CN106875404A - The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image - Google Patents
The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image Download PDFInfo
- Publication number
- CN106875404A CN106875404A CN201710036438.9A CN201710036438A CN106875404A CN 106875404 A CN106875404 A CN 106875404A CN 201710036438 A CN201710036438 A CN 201710036438A CN 106875404 A CN106875404 A CN 106875404A
- Authority
- CN
- China
- Prior art keywords
- region
- image
- epithelial cell
- gray
- area
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/06—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons
- G06N3/061—Physical realisation, i.e. hardware implementation of neural networks, neurons or parts of neurons using biological neurons, e.g. biological neurons connected to an integrated circuit
-
- 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
- G06T2207/10056—Microscopic 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
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Theoretical Computer Science (AREA)
- Molecular Biology (AREA)
- Neurology (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Microelectronics & Electronic Packaging (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image, including using microscope photographing image, carry out image procossing, image processing step includes gray proces, binary conversion treatment, Morphological scale-space, filling, corrosion, the image for obtaining carries out connected component labeling, retain the boundary rectangle minimum value long and wide of each connected region more than 85 and maximum is more than 130, area is more than 4600 and region that connected region of the girth more than 550 is doubtful epithelial cell；Again suspicious region is carried out into image procossing, the characteristic value in the largest connected region being calculated, the characteristic value with epithelial cell is compared, reservation meets the region input BP neural network of feature, determines whether epithelial cell.The present invention can accurately and rapidly draw testing result, eliminate the influence not high of artificial inefficient operation, precision, with important learning value, and have wide prospect, create considerable economic results in society.
Description
Technical field
The invention belongs to medical digital images process field, epithelial cell in a kind of leukorrhea micro-image is referred specifically to
Intelligent identification Method.
Background technology
Leukorrhea routine inspection is the widest inspection of gynecological disease diagnostic application, various in micro-image by observing
The distribution situation of cell judges the cleannes of leukorrhea, has determined whether inflammation.Area wherein shared by epithelial cell is
Determine a key factor of leukorrhea cleannes.Current detection method is by leukorrhea and 0.9%NACL solution mixing systems into glass
Piece, is examined under a microscope by doctor, and because cell category is various in leukorrhea, complicated component, cell is interweaved and size
The features such as not being easily distinguishable, this test mode is the micro-judgment by medical worker, is doped with many subjective factors, same to timeliness
Rate is low, precision is not high, causes persistence, stability and the objectivity of manual identified to be difficult to ensure that.This method utilizes computer
Digital image processing techniques are combined with biomedicine, realize the Intelligent Recognition of cell, in fast and effeciently detecting leukorrhea
Epithelial cell, so as to improve the diagnosis efficiency of clinic.
The content of the invention
The purpose of the present invention is directed to the deficiencies in the prior art in hospital's leukorrhea detection, it is proposed that a kind of leukorrhea micro-image
The intelligent identification Method of middle epithelial cell, so as to reach the easy, epithelial cell efficiently, accurately detected in leukorrhea sample,
The labour intensity of doctor is substantially reduced, the speed and precision of detection is improved.
The present invention adopts the technical scheme that a kind of intelligent identification Method of epithelial cell in leukorrhea micro-image, the method
Comprise the following steps：
Step 1：The image after slide is smeared into solution using microscope photographing leukorrhea and 0.9%NACL solution mixing systems；
Step 2：Gray proces are carried out to the micro-image that step 1 shoots, gray level image is obtained；
Step 3：The complex background of the gray level image that removal step 2 is obtained；
Step 4：The figure obtained to step 3 carries out binaryzation；
Step 5：The figure obtained to step 4 carries out morphological operation, the less region of removal target；
Step 6：The image that step 5 is obtained is filled；
Step 7：The blank map picture obtained to step 6 corrodes；
Step 8：The image obtained to step 7 carries out connected component labeling, and respective regions are extracted according to mark number；
Step 9：Image after being marked to step 8, calculates the boundary rectangle of each connected region, area and girth, according to
The threshold value of setting is contrasted, and filters out the region of doubtful epithelial cell；
Step 10：The coordinate of each connected region screened to step 9, cuts out step 2 in same coordinate correspondence
Region gray level image；
Step 11：Each region come is cut out to step 10, its edge is detected with the weak Model for Edge Detection of level set, point
Cut out target area；
Step 12：Respectively the average gray value of each target area that statistic procedure 11 is obtained, pixel variance, smoothness and
Uniformity, calculates its characteristic value；
Step 13：Each target area being partitioned into step 11 carries out binaryzation；
Step 14：The figure of each binaryzation obtained to step 13, carries out closed operation；
Step 15：Each the closed operation image obtained to step 14 carries out connected component labeling, and finds each region most
Big connected region；
Step 16：The area of each maximum connected region that calculation procedure 15 finds, the feature such as girth and eccentricity, and
Calculate its characteristic value；
Step 17：The characteristic value in each region counted to step 12 and step 16 and the Standard Eigenvalue of epithelial cell
Compare, the region that reservation is consistent with characteristics of epithelial cells；
Step 18：The provincial characteristics input BP neural network stayed to step 17, is determined whether by BP neural network
Epithelial cell；
Step 19：Count the area of epithelial cell, output result；
Wherein step 4 is concretely comprised the following steps：
Step 4-1：Image to having removed background carries out the cap conversion of morphology bottom, obtains cap changing image on earth；
Step 4-2：The gray threshold obtained using maximum variance between clusters to top cap image；
Step 4-3：Each pixel gray value of gray level image is compared with gray threshold, to the gray scale if more than threshold value
Assignment 255, to this gray scale assignment 0 if less than threshold value, obtains bianry image.
Step 5 is concretely comprised the following steps：
Step 5-1：Figure to binaryzation is first corroded with the circular configuration element that radius is 4, the figure after being corroded；
Step 5-2：Again to the circular configuration element expansion that the figure radius for corroding is 3, the figure after being expanded is
Except the figure behind the less region of target；
Step 9 is concretely comprised the following steps：
Step 9-1：The boundary rectangle of connected region is calculated, the length and minimum value wide of reservation boundary rectangle are more than 85 and most
Big connected region of the value more than 130；
Step 9-2：The area of remaining connected region is calculated, by connected region of the area screening Retention area more than 4600
Domain；
Step 9-3：The girth of remaining connected region is calculated, retains connected region of the girth more than 550；
Step 11 is comprised the concrete steps that：
Step 11-1：The level set movements equation obtained using gradient descent method detects the weak edge in each region, level
Collecting EVOLUTION EQUATION is：
Wherein u, λ, v are constant, δεZ () is Dirac functions, g is edge detection function, and φ is level set function；
Step 11-2：Target area is gone out according to edge separation；
Step 12 is concretely comprised the following steps：
Step 12-1：Calculate the average gray value and pixel variance of target area；
Step 12-2：The smoothness of target area is calculated, the computing formula of smoothness is：
R=1-1/ (1+ σ2)
Wherein, σ is standard deviation, and R is smoothness；
Step 12-3：The uniformity of target area is calculated, the computing formula of uniformity is：
Wherein, U is uniformity, p (zi) it is gray level histogram in a region, i=1,2,3 ... L-1 is correspondence
Histogram, L is differentiable number of grayscale levels.
Step 13 is concretely comprised the following steps：
Step 13-1：The gray threshold obtained using maximum variance between clusters to the image for splitting；
Step 13-2：Each pixel gray value of gray level image is compared with gray threshold, to point ash if more than threshold value
Degree assignment 0, to this gray scale assignment 255, the bianry image for being negated if less than threshold value.
Step 14 is concretely comprised the following steps：
Step 14-1：Binary map is expanded with the circular shuttering that radius is 3 first, the figure for being expanded；
Step 14-2：The figure after expansion is corroded with the circular shuttering that radius is 3 again, the figure after being corroded, i.e.,
It is the figure after closed operation；
Step 15 is concretely comprised the following steps：
Step 15-1：Bianry image to each region carries out connected component labeling；
Step 15-2：Calculate each region markd connected domain area, find the maximum connected region of area；
Step 16 is concretely comprised the following steps：
Step 16-1：Calculate the area and girth in each region；
Step 16-2：Eccentricity is calculated, eccentricity computing formula is：
Wherein, e is eccentricity, and c is half focal length, and a is major semiaxis；
Step 17 is concretely comprised the following steps：
Step 17-1：Screened by area, region of the Retention area more than 4600；
Step 17-2：Screened by girth, retain region of the girth more than 550；
Step 17-3：Screened by eccentricity, retain region of the eccentricity more than 0.3；
Step 17-4：Screened by gray value and pixel variance, retain gray value and exist with pixel variance between 90~180
Region between 90 to 2000；
Step 17-5：By the screening of smoothness, smoothness is retained in the region between 0.004~0.006；
Step 17-6：By the screening of uniformity, uniformity is retained in the region between 0.95~1；
Step 18 is concretely comprised the following steps：
Step 18-1：BP neural network model, including input layer, hidden layer and output layer are set up, input layer there are two god
Through unit, hidden layer has five neurons, and output layer is had a neuron, activated using S type functions, to each layer of weights of connection
The random number in an interval (- 1,1) is assigned respectively, while initializing the threshold value of neutral net；
Step 18-2：Training sample input BP neural network is trained, while the characteristic value that input step 15 and 16 is calculated,
It is epithelial cell to set expectation and be output as 1, and 0 is impurity；
Step 18-3：The input and output of each neuron of hidden layer are calculated, in utilization network desired output and reality output,
The partial derivative δ of each neuron of calculation error function pair output layer0(k)；
Step 18-4：Using the δ of each neuron of output layer0K connection weight is corrected in the output of () and each neuron of hidden layer
Value, recycles the partial derivative of each neuron of hidden layer and the Introduced Malaria connection weight of each neuron of input layer；
Step 18-5：Global error is calculated, this is terminated when the error of desired output and reality output is less than 0.0001
The study of one wheel；
Step 18-6：The sample for detecting input will be needed to have learnt the BP neural network for completing, output is upper more than 0.6
Chrotoplast, output is impurity less than 0.6, and reservation is identified as the region of epithelial cell.
Brief description of the drawings
Fig. 1 is the flow chart of the intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image of the invention.
Specific embodiment
Below in conjunction with the accompanying drawings, a kind of automatic testing method of leukorrhea epithelial cell of the invention is described in detail：
Step 1：Using microscope photographing leukorrhea and 0.9%NACL solution mixing systems into solution image；
Step 2：Gray proces are carried out to the micro-image that step 1 shoots, gray level image is obtained；
Step 3：The background of the gray level image that removal step 2 is obtained；
Step 4：The figure obtained to step 3 carries out binaryzation；
Step 4-1：Image to having removed background carries out the cap conversion of morphology bottom, obtains bot-hat transformation image；
Step 4-2：The gray threshold obtained using maximum variance between clusters to top cap image；
Step 4-3：Each pixel gray value of gray level image is compared with gray threshold, to the gray scale if more than threshold value
Assignment 255, to this gray scale assignment 0 if less than threshold value, obtains bianry image.
Step 5：The figure obtained to step 4 carries out morphological operation, the less region of removal target；
Step 5-1：Figure to binaryzation is first corroded with the circular configuration element that radius is 4, the figure after being corroded；
Step 5-2：Again to the circular configuration element expansion that the figure radius for corroding is 3, the figure after being expanded is
Except the figure behind the less region of target；
Step 6：The image that step 5 is obtained is filled；
Step 7：The blank map picture obtained to step 6 corrodes；
Step 8：The image obtained to step 7 carries out connected component labeling, and respective regions are extracted according to mark number；
Step 9：Image after being marked to step 8, calculates the boundary rectangle of each connected region, area and girth, according to
The threshold value of setting is contrasted, and filters out the region of doubtful epithelial cell；
Step 9-1：The boundary rectangle of connected region is calculated, the length and minimum value wide of reservation boundary rectangle are more than 85 and most
Big connected region of the value more than 130；
Step 9-2：The area of remaining connected region is calculated, by connected region of the area screening Retention area more than 4600
Domain；
Step 9-3：The girth of remaining connected region is calculated, retains connected region of the girth more than 550；
Step 10：The coordinate of each connected region screened to step 9, cuts out step 2 in same coordinate correspondence
Region gray level image；
Step 11：Each region come is cut out to step 10, its edge is detected with the weak Model for Edge Detection of level set, point
Cut out target area；
Step 11-1：The level set movements equation obtained using gradient descent method detects the weak edge in each region, level
Collecting EVOLUTION EQUATION is：
Wherein u, λ, v are constant, δεZ () is Dirac functions, g is edge detection function, and φ is level set function；
Step 11-2：Target area is gone out according to edge separation；
Step 12：Respectively the average gray value of each target area that statistic procedure 11 is obtained, pixel variance, smoothness and
Uniformity, calculates its characteristic value；
Step 12-1：Calculate the average gray value and pixel variance of target area；
Step 12-2：The smoothness of target area is calculated, the computing formula of smoothness is：
R=1-1/ (1+ σ2)
Wherein, σ is standard deviation, and R is smoothness；
Step 12-3：The uniformity of target area is calculated, the computing formula of uniformity is：
Wherein, U is uniformity, p (zi) it is gray level histogram in a region, i=1,2,3 ... L-1 is correspondence
Histogram, L is differentiable number of grayscale levels.
Step 13：Each target area being partitioned into step 11 carries out binaryzation；
Step 13-1：The gray threshold obtained using maximum variance between clusters to the image for splitting；
Step 13-2：Each pixel gray value of gray level image is compared with gray threshold, to point ash if more than threshold value
Degree assignment 0, to this gray scale assignment 255, the bianry image for being negated if less than threshold value.
Step 14：The figure of each binaryzation obtained to step 13, carries out closed operation；
Step 15：Each the closed operation image obtained to step 14 carries out connected component labeling, and finds each region most
Big connected region；
Step 15-1：Bianry image to each region carries out connected component labeling；
Step 15-2：Calculate each region markd connected domain area, find the maximum connected region of area；
Step 16：The area of each maximum connected region that calculation procedure 15 finds, the feature such as girth and eccentricity, meter
Calculate its characteristic value；
Step 16-1：Calculate the area and girth in each region；
Step 16-2：Eccentricity is calculated, eccentricity computing formula is：
Wherein, e is eccentricity, and c is half focal length, and a is major semiaxis；
Step 17：The characteristic value in each region counted to step 12 and step 16 and the Standard Eigenvalue of epithelial cell
Compare, the region that reservation is consistent with characteristics of epithelial cells；
Step 17-1：Screened by area, region of the Retention area more than 4600；
Step 17-2：Screened by girth, retain region of the girth more than 550；
Step 17-3：Screened by eccentricity, retain region of the eccentricity more than 0.3；
Step 17-4：Screened by gray value and pixel variance, retain gray value and exist with pixel variance between 90~180
Region between 90 to 2000；
Step 17-5：By the screening of smoothness, smoothness is retained in the region between 0.004~0.006；
Step 17-6：By the screening of uniformity, uniformity is retained in the region between 0.95~1；
Step 18：The provincial characteristics input BP neural network stayed to step 17, is determined whether by BP neural network
Epithelial cell；
Step 18-1：BP neural network model, including input layer, hidden layer and output layer are set up, input layer there are two god
Through unit, hidden layer has five neurons, and output layer is had a neuron, activated using S type functions, to each layer of weights of connection
The random number in an interval (- 1,1) is assigned respectively, while initializing the threshold value of neutral net；
Step 18-2：Training sample input BP neural network is trained, while the characteristic value that input step 15 and 16 is calculated,
It is epithelial cell to set expectation and be output as 1, and 0 is impurity；
Step 18-3：The input and output of each neuron of hidden layer are calculated, in utilization network desired output and reality output,
The partial derivative δ of each neuron of calculation error function pair output layer0(k)；
Step 18-4：Using the δ of each neuron of output layer0K connection weight is corrected in the output of () and each neuron of hidden layer
Value, recycles the partial derivative of each neuron of hidden layer and the Introduced Malaria connection weight of each neuron of input layer；
Step 18-5：Global error is calculated, this is terminated when the error of desired output and reality output is less than 0.0001
The study of one wheel；
Step 18-6：The sample for detecting input will be needed to have learnt the BP neural network for completing, output is upper more than 0.6
Chrotoplast, output is impurity less than 0.6, and reservation is identified as the region of epithelial cell.
Step 19：Count the area of epithelial cell, output result.
By embodiment of above, it is seen that the present invention has the following advantages that：
(1) this method introduces a kind of brand-new level-set segmentation side theoretical based on curve evolvement for meeting human vision
Method, the Level Set Method utilizes the energy hole Evolution Rates of curve, and can detect the weak edge of cell, solves leukorrhea
Cell is difficult to the problem split in micro-image；
(2) to the cell for splitting, BP neural network model training sample is make use of, automatically extracts the feature of cell,
Detection rates are substantially increased, while improve the degree of accuracy of identification.
Above content is the further description for combining specific embodiment to the present patent application, it is impossible to assert this Shen
Specific implementation please is confined to these explanations.For the application person of an ordinary skill in the technical field, do not taking off
On the premise of design of the invention, the deformation made or improve these and belong to protection scope of the present invention.
Claims (12)
1. in a kind of leukorrhea micro-image epithelial cell intelligent identification Method, it is characterised in that the detailed process of the method is：
Step 1：The image after slide is smeared into solution using microscope photographing leukorrhea and 0.9%NACL solution mixing systems；
Step 2：Gray proces are carried out to the micro-image that step 1 shoots, gray level image is obtained；
Step 3：The complex background of the gray level image that removal step 2 is obtained；
Step 4：The figure obtained to step 3 carries out binaryzation；
Step 5：The figure obtained to step 4 carries out morphological operation, the less region of removal target；
Step 6：The image that step 5 is obtained is filled；
Step 7：The blank map picture obtained to step 6 corrodes；
Step 8：The image obtained to step 7 carries out connected component labeling, and respective regions are extracted according to mark number；
Step 9：Image after being marked to step 8, calculates the boundary rectangle of each connected region, area and girth, according to setting
Threshold value contrasted, filter out the region of doubtful epithelial cell；
Step 10：The coordinate of each connected region screened to step 9, cuts out step 2 in the corresponding area of same coordinate
The gray level image in domain；
Step 11：Each region come is cut out to step 10, its edge is detected with the weak Model for Edge Detection of level set, be partitioned into
Target area；
Step 12：Respectively the average gray value of each target area that statistic procedure 11 is obtained, pixel variance, smoothness with it is consistent
Property, calculate its characteristic value；
Step 13：Each target area being partitioned into step 11 carries out binaryzation；
Step 14：The figure of each binaryzation obtained to step 13, carries out closed operation；
Step 15：Each the closed operation image obtained to step 14 carries out connected component labeling, and finds each region maximum
Connected region；
Step 16：The area of each maximum connected region that calculation procedure 15 finds, the feature such as girth and eccentricity, and calculate
Its characteristic value；
Step 17：The characteristic value in each region counted to step 12 and step 16 is carried out with the Standard Eigenvalue of epithelial cell
Compare, the region that reservation is consistent with characteristics of epithelial cells；
Step 18：The provincial characteristics input BP neural network stayed to step 17, epithelium is determined whether by BP neural network
Cell；
Step 19：Count the area of epithelial cell, output result.
2. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 4 is：
Step 4-1：Image to having removed background carries out the cap conversion of morphology bottom, obtains cap changing image on earth；
Step 4-2：The gray threshold obtained using maximum variance between clusters to top cap image；
Step 4-3：Each pixel gray value of gray level image is compared with gray threshold, to this gray scale assignment if more than threshold value
255, to this gray scale assignment 0 if less than threshold value, obtain bianry image.
3. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 5 is：
Step 5-1：Figure to binaryzation is first corroded with the circular configuration element that radius is 4, the figure after being corroded；
Step 5-2：Again to the circular configuration element expansion that the figure radius for corroding is 3, the figure after being expanded is removal mesh
Mark the figure behind less region.
4. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 9 is：
Step 9-1：The boundary rectangle of connected region is calculated, the length and minimum value wide for retaining boundary rectangle are more than 85 and maximum
Connected region more than 130；
Step 9-2：The area of remaining connected region is calculated, by area connected region of the screening Retention area more than 4600；
Step 9-3：The girth of remaining connected region is calculated, retains connected region of the girth more than 550.
5. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 11 is：
Step 11-1：The level set movements equation obtained using gradient descent method detects the weak edge in each region, and level set is drilled
Changing equation is：
Wherein μ, λ, v are constant, δεZ () is Dirac functions, g is edge detection function,It is level set function；
Step 11-2：Target area is gone out according to edge separation.
6. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 12 is：
Step 12-1：Calculate the average gray value and pixel variance of target area；
Step 12-2：The smoothness of target area is calculated, the computing formula of smoothness is：
R=1-1/ (1+ σ2)
Wherein, σ is standard deviation, and R is smoothness；
Step 12-3：The uniformity of target area is calculated, the computing formula of uniformity is：
Wherein, U is uniformity, p (Zi) it is gray level histogram in a region, i=1,2,3 ... L-1 is corresponding Nogata
Figure, L is differentiable number of grayscale levels.
7. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 13 is：
Step 13-1：The gray threshold obtained using maximum variance between clusters to the image for splitting；
Step 13-2：Each pixel gray value of gray level image is compared with gray threshold, the gray scale is assigned if more than threshold value
Value 0, to this gray scale assignment 255, the bianry image for being negated if less than threshold value.
8. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 14 is：
Step 14-1：Binary map is expanded with the circular shuttering that radius is 3 first, the figure for being expanded；Step 14-2：Use again
Radius is that 3 circular shuttering corrodes to the figure after expansion, and the figure after being corroded is the figure after closed operation.
9. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 15 is：
Step 15-1：Bianry image to each region carries out connected component labeling；
Step 15-2：Calculate each region markd connected domain area, find the maximum connected region of area
10. in leukorrhea micro-image according to claim 1 epithelial cell intelligent identification Method, it is characterised in that it is described
The detailed process of step 16 is：
Step 16-1：Calculate the area and girth in each region；
Step 16-2：Eccentricity is calculated, eccentricity computing formula is：
Wherein, e is eccentricity, and c is half focal length, and a is major semiaxis.
The intelligent identification Method of epithelial cell in 11. leukorrhea micro-images according to claim 1, it is characterised in that described
The detailed process of step 17 is：
Step 17-1：Screened by area, region of the Retention area more than 4600；
Step 17-2：Screened by girth, retain region of the girth more than 550；
Step 17-3：Screened by eccentricity, retain region of the eccentricity more than 0.3；
Step 17-4：Screened by gray value and pixel variance, retain gray value and arrived 90 with pixel variance between 90~180
Region between 2000；
Step 17-5：By the screening of smoothness, smoothness is retained in the region between 0.004~0.006；
Step 17-6：By the screening of uniformity, uniformity is retained in the region between 0.95~1.
The intelligent identification Method of epithelial cell in 12. leukorrhea micro-images according to claim 1, it is characterised in that described
The detailed process of step 18 is：
Step 18-1：BP neural network model, including input layer, hidden layer and output layer are set up, input layer there are two neurons,
Hidden layer has five neurons, and output layer is had a neuron, activated using S type functions, to each layer of weights difference of connection
The random number in an interval (- 1,1) is assigned, while initializing the threshold value of neutral net；
Step 18-2：By training sample input BP neural network training, while the characteristic value that input step 15 and 16 is calculated, is set
Desired output is epithelial cell for 1, and 0 is impurity；
Step 18-3：The input and output of each neuron of hidden layer are calculated, network desired output and reality output is being utilized, calculated
Partial derivative δ of the error function to each neuron of output layer0(k)；
Step 18-4：Using the δ of each neuron of output layer0K connection weight is corrected in the output of () and each neuron of hidden layer, then
Using the partial derivative and the Introduced Malaria connection weight of each neuron of input layer of each neuron of hidden layer；
Step 18-5：Global error is calculated, this wheel is terminated when the error of desired output and reality output is less than 0.0001
Study；
Step 18-6：The sample input of detection will be needed to have learnt the BP neural network for completing, output is that epithelium is thin more than 0.6
Born of the same parents, output is impurity less than 0.6, and reservation is identified as the region of epithelial cell.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710036438.9A CN106875404A (en) | 2017-01-18 | 2017-01-18 | The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710036438.9A CN106875404A (en) | 2017-01-18 | 2017-01-18 | The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106875404A true CN106875404A (en) | 2017-06-20 |
Family
ID=59158548
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710036438.9A Pending CN106875404A (en) | 2017-01-18 | 2017-01-18 | The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106875404A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108334835A (en) * | 2018-01-29 | 2018-07-27 | 华东师范大学 | Vaginal fluid micro-image visible component detection method based on convolutional neural networks |
| CN108830858A (en) * | 2018-06-20 | 2018-11-16 | 天津大学 | It is a kind of based on infrared and optical image double-mode imaging information living body method for counting colonies |
| CN109967143A (en) * | 2019-02-27 | 2019-07-05 | 西安理工大学 | A kind of cell size detection method based on micro-fluidic microscopic system |
| CN107330465B (en) * | 2017-06-30 | 2019-07-30 | 清华大学深圳研究生院 | A kind of images steganalysis method and device |
| CN110223307A (en) * | 2018-03-01 | 2019-09-10 | 深圳大森智能科技有限公司 | A kind of blood cell counting method based on image recognition |
| CN110838126A (en) * | 2019-10-30 | 2020-02-25 | 东莞太力生物工程有限公司 | Cell image segmentation method, cell image segmentation device, computer equipment and storage medium |
| CN111651268A (en) * | 2020-05-14 | 2020-09-11 | 武汉兰丁智能医学股份有限公司 | Microscopic image rapid processing system |
| CN113295692A (en) * | 2021-05-25 | 2021-08-24 | 郑州中普医疗器械有限公司 | Cell analysis method based on cell nucleus DNA and TBS double analysis method, computer equipment and storage medium |
| CN116703917A (en) * | 2023-08-07 | 2023-09-05 | 广州盛安医学检验有限公司 | Female genital tract cell pathology intelligent analysis system |
| CN117831033A (en) * | 2024-03-04 | 2024-04-05 | 南京市浦口人民医院(江苏省人民医院浦口分院) | Intelligent extraction and identification method for pathogenic microorganisms based on image analysis |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005114291A1 (en) * | 2004-05-12 | 2005-12-01 | Beth Israel Deaconess Medical Center, Inc. | Method and apparatus for detecting microscopic objects |
| CN104156951A (en) * | 2014-07-30 | 2014-11-19 | 电子科技大学 | Leukocyte detecting method aiming at bronchoalveolar lavage smear |
| CN104846054A (en) * | 2015-05-22 | 2015-08-19 | 电子科技大学 | Method for automatically detecting moulds in leucorrhea based on morphological characteristics |
| CN106295715A (en) * | 2016-08-22 | 2017-01-04 | 电子科技大学 | A kind of leucorrhea cleannes automatic classification method based on BP neural network classifier |
| CN106295588A (en) * | 2016-08-17 | 2017-01-04 | 电子科技大学 | The automatic identifying method of leukocyte in a kind of leucorrhea micro-image |
-
2017
- 2017-01-18 CN CN201710036438.9A patent/CN106875404A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2005114291A1 (en) * | 2004-05-12 | 2005-12-01 | Beth Israel Deaconess Medical Center, Inc. | Method and apparatus for detecting microscopic objects |
| CN104156951A (en) * | 2014-07-30 | 2014-11-19 | 电子科技大学 | Leukocyte detecting method aiming at bronchoalveolar lavage smear |
| CN104846054A (en) * | 2015-05-22 | 2015-08-19 | 电子科技大学 | Method for automatically detecting moulds in leucorrhea based on morphological characteristics |
| CN106295588A (en) * | 2016-08-17 | 2017-01-04 | 电子科技大学 | The automatic identifying method of leukocyte in a kind of leucorrhea micro-image |
| CN106295715A (en) * | 2016-08-22 | 2017-01-04 | 电子科技大学 | A kind of leucorrhea cleannes automatic classification method based on BP neural network classifier |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110334706B (en) * | 2017-06-30 | 2021-06-01 | 清华大学深圳研究生院 | Image target identification method and device |
| CN107330465B (en) * | 2017-06-30 | 2019-07-30 | 清华大学深圳研究生院 | A kind of images steganalysis method and device |
| CN110334706A (en) * | 2017-06-30 | 2019-10-15 | 清华大学深圳研究生院 | A kind of images steganalysis method and device |
| CN108334835B (en) * | 2018-01-29 | 2021-11-19 | 华东师范大学 | Method for detecting visible components in vaginal secretion microscopic image based on convolutional neural network |
| CN108334835A (en) * | 2018-01-29 | 2018-07-27 | 华东师范大学 | Vaginal fluid micro-image visible component detection method based on convolutional neural networks |
| CN110223307B (en) * | 2018-03-01 | 2023-12-15 | 深圳大森智能科技有限公司 | Blood cell counting method based on image recognition |
| CN110223307A (en) * | 2018-03-01 | 2019-09-10 | 深圳大森智能科技有限公司 | A kind of blood cell counting method based on image recognition |
| CN108830858A (en) * | 2018-06-20 | 2018-11-16 | 天津大学 | It is a kind of based on infrared and optical image double-mode imaging information living body method for counting colonies |
| CN108830858B (en) * | 2018-06-20 | 2021-08-03 | 天津大学 | Living body colony counting method based on infrared and optical image dual-mode imaging information |
| CN109967143B (en) * | 2019-02-27 | 2021-06-15 | 西安理工大学 | Cell size detection method based on micro-fluidic microscope system |
| CN109967143A (en) * | 2019-02-27 | 2019-07-05 | 西安理工大学 | A kind of cell size detection method based on micro-fluidic microscopic system |
| CN110838126B (en) * | 2019-10-30 | 2020-11-17 | 东莞太力生物工程有限公司 | Cell image segmentation method, cell image segmentation device, computer equipment and storage medium |
| CN110838126A (en) * | 2019-10-30 | 2020-02-25 | 东莞太力生物工程有限公司 | Cell image segmentation method, cell image segmentation device, computer equipment and storage medium |
| CN111651268A (en) * | 2020-05-14 | 2020-09-11 | 武汉兰丁智能医学股份有限公司 | Microscopic image rapid processing system |
| CN111651268B (en) * | 2020-05-14 | 2023-01-31 | 武汉兰丁智能医学股份有限公司 | Quick processing system for microscopic image |
| CN113295692A (en) * | 2021-05-25 | 2021-08-24 | 郑州中普医疗器械有限公司 | Cell analysis method based on cell nucleus DNA and TBS double analysis method, computer equipment and storage medium |
| CN116703917A (en) * | 2023-08-07 | 2023-09-05 | 广州盛安医学检验有限公司 | Female genital tract cell pathology intelligent analysis system |
| CN116703917B (en) * | 2023-08-07 | 2024-01-26 | 广州盛安医学检验有限公司 | Female genital tract cell pathology intelligent analysis system |
| CN117831033A (en) * | 2024-03-04 | 2024-04-05 | 南京市浦口人民医院(江苏省人民医院浦口分院) | Intelligent extraction and identification method for pathogenic microorganisms based on image analysis |
| CN117831033B (en) * | 2024-03-04 | 2024-05-07 | 南京市浦口人民医院(江苏省人民医院浦口分院) | Intelligent extraction and identification method for pathogenic microorganisms based on image analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106875404A (en) | The intelligent identification Method of epithelial cell in a kind of leukorrhea micro-image | |
| WO2022160771A1 (en) | Method for classifying hyperspectral images on basis of adaptive multi-scale feature extraction model | |
| CN113344849B (en) | Microemulsion head detection system based on YOLOv5 | |
| CN104156734B (en) | A kind of complete autonomous on-line study method based on random fern grader | |
| CN106295588A (en) | The automatic identifying method of leukocyte in a kind of leucorrhea micro-image | |
| CN107680678A (en) | Based on multiple dimensioned convolutional neural networks Thyroid ultrasound image tubercle auto-check system | |
| CN105956582A (en) | Face identifications system based on three-dimensional data | |
| CN107610087A (en) | A kind of tongue fur automatic division method based on deep learning | |
| CN104846054A (en) | Method for automatically detecting moulds in leucorrhea based on morphological characteristics | |
| CN108052886B (en) | A kind of puccinia striiformis uredospore programming count method of counting | |
| CN110555836A (en) | Automatic identification method and system for standard fetal section in ultrasonic image | |
| CN105069818A (en) | Image-analysis-based skin pore identification method | |
| CN108830856B (en) | GA automatic segmentation method based on time series SD-OCT retina image | |
| CN108186051A (en) | A kind of image processing method and processing system of the automatic measurement fetus Double Tops electrical path length from ultrasonoscopy | |
| CN112862849B (en) | Image segmentation and full convolution neural network-based field rice ear counting method | |
| CN109670489B (en) | Weak supervision type early senile macular degeneration classification method based on multi-instance learning | |
| CN105913013A (en) | Binocular vision face recognition algorithm | |
| CN107871316A (en) | A kind of X-ray hand bone interest region extraction method based on deep neural network | |
| TW202004776A (en) | Establishing method of bone age assessment and height prediction model, bone age assessment and height prediction system, and bone age assessment and height prediction method | |
| CN107392929A (en) | A kind of intelligent target detection and dimension measurement method based on human vision model | |
| CN113762009B (en) | Crowd counting method based on multi-scale feature fusion and double-attention mechanism | |
| CN104458747A (en) | Rice chalkiness measurement and calculation method | |
| CN108257124A (en) | A kind of white blood cell count(WBC) method and system based on image | |
| CN109003275A (en) | The dividing method of weld defect image | |
| CN102567734A (en) | Specific value based retina thin blood vessel 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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170620 |