CN113129314B - Intelligent image processing method for lung shadows - Google Patents
Intelligent image processing method for lung shadows Download PDFInfo
- Publication number
- CN113129314B CN113129314B CN202110359759.9A CN202110359759A CN113129314B CN 113129314 B CN113129314 B CN 113129314B CN 202110359759 A CN202110359759 A CN 202110359759A CN 113129314 B CN113129314 B CN 113129314B
- Authority
- CN
- China
- Prior art keywords
- lung
- image processing
- node
- processing method
- edge
- 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
- 210000004072 lung Anatomy 0.000 title claims abstract description 59
- 238000003672 processing method Methods 0.000 title claims abstract description 19
- 230000002685 pulmonary effect Effects 0.000 claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 17
- 206010056342 Pulmonary mass Diseases 0.000 claims abstract description 12
- 238000007635 classification algorithm Methods 0.000 claims abstract description 11
- 238000000605 extraction Methods 0.000 claims abstract description 11
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 238000000513 principal component analysis Methods 0.000 claims abstract description 9
- 239000011159 matrix material Substances 0.000 claims description 17
- 230000009466 transformation Effects 0.000 claims description 9
- 238000000034 method Methods 0.000 claims description 8
- 108050006002 RNA polymerase sigma factor FliA Proteins 0.000 claims description 7
- 230000000877 morphologic effect Effects 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 7
- 239000000853 adhesive Substances 0.000 claims description 6
- 230000001070 adhesive effect Effects 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 6
- 206010058467 Lung neoplasm malignant Diseases 0.000 description 12
- 201000005202 lung cancer Diseases 0.000 description 12
- 208000020816 lung neoplasm Diseases 0.000 description 12
- 206010028980 Neoplasm Diseases 0.000 description 3
- 238000005452 bending Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 238000001574 biopsy Methods 0.000 description 2
- 238000013276 bronchoscopy Methods 0.000 description 2
- 201000011510 cancer Diseases 0.000 description 2
- 238000002591 computed tomography Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 206010027476 Metastases Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005314 correlation function Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000002161 passivation Methods 0.000 description 1
- 231100000915 pathological change Toxicity 0.000 description 1
- 230000036285 pathological change Effects 0.000 description 1
- 230000001575 pathological effect Effects 0.000 description 1
- 230000007170 pathology Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 208000024891 symptom Diseases 0.000 description 1
Images
Classifications
-
- 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
-
- G06T5/70—
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- 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
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/28—Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/30061—Lung
- G06T2207/30064—Lung nodule
Abstract
The invention provides an intelligent image processing method for lung shadows, which comprises the following steps of S1: extracting an original lung image, preprocessing the original lung image, and then segmenting to obtain lung nodules; s2: carrying out feature extraction on pulmonary nodules to obtain feature parameters; s3: carrying out feature selection on pulmonary nodules by adopting a principal component analysis strategy; s4: and identifying the pulmonary nodules by adopting a classification algorithm to obtain a positive data set and a negative data set. According to the invention, the regional structure of the lung nodule adhered to the region outside the lung in the image display is captured and supplemented through the edge filling algorithm, so that the reliability of data in the lung image processing is improved.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an intelligent image processing method for lung shadows.
Background
At present, the death rate of lung cancer in cancer diseases is very high, and according to data counted in 2015, the total number of newly diagnosed lung cancer cases in China is about 78.7 thousands, about 630500 lung cancer patients die, which is equivalent to that 1700 lung cancer patients die averagely every day, the total number of lung cancer patients is at the top of the world, the incidence rate of lung cancer is gradually increased, and the health of the people in China is seriously harmed. In recent years, the accurate screening and staging accuracy of lung cancer is improved by the technical development of high-resolution computed tomography, PET-CT, bronchoscopy, percutaneous tumor puncture and the like. PET-CT is of great value not only for assessing the whole body, especially for possible distant metastases, but also for finding some lesions that are difficult to diagnose with CT. The pathological biopsy is the 'gold standard' for determining the cancer, the bronchoscopy has high value for biopsy or brush cytology examination of specimens in the central lung cancer, more than 95% of pathological changes can be confirmed through pathology, and the early screening of the lung cancer is very important.
The pulmonary nodules serve as basic characteristics of lung cancer focuses and are important basis for lung cancer screening, and with the continuous improvement of a CT image scanning technology, the problem that an image recognition processing technology is matched with the CT image scanning technology needs to be solved urgently, the method is particularly critical in the aspect of extracting lung feature integrity, because the edge extraction of lung images is influenced by high-density lung edge adhesion, the similarity of lung edge tissues is high, partial features can be omitted in a traditional edge extraction algorithm, and therefore the remained lung edges with feature parameters are eliminated, and the problem of data loss in the image processing process is caused.
Disclosure of Invention
In view of the above, the present invention provides an intelligent image processing method for lung shadows.
In order to solve the technical problems, the invention adopts the technical scheme that: an intelligent image processing method aiming at lung shadow comprises
S1: extracting an original lung image, preprocessing the original lung image, and then segmenting to obtain lung nodules;
s2: carrying out feature extraction on pulmonary nodules to obtain feature parameters;
s3: performing feature selection on pulmonary nodules by adopting a principal component analysis strategy;
s4: and identifying the pulmonary nodules by adopting a classification algorithm to obtain a positive data set and a negative data set.
In the present invention, preferably, the preprocessing of the lung image specifically includes the following steps:
s11: denoising the original lung image;
s12: selecting a proper threshold value to convert the gray level image into a binary image;
s13: the lung margins are detected and obtained.
In the present invention, it is preferable that whether the lung edge is closed in step S13 is determined, if yes, step S2 is performed, otherwise, the lung edge is repaired by using an edge filling algorithm, and then step S2 is performed.
In the present invention, preferably, the edge filling algorithm includes the following specific steps:
t1: discretizing and sampling the lung edges in the step S13 to obtain a plurality of nodes, and sequentially storing sampling results in a two-dimensional array a;
t2: carrying out stress analysis on each node, and calculating to obtain a resultant force sigma F corresponding to each node;
t3: judging whether the resultant force sigma F of the current node points to the inside of the edge, if so, storing the corresponding node into the array b, and continuing to obtain the next node; otherwise, judging whether the current node meets the node verification condition, if so, executing a step T4, otherwise, deleting the current node;
t4: storing the current corresponding node into an array b, judging whether the absolute value of the difference of the last ten elements of the array b is less than five, and if so, finishing the instruction; otherwise, step T3 is repeated.
In the present invention, preferably, the node verification condition is specifically that the resultant force ∑ F is smaller than the adhesive force F of the edge a The case (1).
In the present invention, preferably, the feature extraction of pulmonary nodules specifically includes the following steps:
s21: sequentially storing boundary points of pulmonary nodules;
s22: finding out key points on the boundary;
s23: respectively solving the attributes of triangles formed by the adjacent three points;
s24: the number of key points, minimum convex angle and minimum concave angle are extracted.
In the present invention, preferably, the feature parameters include morphological features, grayscale features, and texture features.
In the present invention, preferably, the principal component analysis strategy specifically includes the following steps:
selecting two observation variables X for N samples respectively 1 、X 2 Linear combination is carried out, and a rotation transformation matrix U is solved through coordinate transformation;
obtaining principal component Z from the eigenvector i ；
And estimating the correlation matrix and the total covariance by using the sample data, and converting the calculation principal component problem into an eigenvector and an eigenvalue for solving the covariance matrix of the sample.
In the present invention, preferably, the classification algorithm employs any one of a BP neural network algorithm, a fisher classification algorithm, or an SVM algorithm.
In the present invention, it is preferable that the denoising process employs any one of a morphological filter, a gaussian filter, or a median filter.
The invention has the advantages and positive effects that: pulmonary nodules are easily omitted in the process of executing the selected pulmonary edge, so that the detected pulmonary edge is not closed, the regional structure of the pulmonary nodules adhered to the region outside the lung in image display is captured and supplemented through an edge filling algorithm, a plurality of nodes are taken, the final pulmonary edge is determined by deleting the nodes belonging to the salient points, the parts of the nodules possibly connected with the region outside the lung can be supplemented to original data of image processing, the reliability of the data in the process of processing the pulmonary image is improved, and the problem of inaccurate edge extraction of the pulmonary image caused by adhesion of the pulmonary edge is solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings:
FIG. 1 is a flow chart of an intelligent image processing method for lung shadows according to the present invention;
FIG. 2 is a schematic diagram of a preprocessing step of an intelligent image processing method for lung shadows according to the present invention;
FIG. 3 is a schematic diagram of a preprocessing step of an intelligent image processing method for lung shadows according to the present invention;
FIG. 4 is a flowchart of feature extraction for pulmonary nodules according to an intelligent image processing method for pulmonary shadows provided by the present invention;
fig. 5 is a schematic diagram of a principal component analysis strategy of an intelligent image processing method for lung shadows according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
It will be understood that when an element is referred to as being "secured to" another element, it can be directly on the other element or intervening elements may also be present. When a component is referred to as being "connected" to another component, it can be directly connected to the other component or intervening components may also be present. When a component is referred to as being "disposed on" another component, it can be directly on the other component or intervening components may also be present. The terms "vertical," "horizontal," "left," "right," and the like as used herein are for illustrative purposes only.
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 terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
As shown in FIGS. 1 to 5, the present invention provides an intelligent image processing method for lung shadow, which comprises
S1: extracting an original lung image, preprocessing the original lung image, and then segmenting to obtain lung nodules;
s2: carrying out feature extraction on pulmonary nodules to obtain feature parameters;
s3: carrying out feature selection on pulmonary nodules by adopting a principal component analysis strategy;
s4: and identifying the pulmonary nodules by adopting a classification algorithm to obtain a positive data set and a negative data set, and detecting and classifying the pulmonary nodules by adopting a proper classifier according to the characteristics by adopting the classification algorithm.
In this embodiment, further, the preprocessing the lung image specifically includes the following steps:
s11: denoising the original lung image;
s12: selecting a proper threshold value to convert the gray level image into a binary image;
s13: the lung margins are detected and obtained.
In this embodiment, further, whether the lung edge is closed in step S13 is determined, if yes, step S2 is performed, otherwise, the lung edge is repaired by using an edge filling algorithm, and then step S2 is performed.
Since the lung nodules inside the lung may be connected to the regions outside the lung, it can be seen that the lung edges obtained by the preliminary detection are not closed, and some lung nodules are often lost, so that the open lung nodules need to be supplemented by using an edge filling algorithm.
In this embodiment, further, the edge filling algorithm includes the following specific steps:
t1: discretizing and sampling the lung edges in the step S13 to obtain a plurality of nodes, and sequentially storing sampling results in a two-dimensional array a;
t2: and (3) carrying out stress analysis on each node, calculating to obtain a resultant force sigma F corresponding to each node, marking the current node as i, respectively representing the previous node and the next node as i-1 and i +1, and respectively representing the interaction force between the i-1 and the i +1 as i-1 and i +1
And
then
Then a resultant force
T3: judging whether the resultant force sigma F of the current node points to the inside of the edge, if so, storing the corresponding node into the array b, and continuing to obtain the next node; otherwise, judging whether the current node meets the node verification condition, if so, executing a step T4, otherwise, deleting the current node;
t4: storing the current corresponding node into an array b, judging whether the absolute value of the difference of the last ten elements of the array b is less than five, and if so, finishing the instruction; otherwise, step T3 is repeated.
In the present embodiment, further, the node verification condition is specifically that the resultant force F is smaller than the adhesive force F of the edge a The case (1). When sigma F<F a When, it means that the current node is a concave point, it means that the node is on the boundary where the bending is large, i.e., the adhesive force F a Attracting the current node to prevent it from moving; when ≧ F a When the node is positioned on the boundary with a smaller bending radius, namely the adhesive force cannot attract the current node, the current node moves outwards under the action of the resultant force, and the bending radius is increased until the node stops moving. It can be seen that by comparing the magnitude of the resultant force and the adhesive force, which current nodes are pits can be judged, and further the determination of the lung edge is realized.
In this embodiment, further, the feature extraction of the pulmonary nodules specifically includes the following steps:
s21: sequentially storing boundary points of pulmonary nodules;
s22: finding out key points on the boundary;
s23: respectively solving the attributes of triangles formed by the adjacent three points;
s24: the number of key points, minimum convex angle and minimum concave angle are extracted.
Since the number of pulmonary nodule boundary keypoints can be evaluated for its curvature, the greater the number of boundary keypoints, the greater the curvature. The smallest angle among all the pits on the pulmonary nodule boundary is called the minimum pit angle, and the occurrence indicates the possible presence of lobulation. The minimum lobe is the smallest angle among all the salient points on the pulmonary nodule contour, indicatingBurrs may be present. The ratio of arc length to chord length is expressed as: r i Where C takes two integers of the same size but of opposite sign to indicate whether the pulmonary nodule boundary keypoint belongs to a convex arc segment or a concave arc segment. B selects a positive integer value to reflect the sharpness of the image boundary. The main role of the boundary clarity factor is to correct the influence of the original image boundary blurring on the contours obtained after segmentation. Each point on the boundary can calculate its respective R i 。
In the present embodiment, further, the feature parameters include morphological features, grayscale features, and texture features. According to the invention, the geometrical characteristics of the diameter, roundness, area, sphericity, compactness and eccentricity of the lung nodule in shape are utilized, and specifically, the conditions of leaf separation and burr in the lung nodule image are obtained through the geometrical characteristics, so that the potential lung cancer symptom is obtained. The lobe condition refers to the fact that a plurality of arcs with unevenness exist on the surface of a pulmonary nodule, so that the edge curvature of the pulmonary nodule is increased, and the burr condition refers to the fact that a straight, powerful, unbranched, radial, thin and short line image which extends from the edge of the pulmonary nodule to the periphery exists. Since the lung CT scanning image is a gray level image, the lung structure can be distinguished through gray level values among regions, and the gray level value standard deviation of pixel points inside lung nodules is obtained. The texture features are used for measuring the spatial information of the image within a certain range, mathematical statistics is generally adopted in the process of extracting the texture features, and the gray level distribution in the texture can be fully reflected by analyzing the texture features of the image by utilizing the gray level co-occurrence matrix, and particularly, the combination information including the change amplitude, the direction and the adjacent interval of the gray level of the image is included. When the texture features are represented as thicker, the corresponding energy values are larger, and conversely, when the texture features are thinner, the corresponding energy values are smaller.
In this embodiment, further, the principal component analysis strategy specifically includes the following steps:
w1: selecting two observation variables X for N samples respectively 1 、X 2 Linear combination is carried out, and a rotation transformation matrix U is solved through coordinate transformation;
w2: obtaining principal component Z from the eigenvector i ；
W3: and estimating the correlation matrix and the total covariance by using the sample data, and converting the calculation principal component problem into an eigenvector and an eigenvalue for solving the covariance matrix of the sample. The principal component analysis strategy is to regenerate a group of irrelevant features by computing and analyzing a plurality of features with certain relevance, and the new combined features can cover the original information quantity as much as possible. Generally speaking, the first comprehensive index F is selected from the original P characteristics 1 The variance is calculated, the more information covered the greater the variance. If F 1 Is the total linear combined inner variance (Var (F) 1 ) The largest group) is said to be the first principal component. When F is present 1 When the data information contained in the original characteristic combination is not sufficiently reflected, the second principal component F needs to be selected 2 . Because F 1 And F 2 Not related, so there is Cov [ F ] 1 ,F 2 ]0. At step W1 for two observed variables X 1 、X 2 When linear combination is carried out, p random variables X are selected 1 ，X 2 ，…，X p Definition of a linear combination of F 1 ，F 2 ，…，F k (k≤p)，F i The expression of (a) is:
the problem of finding p random variables is converted into the problem of finding linear combinations, linear combinations F i The following three conditions need to be met:
COV(F i ,F j ) This is for each principal component to contain no redundant information, 0, i ≠ j, i, j ═ 1,2, …, k;
Var(F 1 )≥Var(F 2 )≥…≥Var(F k ) The variance of the principal component decreases, i.e., the amount of information covered decreases.
Specifically, in the present embodiment, the number of NTaking two observed variables X in a sample 1 、X 2 Mapping N samples to coordinate systems of two observed variables, wherein the samples are along X 1 、X 2 The dispersion of (A) is expressed as Var (X) 1 ) And Var (X) 2 ) In order to ensure that the information amount is not distorted, two observation variables are linearly combined and subjected to coordinate transformation
Then, by
And U T ＝U -1 ,U T And obtaining a rotation transformation matrix U when U is I.
The sample mean value is subjected to standardized transformation according to the sample mean value to obtain a covariance matrix;
the correlation matrix is represented as R, resulting in
Wherein r is ij ＝x i ×x j I, j ═ 1,2, …, p. And (3) carrying out standardized transformation on the correlation matrix R to obtain a covariance matrix, estimating the correlation matrix and the total covariance by using sample data, and converting the calculation principal component problem into an eigenvector and an eigenvalue for solving the covariance matrix of the sample.
In this embodiment, further, the classification algorithm adopts any one of a BP neural network algorithm, a fisher classification algorithm, or an SVM algorithm. The SVM algorithm may specifically define a convex hull:
wherein a is i The co-correlation function is expressed, and the problem of linear inseparability is solved by setting a proper parameter mu.
In the present embodiment, further, the denoising process employs any one of a morphological filter, a gaussian filter, or a median filter. The Gaussian filter traverses all pixel points of the image through convolution, and the weighted average gray value of the pixel points in the neighborhood of the template is obtained by weighted averaging the value of each pixel point in the image and the value of the pixel point in the neighborhood of the pixel point in order to filter Gaussian noise. The principle of the median filter is that the average value of each point value in the neighborhood of the pixel point is used to replace the pixel value in the digital image or the sequence, so that the values of the surrounding pixel points are closer to the true value, and isolated noise points can be removed. The morphological filter is characterized in that a part of the graph, which is different from the structural element form, is filtered by open operation while a similar area is kept, and a part of the graph, which is different from the structural element form, is filled by closed operation, so that the efficient filtering is realized, and the image passivation condition is avoided.
The embodiments of the present invention have been described in detail, but the present invention is only the preferred embodiments of the present invention, and is not to be considered as limiting the scope of the present invention. All equivalent changes and modifications made within the scope of the present invention should be covered by the present patent.
Claims (7)
1. An intelligent image processing method for lung shadow is characterized by comprising
S1: extracting an original lung image, preprocessing the original lung image, and then segmenting to obtain lung nodules;
s2: carrying out feature extraction on pulmonary nodules to obtain feature parameters;
s3: performing feature selection on pulmonary nodules by adopting a principal component analysis strategy;
s4: identifying pulmonary nodules by adopting a classification algorithm to obtain a positive data set and a negative data set;
the preprocessing of the lung image specifically comprises the following steps:
s11: denoising the original lung image;
s12: selecting a proper threshold value to convert the gray level image into a binary image;
s13: detecting and obtaining lung margins;
judging whether the lung edge in the step S13 is closed, if so, executing a step S2, otherwise, adopting an edge filling algorithm to repair the lung edge, and then executing a step S2;
the edge filling algorithm comprises the following specific steps:
t1: discretizing and sampling the lung edges in the step S13 to obtain a plurality of nodes, and sequentially storing sampling results in a two-dimensional array a;
t2: carrying out stress analysis on each node, and calculating to obtain a resultant force sigma F corresponding to each node;
t3: judging whether the resultant force sigma F of the current node points to the inside of the edge one by one, if so, storing the corresponding node into the array b, and continuing to obtain the next node; otherwise, judging whether the current node meets the node verification condition, if so, executing a step T4, otherwise, deleting the current node;
t4: storing the current corresponding node into an array b, judging whether the absolute value of the difference of the last ten elements of the array b is less than five, and if so, finishing the instruction; otherwise, step T3 is repeated.
2. An intelligent image processing method for lung shadows according to claim 1, wherein the node verification condition is specifically that the resultant force ∑ F is smaller than the adhesive force F of the edge a The case (1).
3. The intelligent image processing method for lung shadows according to claim 1, wherein the feature extraction of lung nodules specifically comprises the following steps:
s21: sequentially storing boundary points of pulmonary nodules;
s22: finding out key points on the boundary;
s23: respectively solving the attributes of triangles formed by the adjacent three points;
s24: the number of key points, minimum convex angle and minimum concave angle are extracted.
4. An intelligent image processing method for lung shadows according to claim 1, wherein the feature parameters comprise morphological features, gray-scale features and texture features.
5. An intelligent image processing method for lung shadows according to claim 1, wherein the principal component analysis strategy specifically comprises the following steps:
w1: selecting two observation variables X for N samples respectively 1 、X 2 Linear combination is carried out, and a rotation transformation matrix U is solved through coordinate transformation;
w2: obtaining principal component Z from the eigenvector i ；
W3: and estimating the correlation matrix and the total covariance by using the sample data, and converting the calculation principal component problem into an eigenvector and an eigenvalue for solving the covariance matrix of the sample.
6. An intelligent image processing method for lung shadows according to claim 1, wherein the classification algorithm adopts any one of a BP neural network algorithm, a fisher classification algorithm or an SVM algorithm.
7. An intelligent image processing method for lung shadows according to claim 1, wherein the denoising process employs any one of a morphological filter, a gaussian filter or a median filter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110359759.9A CN113129314B (en) | 2021-04-01 | 2021-04-01 | Intelligent image processing method for lung shadows |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110359759.9A CN113129314B (en) | 2021-04-01 | 2021-04-01 | Intelligent image processing method for lung shadows |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113129314A CN113129314A (en) | 2021-07-16 |
| CN113129314B true CN113129314B (en) | 2022-08-02 |
Family
ID=76775489
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110359759.9A Active CN113129314B (en) | 2021-04-01 | 2021-04-01 | Intelligent image processing method for lung shadows |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113129314B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108334883A (en) * | 2018-01-29 | 2018-07-27 | 南京信息工程大学 | A kind of optimization algorithm for extracting independent sift key points based on principal component analysis |
| CN109035227A (en) * | 2018-07-13 | 2018-12-18 | 哈尔滨理工大学 | The system that lung tumors detection and diagnosis is carried out to CT image |
| CN109300113A (en) * | 2018-08-28 | 2019-02-01 | 东北大学 | A kind of Lung neoplasm assisted detection system and method based on improvement Convex Hull Method |
| CN109389590A (en) * | 2017-09-28 | 2019-02-26 | 上海联影医疗科技有限公司 | Colon image data processing system and method |
| CN110133741A (en) * | 2019-05-08 | 2019-08-16 | 常州大学 | A kind of automatic safety inspection method of multi-angle of view matching and tracking |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105976367B (en) * | 2016-04-29 | 2019-06-28 | 上海联影医疗科技有限公司 | Image partition method, pulmonary nodule detection method and its computer-aided detection system |
| CN108335277A (en) * | 2018-03-05 | 2018-07-27 | 北京中科嘉宁科技有限公司 | A kind of image recess method for repairing and mending based on double Graham scanning methods |
-
2021
- 2021-04-01 CN CN202110359759.9A patent/CN113129314B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109389590A (en) * | 2017-09-28 | 2019-02-26 | 上海联影医疗科技有限公司 | Colon image data processing system and method |
| CN108334883A (en) * | 2018-01-29 | 2018-07-27 | 南京信息工程大学 | A kind of optimization algorithm for extracting independent sift key points based on principal component analysis |
| CN109035227A (en) * | 2018-07-13 | 2018-12-18 | 哈尔滨理工大学 | The system that lung tumors detection and diagnosis is carried out to CT image |
| CN109300113A (en) * | 2018-08-28 | 2019-02-01 | 东北大学 | A kind of Lung neoplasm assisted detection system and method based on improvement Convex Hull Method |
| CN110133741A (en) * | 2019-05-08 | 2019-08-16 | 常州大学 | A kind of automatic safety inspection method of multi-angle of view matching and tracking |
Non-Patent Citations (1)
| Title |
|---|
| 基于改进凸包算法的肺实质分割研究;李金等;《中国生物医学工程学报》;20130831;第484-490页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113129314A (en) | 2021-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1593094B1 (en) | Image analysis for the purpose of assessing cancer | |
| CN110443128B (en) | Finger vein identification method based on SURF feature point accurate matching | |
| CN107341812B (en) | A kind of sequence Lung neoplasm image partition method based on super-pixel and Density Clustering | |
| CN108537751B (en) | Thyroid ultrasound image automatic segmentation method based on radial basis function neural network | |
| CN109410238B (en) | Wolfberry identification and counting method based on PointNet + + network | |
| CN111598875A (en) | Method, system and device for building thyroid nodule automatic detection model | |
| CN110706225B (en) | Tumor identification system based on artificial intelligence | |
| Naresh et al. | Early detection of lung cancer using neural network techniques | |
| CN113706434B (en) | Post-processing method for chest enhancement CT image based on deep learning | |
| CN110598030A (en) | Oracle bone rubbing classification method based on local CNN framework | |
| CN112862808A (en) | Deep learning-based interpretability identification method of breast cancer ultrasonic image | |
| WO2013019856A1 (en) | Automated malignancy detection in breast histopathological images | |
| US8983154B2 (en) | Method for identifying colon polyps in a computed tomography colonography dataset | |
| Zambal et al. | Shape and appearance models for automatic coronary artery tracking | |
| CN111340770A (en) | Method for constructing cancer prognosis model by combining global weighted LBP (local binary pattern) and texture analysis | |
| CN116630971B (en) | Wheat scab spore segmentation method based on CRF_Resunate++ network | |
| CN114155202A (en) | Thyroid nodule ultrasonic image classification method based on feature fusion and transfer learning | |
| Chen et al. | Segmentation of overlapping cervical cells with mask region convolutional neural network | |
| CN114140445B (en) | Breast cancer pathological image identification method based on key attention area extraction | |
| CN109934102B (en) | Finger vein identification method based on image super-resolution | |
| CN111860500B (en) | Shoe stamp wearing area detection and edge tracing method | |
| CN113129314B (en) | Intelligent image processing method for lung shadows | |
| CN116664932A (en) | Colorectal cancer pathological tissue image classification method based on active learning | |
| CN114862883A (en) | Target edge extraction method, image segmentation method and system | |
| CN111415350B (en) | Colposcope image identification method for detecting cervical lesions |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20230626 Address after: Room 219, Building 8, No. 41 Jinlang Road, Langxia Town, Jinshan District, Shanghai, 200000 Patentee after: Shanghai Hanhu Intelligent Technology Co.,Ltd. Address before: No.85 Wujin Road, Hongkou District, Shanghai Patentee before: SHANGHAI FIRST PEOPLE'S Hospital |
|
| TR01 | Transfer of patent right |