lower side of each pixel pointNeighborhood pixel points of each pixel point are respectively marked as the first pixel pointThe first neighborhood pixel point, the second neighborhood pixel point, the third neighborhood pixel point and the fourth neighborhood pixel point of the pixel points;
If it is Greater thanAnd less than or equal toWill be the firstThe first neighborhood pixel point and the second neighborhood pixel point of each pixel point are respectively marked as the first neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstThe second neighborhood pixel point and the third neighborhood pixel point of each pixel point are marked as the second neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstThe third neighborhood pixel point and the fourth neighborhood pixel point of each pixel point are marked as the fourth neighborhood pixel pointNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point; if it isGreater thanAnd less than or equal toWill be the firstA fourth adjacent pixel point and a first adjacent pixel point of each pixel point are marked as the firstNeighborhood circle of each pixel pointA first reference pixel point and a second reference pixel point of each point;
Said first The specific acquisition method of the neighborhood circle of each pixel point comprises the following steps:
in the first place The center of each pixel point is used as the center of a circle, and a circle with any radius is made and marked as the firstNeighborhood circles of the pixel points;
The method for obtaining the second liver cancer edge factor of the pixel point according to the gradient direction of the pixel point and the first liver cancer edge factor of the pixel point in all combinations of the pixel point comprises the following specific steps:
for the first The first pixel pointCombinations of (a) and (b)The first pixel pointThe pixel points in the combinations are paired pairwise to obtain the first pixel pointThe first pixel pointA plurality of pixel point pairs in the plurality of combinations;
for the first The first pixel pointIn the first combinationA pixel point pair is used for calculating the firstThe first pixel pointIn the first combinationCosine similarity of gradient direction between two pixels in each pixel pair is marked as the firstThe first pixel pointIn the first combinationSimilarity values for the pairs of pixels;
Will be for the first The first pixel pointThe average value of the similarity values of all pixel point pairs in each combination is recorded as the firstThe first pixel pointSimilarity values for each combination;
according to the first The first pixel pointCharacteristic values of each combinationThe first pixel pointThe first liver cancer edge factor of each pixel point in each combination is obtainedThe first pixel pointThe liver cancer edge coefficients of each combination;
maximum liver cancer edge coefficient The combination of the pixel points is marked as the firstCombining the most similar edges of the pixel points and combining the first pixel pointLiver cancer edge coefficient of the most similar edge combination of each pixel point is used as the firstA second liver cancer edge factor of each pixel point;
The acquisition of the first The first pixel pointThe specific calculation method of the liver cancer edge coefficients comprises the following steps:
In the method, in the process of the invention, Represent the firstThe first pixel pointThe liver cancer edge coefficients of each combination; Represent the first The first pixel pointSimilarity values for each combination; representing a preset window side length; Represent the first The first pixel pointIn the first combinationA first liver cancer edge factor of each pixel point; Representing a maximum minimum normalization function.
2. The method for enhancing liver cancer pathology image based on artificial intelligence according to claim 1, wherein a local range is preset, and local pixel points of each pixel point are obtained; combining the pixel points and local pixel points of the pixel points to obtain a plurality of combinations of the pixel points, wherein the specific method comprises the following steps:
for the first A first pixel pointConstructing one pixel point as centerDividing the window by the first windowAll the pixel points except the pixel point are marked as the first pixel pointLocal pixel points of the pixel points;
will be the first Pixel dot and slaveSelecting local pixel points of the pixel pointsThe pixel point is taken as the firstA combination of pixel points to obtain the firstSeveral combinations of pixels.
3. The liver cancer pathological image enhancement method based on artificial intelligence according to claim 1, wherein the laplace filter weight of the pixel point is obtained according to the second liver cancer edge factor of the pixel point; according to the Laplace filtering weight of the pixel point, a convolution kernel for carrying out Laplace filtering on the pixel point is obtained, and the specific method comprises the following steps:
for the first A pixel point is acquiredSecond liver cancer edge factors of all pixel points in the most similar edge combination of the pixel pointsThe average value of the second liver cancer edge factors of all the pixel points in the most similar edge combination of the pixel points is recorded as the firstThe liver cancer edge characteristic value of each pixel point is compared with the first curve tangent functionNormalizing the liver cancer edge characteristic value of each pixel point to obtain the first pixel pointThe Laplacian filtering weight of each pixel point is recorded as；
Presetting an initial Laplace convolution kernel; according to the firstThe Laplace filtering weight of each pixel point weights the initial Laplace convolution kernel to obtain the second pixel pointA convolution kernel of Laplace filtering is carried out on each pixel point;
according to the pair of Convolution kernel of Laplace filtering is carried out on each pixel point, and the first pixel point is subjected toAnd carrying out Laplacian filtering on each pixel point.
4. The method for enhancing liver cancer pathology image based on artificial intelligence according to claim 1, wherein the method for acquiring liver region in CT image comprises the following specific steps:
and acquiring a liver region in the CT image by using a semantic segmentation algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410704759.1A CN118297812B (en) | 2024-06-03 | 2024-06-03 | Liver cancer pathological image enhancement method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410704759.1A CN118297812B (en) | 2024-06-03 | 2024-06-03 | Liver cancer pathological image enhancement method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118297812A CN118297812A (en) | 2024-07-05 |
| CN118297812B true CN118297812B (en) | 2024-09-10 |
Family
ID=91687564
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410704759.1A Active CN118297812B (en) | 2024-06-03 | 2024-06-03 | Liver cancer pathological image enhancement method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118297812B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118674718B (en) * | 2024-08-22 | 2024-10-25 | 大石桥市中宁冶金矿产有限责任公司 | Visual detection method and system for refractory material processing quality |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117745751A (en) * | 2024-02-21 | 2024-03-22 | 中国人民解放军总医院第八医学中心 | Pulmonary tuberculosis CT image segmentation method based on feature extraction |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115187592B (en) * | 2022-09-08 | 2022-12-09 | 山东奥洛瑞医疗科技有限公司 | Image segmentation method suitable for nuclear magnetic resonance image |
| CN116883537B (en) * | 2023-09-06 | 2023-12-01 | 微山县人民医院 | Image enhancement-based common surgical image acquisition method |
-
2024
- 2024-06-03 CN CN202410704759.1A patent/CN118297812B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117745751A (en) * | 2024-02-21 | 2024-03-22 | 中国人民解放军总医院第八医学中心 | Pulmonary tuberculosis CT image segmentation method based on feature extraction |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118297812A (en) | 2024-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116993628B (en) | CT image enhancement system for tumor radio frequency ablation guidance | |
| CN118297812B (en) | Liver cancer pathological image enhancement method based on artificial intelligence | |
| CN117422628B (en) | Optimized enhancement method for cardiac vascular ultrasonic examination data | |
| Liu et al. | Automatic lung segmentation based on image decomposition and wavelet transform | |
| CN117764864B (en) | Nuclear magnetic resonance tumor visual detection method based on image denoising | |
| CN117237342B (en) | Intelligent analysis method for respiratory rehabilitation CT image | |
| CN117522719B (en) | Bronchoscope image auxiliary optimization system based on machine learning | |
| CN116630762A (en) | Multi-mode medical image fusion method based on deep learning | |
| CN118097160B (en) | Critical critical illness state monitoring system based on vision technology | |
| CN117830308B (en) | Intelligent contrast analysis method for angiography before and after interventional operation | |
| CN116128774B (en) | Stomach spiral CT data enhancement processing method | |
| CN113658193B (en) | Liver CT image tumor segmentation method based on information fusion | |
| CN116993764A (en) | Stomach CT intelligent segmentation extraction method | |
| CN116485819A (en) | Ear-nose-throat examination image segmentation method and system | |
| Mehmood et al. | Enhancement of CT images for visualization | |
| CN115861230A (en) | Method and system for determining lung region by thoracic cavity CT image segmentation | |
| CN114240893A (en) | Method for measuring and calculating spinal Cobb angle in external image | |
| CN118196123B (en) | Lung X-ray image processing method for lung fiber inhalation pathogenicity | |
| CN108921860B (en) | Full-automatic segmentation method for prostate magnetic resonance image | |
| CN118397031B (en) | Electronic medical record data analysis method for tuberculosis patients | |
| CN117974692B (en) | Ophthalmic medical image processing method based on region growing | |
| Hou et al. | Research on tumor image segmentation in medical imaging based on extremum adaptive median filtering | |
| CN118135335B (en) | Brain injury CT image optimization classification method based on machine learning | |
| CN116167949B (en) | Clinical auxiliary decision-making system based on medical image big data | |
| CN117952994B (en) | Cardiovascular CT image intelligent segmentation method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20240820 Address after: 110000 No. 155 North Nanjing street, Heping District, Liaoning, Shenyang Applicant after: THE FIRST HOSPITAL OF CHINA MEDICIAL University Country or region after: China Address before: Floor 2, Gate 7, No. 66 Nanjing South Street, Heping District, Shenyang City, Liaoning Province, 110000 Applicant before: Shenyang Hongqi Health Management Co.,Ltd. Country or region before: China |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |