CN110136112A - One kind is based on mammary X-ray photography calcification computer aided detection algorithm - Google Patents
One kind is based on mammary X-ray photography calcification computer aided detection algorithm Download PDFInfo
- Publication number
- CN110136112A CN110136112A CN201910397979.3A CN201910397979A CN110136112A CN 110136112 A CN110136112 A CN 110136112A CN 201910397979 A CN201910397979 A CN 201910397979A CN 110136112 A CN110136112 A CN 110136112A
- Authority
- CN
- China
- Prior art keywords
- image
- calcification
- pixel
- mammary
- cluster
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
- G06T7/11—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
- G06T2207/10116—X-ray image
-
- 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
- G06T2207/20081—Training; Learning
-
- 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
- G06T2207/30068—Mammography; Breast
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The present invention provides one kind based on mammary X-ray photography calcification computer aided detection algorithm, comprising the following steps: reads in image, pretreatment, image filtering, image equilibration, image segmentation, gray scale rendition, LBP and GLDM textural characteristics and random forests algorithm classification;The verification and measurement ratio of Breast Calcifications both can be improved by the method for image procossing by the present invention, accurately represent the azimuth information of Breast Calcifications, improve the accuracy of Breast Calcifications detection, the present invention can be used for being quickly detected from suspicious calcified regions from breast molybdenum target target line image, pass through the method for Computer Image Processing, suspicious calcified regions are marked for reference, provide advantageous information for last good pernicious Distinguishing diagnosis.
Description
[technical field]
Present invention relates particularly to one kind based on mammary X-ray photography calcification computer aided detection algorithm.
[background technique]
Breast cancer is one of the most common malignant tumors in women, annual new breast cancer case about 27.4W in the whole nation, Zhan Nv
Property Incidence 16%, rank the first in female tumor, the key of prevention and treatment be early diagnose and early treatment, it is micro-
Calcification is that breast cancer is most important, most apparent pathological sign, and especially for the breast cancer of early stage " invisible ", Microcalcification is usual
It is very important reference index, is also the important intermediate result information of early-stage breast cancer.But generally very due to Microcalcification
It is small, therefore often cause to be easy the problem of failing to pinpoint a disease in diagnosis since radiologist's human eye is difficult to find.In order to mitigate doctor's rule of thumb meat
The time and efforts of eye label, the present invention can be used for quickly and easily detecting suspicious calcified regions from breast molybdenum target image,
By the method for Computer Image Processing, suspicious calcified regions are marked and are referred to for doctor, good pernicious sentenced for doctor is last
Average information Zhen Duan be provided.It proposes to be directed to molybdenum target X-ray Breast Calcifications cluster aided detection algorithms, provides second with reference to meaning for doctor
See.
In recent years, experts and scholars start to study Breast Calcifications image procossing.Currently, mammography X is examined
The research method for surveying calcification point is broadly divided into three classes: (1) based on image enchancing method, McLoughlin proposes one kind and passes through letter
Single ground gray level square root establishes quantum noise model, improves local contrast, realizes the detection of Breast Calcifications;(2) it is based on
The method of multi-resolution decomposition, Wang etc. propose it is a kind of extract high-frequency signal and noise using error image, recycle small echo contracting
Subtract and noise is inhibited, last neural network classification algorithm extracts Microcalcification information;(3) based on the method for machine learning,
Issam et al. is detected, with determination using there is the study of supervision to detect image segmentation with every piece of region of the SVM to image
It whether include microcalcifications.
[summary of the invention]
The technical problem to be solved in the present invention is to provide a kind of based on mammary X-ray photography calcification computer aided detection
Algorithm, for solving the problems, such as that center Breast Calcifications detection accuracy is low.
The present invention is implemented as follows: a kind of be based on mammary X-ray photography calcification computer aided detection algorithm, including following
Step:
(1) it reads in image: reading the calcification lesion picture of mammary X-ray photography detection and normally without calcification picture;
(2) it pre-processes: picture being pre-processed, mammary region is extracted, obtains image;
(3) image filtering: being decomposed using contourlet transform, and image is carried out mathematical morphology transformation;
(4) image equilibration: the brightness of brightness vicinity in image is increased using partial histogram equalization algorithm
By force, the brightness deterioration near brightness minimum, promotes the overall contrast of image;
(5) image segmentation: carrying out more mean iteratives of breast tissue with K-means algorithm to carry out body of gland segmentation,
Extract calcification area-of-interest and breast edge contour;
(6) it gray scale rendition: extracts calcification area-of-interest and breast edge contour carries out and operation, it is interested to obtain calcification
The image detection of Breast Calcifications is completed in region contour position;
(7) LBP and GLDM textural characteristics: cream is carried out using two textural characteristics of local binary patterns and gray level co-occurrence matrixes
Detection is extracted in the true and false positive calcification of gland;
(8) random forests algorithm is classified: completing true and false positive calcification classification with random forests algorithm.
Further, the specific operation method is as follows for the filtering image:
(3a) carries out multi-resolution decomposition to image by pyramid transform, to capture horizontally and vertically unusual
Point reduces dimension and completes multiscale analysis;
(3b) with anisotropic filter group will be distributed over the singular point in same direction synthesize coefficient complete it is multi-direction
Property analysis;
(3c) carries out opening operation Processing Algorithm to the filtered image of anisotropic filter group with structural element;
(3d) will obtain image and carry out contourlet transform reconstruction.
Further, the specific operation method is as follows for described image equalization:
(4a) defines a rectangular sub blocks on the image;
(4b) utilizes the histogram information of the rectangular sub blocks, carries out pixels statistics to the pixel at rectangular sub blocks center;
(4c) Gray Histogram grade occur number of pixels divided by pixel sum;
(4d) moves at rectangular sub blocks center pixel-by-pixel, and repeats the above 4b-4c process, until the institute of traversal input picture
There is pixel.
Further, described image segmentation particular content is as follows:
The number 4 of (5a) input cluster;
(5b) divides equally pixel as initial cluster from the pixel coverage of image;
Each pixel is assigned to cluster most like in initial cluster according to the mean value of pixel in initial cluster by (5c);
(5d) counts pixel pixel value summation and each cluster pixel number, and the two, which is divided by, to be obtained mean value and update initial cluster,
New mean value as cluster;
(5e) if the new mean value of cluster changes with the former mean value of the cluster, repeatedly step 5c-5e, until the mean value of cluster is not
It changes again, exports the mean value of 4 clusters;
(5f) finds calcification position according to highest mean value, and extracts calcification area-of-interest;Cream is carried out according to minimum mean value
The segmentation of room edge contour.
Further, the specific operation method is as follows for the step (7):
(7a) divides an image into fritter;
(7b) is by pixel centered on each pixel in fritter, 8 neighbours more adjacent thereto;
(7c) writes 0 if the neighbour value occupied of the value of center pixel is big;If it is not, then writing 1, one 8 numbers are constructed;
(7d) calculates the frequency histogram of each " number " appearance in fritter;
(7e) standardizes to histogram;
(7f) connects the histogram of all fritters, forms the LBP feature vector of entire window;
(7g) calculates image in 0 degree, 45 degree, 90 degree and 135 degree of energy, entropy, the degree of correlation, right using gray level co-occurrence matrixes
Than four characteristic angles of degree.
Further, the random forests algorithm classification includes following content:
(8a) defines an original training set S, puts back to sampling at random and selects M sample every time, carries out n times sampling altogether, raw
At N number of training set, N number of decision-tree model is respectively trained;
(8b) is directed to single decision-tree model, selects the feature of predetermined amount every time, calculates the Gini coefficient of feature, selection is most
An excellent feature is divided as the left and right subtree of decision tree nodes;
(8c) each tree is gone down all in accordance with the method division of (8b), until the node reaches termination condition, and each tree
It will predict that the M sample belongs to certain one kind, each tree is grown to the maximum extent, the not beta pruning in the fission process of decision tree;
More N decision trees of generation are formed random forest by (8d), choose final classification knot in a vote by N Tree Classifier
Fruit.
The present invention has the advantages that the verification and measurement ratio of Breast Calcifications both can be improved, the side of Breast Calcifications is accurately represented
Position information, improves the accuracy of Breast Calcifications detection, and the present invention can be used for from mammary gland aluminium target line image being quickly detected from
Suspicious calcified regions, and suspicious calcified regions are marked.
[specific embodiment]
One kind is based on mammary X-ray photography calcification computer aided detection algorithm, comprising the following steps:
(1) it reads in image: reading the calcification lesion picture of mammary X-ray photography detection and normally without calcification picture;
(2) it pre-processes: picture being pre-processed, mammary region is extracted, obtains image;
Remove the personal information in relation to patient, label and the background areas such as name, ID card No., the telephone number of patient
Domain be typically all in the fixed position in the image upper left corner or the upper right corner, Breast Calcifications point region be necessarily present in mammary region it
It is interior, so label and background area is only needed to shear, mammary region can be extracted;
(3) it image filtering: is decomposed using contourlet transform (Contourlet transform), by image (band logical subgraph
Picture) carry out mathematical morphology transformation;Specifically:
(3a) carries out multi-resolution decomposition to image by pyramid transform (Laplacian Pyramid), to capture level side
To the singular point with vertical direction, reduces dimension and complete multiscale analysis;
(3b) will be distributed over unusual in same direction with anisotropic filter group (Directional filter banks)
Point synthesizes a coefficient and completes multidirectional analysis;
(3c) carries out opening operation Processing Algorithm to the filtered image of anisotropic filter group with structural element;
(3d) will obtain image and carry out contourlet transform reconstruction.
(4) image equilibration: the brightness of brightness vicinity in image is increased using partial histogram equalization algorithm
By force, the brightness deterioration near brightness minimum, promotes the overall contrast of image;Specifically:
(4a) defines a rectangular sub blocks on the image;
(4b) utilizes the histogram information of the rectangular sub blocks, carries out pixels statistics to the pixel at rectangular sub blocks center;
(4c) Gray Histogram grade occur number of pixels (i.e. frequency) divided by pixel sum;
(4d) moves at rectangular sub blocks center pixel-by-pixel, and repeats the above 4b-4c process, until the institute of traversal input picture
There is pixel.
This method not only makes image local details obtain sufficient contrast enhancing, while eliminating blocking artifact due to sub-block
Balanced total degree is equal to the sum of all pixels of input picture.
(5) image segmentation: carrying out more mean iteratives of breast tissue with K-means algorithm to carry out body of gland segmentation,
Extract calcification area-of-interest (region of interest) and breast edge contour;Specifically:
The number 4 of (5a) input cluster;
(5b) divides equally pixel as initial cluster from the pixel coverage of image;
Each pixel is assigned to cluster most like in initial cluster according to the mean value of pixel in initial cluster by (5c);
(5d) counts pixel pixel value summation and each cluster pixel number, and the two, which is divided by, to be obtained mean value and update initial cluster,
New mean value as cluster;
(5e) if the new mean value of cluster changes with the former mean value of the cluster, repeatedly step 5c-5e, until the mean value of cluster is not
It changes again, exports the mean value of 4 clusters;
(5f) finds calcification position according to highest mean value, and extracts calcification area-of-interest;Cream is carried out according to minimum mean value
The segmentation of room edge contour.
(6) it gray scale rendition: extracts calcification area-of-interest and breast edge contour carries out and operation, it is interested to obtain calcification
The image detection of Breast Calcifications is completed in region contour position;
(7) local binary patterns (Local Binary Patterns, LBP) and ash LBP and GLDM textural characteristics: are utilized
It spends co-occurrence matrix (Gray level co-occurrence matrix, GLCM) two textural characteristics and carries out the true and false positive of mammary gland
Detection is extracted in calcification;Specifically:
(7a) divides an image into fritter;
(7b) by pixel centered on each pixel in fritter, 8 neighbours more adjacent thereto (upper left, it is left, it is left
Under, upper right, etc.);
(7c) writes 0 if the neighbour value occupied of the value of center pixel is big;If it is not, then writing 1, one 8 numbers are constructed;
(7d) calculates the frequency histogram of each " number " appearance in fritter;
(7e) standardizes to histogram;
(7f) connects (standardization) histogram of all fritters, forms the LBP feature vector of entire window;
(7g) calculates image in 0 degree, 45 degree, 90 degree and 135 degree of energy, entropy, the degree of correlation, right using gray level co-occurrence matrixes
Than four characteristic angles of degree.
Since co-occurrence matrix is usually larger and sparse, the various measurements of matrix are generallyd use to obtain more useful spy
Collection;
Wherein, being extracted using gray level co-occurrence matrixes includes the textural characteristics such as energy, entropy, contrast, the degree of correlation;
(1) LBP textural characteristics:
Wherein ix,yIt is the coordinate with the center pixel of gray value, ipIndicate that the gray value of surrounding P pixel, S are a symbols
Number function, is specifically defined are as follows:
The two functions are calculated to obtain binary system LBP value, and LBP characteristic value is retouched by the vector comprising ten features
It states, gray level co-occurrence matrixes include the following:
(2) energy also becomes angular second moment (Angular Second Moment, ASM)
(3) entropy (Entropy, ENT)
(4) degree of correlation (CORRLN)
f3=[∑i∑j((ij)P(i,j))-μxμy]/σxσy
(5) contrast (CON)
f4=∑i∑j(i-j)2P(i,j)
(8) random forests algorithm is classified: completing true and false positive calcification classification with random forests algorithm;Specifically:
(8a) defines an original training set S, puts back to sampling at random and selects M sample every time, carries out n times sampling altogether, raw
At N number of training set, N number of decision-tree model is respectively trained;
(8b) is directed to single decision-tree model, it is assumed that the number of training sample feature is X, and decision tree training is one or so
Subtree division process, every time select predetermined amount feature, calculate the Gini coefficient of feature, select an optimal feature as
The left and right subtree of decision tree nodes divides;
(8c) each tree is gone down all in accordance with the method division of (8b), until the node reaches termination condition, and each tree
It will predict that the M sample belongs to certain one kind, each tree is grown to the maximum extent, the not beta pruning in the fission process of decision tree;
More N decision trees of generation are formed random forest by (8d), choose final classification knot in a vote by N Tree Classifier
Fruit.
Inventive algorithm has been over the accuracy rate of the mankind, contact mammary X-ray photography naked eyes diagnosis one in some aspects
The accuracy rate of the low seniority resident 1 in year is 80%, the low seniority of contact mammary X-ray photography naked eyes diagnosis about 5 years cures mainly doctor
The accuracy rate of teacher 2 is 85%.As shown in Table 1, RF algorithm can distinguish normal sample and calcification sample well, detection
Accurate rate has respectively reached 90%, 81.5% and 87.5%.Result explanation, using based on traditional Machine learning classifiers
Area of computer aided Breast Calcifications lesion detection algorithm can obtain good true positive rate, and have lower false positive rate.
1 computer aided algorithm of table and the comparison of doctor's naked eyes calcification testing result
Method | Sen | Spec | Acc |
Our | 0.8360 | 0.9110 | 0.9000 |
Doctor 1 | 0.5000 | 0.8000 | 0.8000 |
Doctor 2 | 0.5200 | 0.8889 | 0.8500 |
The present invention use by breast image carry out multi-scale filtering technology denoising and adaptive local histogram it is equal
Weighing apparatusization enhances the contrast metric of mammogram.Secondly, being changed by KNN algorithm to carry out the multi-threshold of breast tissue
Generation;Finally, calcification and normal picture is detected according to gray scale and textural characteristics, pass through LBP, two statistical nature benefits of GLDM
The performance of assessment breast calcification detection algorithm is compared to random forest method.Breast Calcifications both can be improved in the present invention
Verification and measurement ratio accurately represents the azimuth information of Breast Calcifications, improves the accuracy of Breast Calcifications detection, and the present invention can be used for
Suspicious calcified regions are quickly detected from from mammary gland aluminium target line image, and suspicious calcified regions are marked.
Claims (6)
1. one kind is based on mammary X-ray photography calcification computer aided detection algorithm, it is characterised in that: the following steps are included:
(1) it reads in image: reading the calcification lesion picture of mammary X-ray photography detection and normally without calcification picture;
(2) it pre-processes: picture being pre-processed, mammary region is extracted, obtains image;
(3) image filtering: being decomposed using contourlet transform, and image is carried out mathematical morphology transformation;
(4) image equilibration: being enhanced the brightness of brightness vicinity in image using partial histogram equalization algorithm, bright
The brightness deterioration near minimum is spent, the overall contrast of image is promoted;
(5) image segmentation: more mean iteratives of breast tissue are carried out with K-means algorithm to carry out body of gland segmentation, extracted
Calcification area-of-interest and breast edge contour;
(6) it gray scale rendition: extracts calcification area-of-interest and breast edge contour carries out and operation, obtain calcification area-of-interest
Outline position completes the image detection of Breast Calcifications;
(7) LBP and GLDM textural characteristics: it is true that mammary gland is carried out using two textural characteristics of local binary patterns and gray level co-occurrence matrixes
Detection is extracted in false positive calcification;
(8) random forests algorithm is classified: completing true and false positive calcification classification with random forests algorithm.
2. as described in claim 1 a kind of based on mammary X-ray photography calcification computer aided detection algorithm, it is characterised in that:
The specific operation method is as follows for the filtering image:
(3a) carries out multi-resolution decomposition to image by pyramid transform, to capture singular point horizontally and vertically, drop
Low dimensional completes multiscale analysis;
(3b) will be distributed over the singular point in same direction with anisotropic filter group and synthesize a coefficient completion multidirectional point
Analysis;
(3c) carries out opening operation Processing Algorithm to the filtered image of anisotropic filter group with structural element;
(3d) will obtain image and carry out contourlet transform reconstruction.
3. as described in claim 1 a kind of based on mammary X-ray photography calcification computer aided detection algorithm, it is characterised in that:
The specific operation method is as follows for described image equalization:
(4a) defines a rectangular sub blocks on the image;
(4b) utilizes the histogram information of the rectangular sub blocks, carries out pixels statistics to the pixel at rectangular sub blocks center;
(4c) Gray Histogram grade occur number of pixels divided by pixel sum;
(4d) moves at rectangular sub blocks center pixel-by-pixel, and repeats the above 4b-4c process, until all pictures of traversal input picture
Element.
4. as claimed in claim 3 a kind of based on mammary X-ray photography calcification computer aided detection algorithm, it is characterised in that:
It is as follows that described image divides particular content:
The number 4 of (5a) input cluster;
(5b) divides equally pixel as initial cluster from the pixel coverage of image;
Each pixel is assigned to cluster most like in initial cluster according to the mean value of pixel in initial cluster by (5c);
(5d) counts pixel pixel value summation and each cluster pixel number, and the two, which is divided by, to be obtained mean value and update initial cluster, as
The new mean value of cluster;
(5e) if the new mean value of cluster changes with the former mean value of the cluster, repeatedly step 5c-5e, until the mean value of cluster is no longer sent out
Changing exports the mean value of 4 clusters;
(5f) finds calcification position according to highest mean value, and extracts calcification area-of-interest;Breast side is carried out according to minimum mean value
Edge contours segmentation.
5. as described in claim 1 a kind of based on mammary X-ray photography calcification computer aided detection algorithm, it is characterised in that:
The specific operation method is as follows for the step (7):
(7a) divides an image into fritter;
(7b) is by pixel centered on each pixel in fritter, 8 neighbours more adjacent thereto;
(7c) writes 0 if the neighbour value occupied of the value of center pixel is big;If it is not, then writing 1, one 8 numbers are constructed;
(7d) calculates the frequency histogram of each " number " appearance in fritter;
(7e) standardizes to histogram;
(7f) connects the histogram of all fritters, forms the LBP feature vector of entire window;
(7g) calculates energy, entropy, the degree of correlation, contrast of the image at 0 degree, 45 degree, 90 degree and 135 degree using gray level co-occurrence matrixes
Four characteristic angles.
6. as described in claim 1 a kind of based on mammary X-ray photography calcification computer aided detection algorithm, it is characterised in that:
The random forests algorithm classification includes following content:
(8a) defines an original training set S, puts back to sampling at random and selects M sample every time, carry out n times sampling altogether, generates N number of
N number of decision-tree model is respectively trained in training set;
(8b) is directed to single decision-tree model, selects the feature of predetermined amount every time, calculates the Gini coefficient of feature, select optimal
One feature is divided as the left and right subtree of decision tree nodes;
(8c) each tree is gone down all in accordance with the method division of (8b), and until the node reaches termination condition, and each tree all can
Predict that the M sample belongs to certain one kind, each tree is grown to the maximum extent, the not beta pruning in the fission process of decision tree;
More N decision trees of generation are formed random forest by (8d), choose final classification result in a vote by N Tree Classifier.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910397979.3A CN110136112B (en) | 2019-05-14 | 2019-05-14 | Computer-aided detection algorithm based on mammary gland X-ray photography calcification |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910397979.3A CN110136112B (en) | 2019-05-14 | 2019-05-14 | Computer-aided detection algorithm based on mammary gland X-ray photography calcification |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110136112A true CN110136112A (en) | 2019-08-16 |
CN110136112B CN110136112B (en) | 2022-11-04 |
Family
ID=67573694
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910397979.3A Active CN110136112B (en) | 2019-05-14 | 2019-05-14 | Computer-aided detection algorithm based on mammary gland X-ray photography calcification |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110136112B (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110648335A (en) * | 2019-09-20 | 2020-01-03 | 俞喆祺 | Low-calculation method for assisting detection of esophageal adenocarcinoma and colorectal cancer |
CN110766704A (en) * | 2019-10-22 | 2020-02-07 | 深圳瀚维智能医疗科技有限公司 | Breast point cloud segmentation method, device, storage medium and computer equipment |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5598481A (en) * | 1994-04-29 | 1997-01-28 | Arch Development Corporation | Computer-aided method for image feature analysis and diagnosis in mammography |
US20090041326A1 (en) * | 2007-08-06 | 2009-02-12 | Shoupu Chen | Line structure detection and analysis for mammography cad |
CN104732213A (en) * | 2015-03-23 | 2015-06-24 | 中山大学 | Computer-assisted lump detecting method based on mammary gland magnetic resonance image |
CN107392204A (en) * | 2017-07-20 | 2017-11-24 | 东北大学 | A kind of galactophore image microcalcifications automatic checkout system and method |
CN107958453A (en) * | 2017-12-01 | 2018-04-24 | 深圳蓝韵医学影像有限公司 | Detection method, device and the computer-readable storage medium of galactophore image lesion region |
CN109146848A (en) * | 2018-07-23 | 2019-01-04 | 东北大学 | A kind of area of computer aided frame of reference and method merging multi-modal galactophore image |
-
2019
- 2019-05-14 CN CN201910397979.3A patent/CN110136112B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5598481A (en) * | 1994-04-29 | 1997-01-28 | Arch Development Corporation | Computer-aided method for image feature analysis and diagnosis in mammography |
US20090041326A1 (en) * | 2007-08-06 | 2009-02-12 | Shoupu Chen | Line structure detection and analysis for mammography cad |
CN104732213A (en) * | 2015-03-23 | 2015-06-24 | 中山大学 | Computer-assisted lump detecting method based on mammary gland magnetic resonance image |
CN107392204A (en) * | 2017-07-20 | 2017-11-24 | 东北大学 | A kind of galactophore image microcalcifications automatic checkout system and method |
CN107958453A (en) * | 2017-12-01 | 2018-04-24 | 深圳蓝韵医学影像有限公司 | Detection method, device and the computer-readable storage medium of galactophore image lesion region |
CN109146848A (en) * | 2018-07-23 | 2019-01-04 | 东北大学 | A kind of area of computer aided frame of reference and method merging multi-modal galactophore image |
Non-Patent Citations (8)
Title |
---|
刘怡: "基于Contourlet变换的纹理图像分割方法研究", 《大众商务》 * |
彭庆涛 等: "基于小波分析和灰度纹理特征的乳腺X线图像微钙化点区域的提取", 《北京生物医学工程》 * |
王瑞平 等: "独立分量分析在乳腺钼靶X片感兴趣区域提取中的应用", 《中国生物医学工程学报》 * |
程运福 等: "基于乳腺X线图像的微钙化点区域自动检测算法研究", 《中国医学物理学杂志》 * |
蔡雅丽 等: "计算机辅助***在乳腺钙化性病变X线摄影诊断中的应用", 《中国医学影像学杂志》 * |
谷宇 等: "基于NSCT和CLAHE的乳腺钼靶X线图像微钙化点增强方法", 《光学技术》 * |
郝欣 等: "基于医学图像内容检索的计算机辅助乳腺X线影像诊断技术", 《中国生物医学工程学报》 * |
钟明霞: "基于形态滤波和Sobel算子的微钙化簇边缘检测算法", 《计算机时代》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110648335A (en) * | 2019-09-20 | 2020-01-03 | 俞喆祺 | Low-calculation method for assisting detection of esophageal adenocarcinoma and colorectal cancer |
CN110766704A (en) * | 2019-10-22 | 2020-02-07 | 深圳瀚维智能医疗科技有限公司 | Breast point cloud segmentation method, device, storage medium and computer equipment |
CN110766704B (en) * | 2019-10-22 | 2022-03-08 | 深圳瀚维智能医疗科技有限公司 | Breast point cloud segmentation method, device, storage medium and computer equipment |
Also Published As
Publication number | Publication date |
---|---|
CN110136112B (en) | 2022-11-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Tai et al. | An automatic mass detection system in mammograms based on complex texture features | |
Al-Shamlan et al. | Feature extraction values for breast cancer mammography images | |
Biswas et al. | Mammogram classification using gray-level co-occurrence matrix for diagnosis of breast cancer | |
CN107358267B (en) | Mammary gland ultrasonic image multi-element classification system and method based on cross-correlation characteristics | |
Beheshti et al. | Classification of abnormalities in mammograms by new asymmetric fractal features | |
Das et al. | A fast and automated segmentation method for detection of masses using folded kernel based fuzzy c-means clustering algorithm | |
Saki et al. | A novel opposition-based classifier for mass diagnosis in mammography images | |
Bose et al. | Detection of microcalcification in mammograms using soft computing techniques | |
CN110136112A (en) | One kind is based on mammary X-ray photography calcification computer aided detection algorithm | |
Fazilov et al. | Improvement of Image Enhancement Technique for Mammography Images | |
Pashoutan et al. | Automatic breast tumor classification using a level set method and feature extraction in mammography | |
CN104331864B (en) | Based on the processing of the breast image of non-down sampling contourlet and the significant model of vision | |
Rizzi et al. | A supervised method for microcalcification cluster diagnosis | |
Krishnaveni et al. | Study of Mammogram Microcalcification to aid tumour detection using Naive Bayes Classifier | |
Johny et al. | Breast cancer detection in mammogram using fuzzy C-means and random forest classifier | |
Sehrawat et al. | Detection and classification of tumor in mammograms using discrete wavelet transform and support vector machine | |
Arpana et al. | Feature extraction values for digital mammograms | |
CN109377461A (en) | A kind of breast X-ray image self-adapting enhancement method based on NSCT | |
Rizzi et al. | A fully automatic system for detection of breast microcalcification clusters | |
Dolejší et al. | Automatic two-step detection of pulmonary nodules | |
Htay et al. | Analysis on Results Comparison of Feature Extraction Methods for Breast Cancer Classification | |
Lu et al. | 3d tomosynthesis to detect breast cancer | |
Ding | Breast cancer pathological image auto-classification using weighted GLCM-SVM | |
Eqbal et al. | Medical image feature extraction using wavelet transform | |
Sinha | CAD based medical image processing: emphasis to breast cancer detection |
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 |