CN110415255A - A kind of immunohistochemistry pathological image CD3 positive nucleus dividing method and system - Google Patents
A kind of immunohistochemistry pathological image CD3 positive nucleus dividing method and system Download PDFInfo
- Publication number
- CN110415255A CN110415255A CN201910517113.1A CN201910517113A CN110415255A CN 110415255 A CN110415255 A CN 110415255A CN 201910517113 A CN201910517113 A CN 201910517113A CN 110415255 A CN110415255 A CN 110415255A
- Authority
- CN
- China
- Prior art keywords
- image
- nucleus
- segmentation
- area
- processed
- 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
- 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
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Investigating Or Analysing Biological Materials (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of immunohistochemistry pathological image CD3 positive nucleus dividing method and system, steps of the method are: color deconvolution is carried out to immunohistochemistry pathological image, separates dyeing channel；Image, at irregular block of pixels, is carried out kmeans cluster and distinguishes image background and remove using super-pixel segmentation；Image segmentation is carried out based on morphological feature, obtains primary segmentation good nucleus first area image L1 and the first image C1 to be processed；First image C1 to be processed is subjected to local threshold bernsen segmentation and morphological feature segmentation, obtains the nucleus second area image L2 divided and the second image C2 to be processed；Second image C2 to be processed is subjected to prospect label, and nucleus third area image L3 is partitioned into using watershed algorithm；Nucleus first area image L1, nucleus second area image L2, nucleus third area image L3 are constituted into nucleus segmented image.Robustness of the present invention is high, and segmentation precisely, can satisfy practical application request.
Description
Technical field
The present invention relates to the technical fields of image procossing, and in particular to a kind of immunohistochemistry pathological image CD3 positive cell
Core dividing method and system.
Background technique
CD3 is a kind of important leukocyte differentiation antigen, and almost all of T cell surface exists, and is to constitute T cell to resist
The membranous antigen of original receptor (TCR).When antigen is in conjunction with TCR, its signal is participated in intracellular reception and registration, is various each with discovery
The related important membranous antigen of the T cell function of sample.Medical staff carries out segmentation, the label of CD3 cell to pathological image at present
It to be expended considerable time and effort with counting, therefore, it is necessary to a kind of more accurate pathology figure cell segmentation means to mitigate
The pressure of medical staff.
Summary of the invention
In order to mitigate the burden of medical staff, the automatic segmentation that CD3 cell is carried out for pathological image is realized, the present invention mentions
For a kind of immunohistochemistry pathological image CD3 positive nucleus dividing method and system, robustness of the present invention is high, and segmentation precisely, can
To meet practical application request.
In order to achieve the above object, the invention adopts the following technical scheme:
The present invention provides a kind of immunohistochemistry pathological image CD3 positive nucleus dividing method, includes the following steps:
S1: color deconvolution is carried out to immunohistochemistry pathological image, separates dyeing channel；
S2: being transformed into Lab color space from RGB color for preexisting immunity group pathological image, using super-pixel point
It is cut into irregular block of pixels, the irregular block of pixels carries out kmeans cluster and distinguishes image background, and image background is gone
It removes；
S3: based on morphological feature carry out image segmentation, obtain the good nucleus first area image L1 of primary segmentation with
And the first image C1 to be processed；
S4: the first image C1 to be processed is subjected to local threshold bernsen segmentation and morphological feature is divided, is divided
The nucleus second area image L2 and the second image C2 to be processed cut；
S5: the second image C2 to be processed is subjected to prospect label, and nucleus third is partitioned into using watershed algorithm
Area image L3；
S6: by nucleus first area image L1, nucleus second area image L2, nucleus third area image L3 with
Immunohistochemistry pathological image original image carries out mask process, obtains nucleus segmented image.
Color deconvolution is carried out to immunohistochemistry pathological image described in step S1 as a preferred technical solution, is calculated
Formula are as follows:
C=M-1[y]
Wherein, C indicates that isolated H and DAB dyeing channel, M indicate different dyeing parameter matrixes, the row difference of parameter matrix
Indicate H, eosin and DAB dyeing, the column of parameter matrix indicate RGB Staining Protocol parameter size, and y indicates that the light of each pixel is close
Degree.
Irregular block of pixels described in step S2 carries out kmeans cluster and distinguishes image back as a preferred technical solution,
Scape, by L layers in the Lab color space of the irregular block of pixels object as kmeans cluster, the specific following institute of calculation formula
It states:
Wherein, E indicates least squares error, and x indicates each irregular block of pixels, and u indicates mass center, g presentation class
Cluster, when E minimum, then kmeans cluster is completed.
Image segmentation, specific steps are carried out based on morphological feature described in step S3 as a preferred technical solution, are as follows:
S31: by the image and DAB dyeing channel progress mask process after removal background, the DAB of acquisition removal background is dyed
Channel image, and use watershed algorithm segmented image；
S32: extracting characteristics of image after watershed algorithm is divided, removes the region of doubtful dust, retains nucleus the
One area image L1 and the first image C1 to be processed.
Described image feature includes gray average, contrast, compactness and elemental area as a preferred technical solution,
The region of the doubtful dust is set as contrast less than 0.04 or gray average is greater than 200 image-region, the nucleus
First area image L1 is set as the elemental area of 40X amplification factor less than 2000 and density is greater than 0.93 image-region, described
The elemental area that first image C1 to be processed is set as 40X amplification factor is greater than 2000 image-region.
The first image C1 to be processed is subjected to local threshold bernsen described in step S4 as a preferred technical solution,
Segmentation and morphological feature segmentation, specific steps are as follows:
The active window of S41: local threshold bernsen segmentation is set as 77*77 pixel size, and local threshold bernsen divides
The binary image of the first image C1 to be processed is obtained after cutting, and carries out exposure mask with DAB dyeing channel, obtains DAB dyeing channel
The nuclear area the first image C1 to be processed area image, execute opening operation after carry out watershed segmentation again；
S42: the image handled by local threshold and watershed segmentation carries out the image segmentation based on morphological feature,
Characteristics of image is extracted, nucleus second area image L2 and the second image C2 to be processed is retained.
The nucleus second area image L2 is set as the elemental area of 40X amplification factor as a preferred technical solution,
Less than 2000 and density is greater than 0.93 image-region, and the second image C2 to be processed is set as the pixel of 40X amplification factor
Area is greater than 2000 image-region.
The second image C2 to be processed is subjected to prospect label described in step S5 as a preferred technical solution, and is adopted
Nucleus third area image L3, specific steps are partitioned into watershed algorithm are as follows:
S51: exposure mask is carried out with DAB dyeing channel after the second image C2 progress binary conversion treatment to be processed, is obtained wait locate
The DAB staining cell cut zone of reason using the opening and closing operations based on reconstruction and takes local maximum to carry out prospect label；
S52: the image after prospect label is overlapped with DAB staining cell cut zone to be processed in step S51, will
Prospect marks the local minimum for being set as image nuclear centers, carries out image segmentation using fractional spins and obtains carefully
Karyon third area image L3.
Further include the steps that edge line is arranged in nucleus as a preferred technical solution, obtains nucleus in step s 6
Boundary line is arranged in nuclear area edge in segmented image, and multiple summits borderline linking generates edge line.
The present invention also provides a kind of immunohistochemistry pathological image CD3 positive nucleus segmenting systems, comprising: dyeing channel point
From module, background removal module, image primary segmentation module, local threshold bernsen segmentation module and watershed segmentation module；
The dyeing channel separation module is equipped with color warp product unit, and color warp product unit is used for immunohistochemistry disease
It manages image and carries out color deconvolution, dyeing channel is separated；
The background removal module includes super-pixel segmentation unit and kmeans cluster cell, the super-pixel segmentation unit
Irregular block of pixels is divided the image into, the kmeans cluster cell is for distinguishing background image；
Described image primary segmentation module is used for morphological feature image segmentation, obtains the good nucleus first of primary segmentation
Area image L1 and the first image C1 to be processed；
The local threshold bernsen segmentation module is used to the first image C1 to be processed carrying out local threshold
Bernsen segmentation, obtains nucleus second area image L2 and the second image C2 to be processed；
The watershed segmentation module is used to carrying out the second image C2 to be processed into prospect label and watershed algorithm point
Nucleus third area image L3 is cut out,
The nucleus first area image L1, nucleus second area image L2 and nucleus third area image L3 with
After immunohistochemistry pathological image original image carries out mask process, immunohistochemistry pathological image CD3 positive nucleus segmentation result is obtained
Image.
Compared with the prior art, the invention has the following advantages and beneficial effects:
(1) present invention divides the image into irregular block of pixels using super-pixel segmentation when Kmeans clusters removal background,
It avoids dyeing shallower part in cell and is misidentified as background removal, improve the accuracy of image procossing.
(2) present invention uses local threshold bernsen segmented image, due to the situation uneven there may be dyeing, causes
Color of image is partially deep or partially shallow, and local threshold bernsen segmented image eliminates the interference of dyeing difference, improves at image
The precision of reason.
(3) present invention is solved and watershed algorithm point is used alone using prospect label and watershed algorithm segmented image
The problem of overlapping cell can not be divided when cutting, the technical effect of accurate segmentation is reached.
Detailed description of the invention
Fig. 1 is the flow diagram of the present embodiment immunohistochemistry pathological image CD3 positive nucleus dividing method；
Fig. 2 is the pathological image nucleus point of the present embodiment immunohistochemistry pathological image CD3 positive nucleus dividing method
Cut effect diagram.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and
It is not used in the restriction present invention.
Embodiment
As shown in Figure 1, the present embodiment provides a kind of immunohistochemistry pathological image CD3 positive nucleus dividing methods, including
Following step:
S1: color deconvolution is carried out for the immunohistochemistry pathological image of original RGB coding, by haematoxylin
(Haematoxylin, H) and two dyeing channels of diaminobenzidine (3,3'-Diaminobenzidine, DAB) are divided
From, only to DAB dyeing CD3 positive cell be split；
In the present embodiment, color Deconvolution Algorithm Based on Frequency is directed to the colouring information that RGB video camera obtains, and is based on immunohistochemistry skill
The specificity for the coloring agent RGB component light that art uses absorbs, and calculates separately every kind of coloring agent to the function and effect of image, deconvolution
Refer to that known output and part input, calculate the process of Unknown worm, wherein output is CD3 colored graph, it is known that input is H=
[0.6500286,0.704031,0.2860126], DAB=[0.26814753,0.57031375,0.77642715], thus
H coloured portions and DAB coloured portions, formula are as follows out:
C=M-1[y]
Wherein, C is H the and DAB dyeing channel separated, and M is different dyeing parameter matrixes, the row difference of parameter matrix
It is dyed for H, eosin (eosin) and DAB, eosin is only not used as background color in the present embodiment, and parameter matrix is classified as
RGB Staining Protocol parameter size, y are the optical density of each pixel；
S2: super-pixel segmentation and Kmeans cluster removal background:
The present embodiment first has to removal background, is contaminated according to background in order to carry out feature extraction and segmentation to cellular portions
Preexisting immunity group pathological image is transformed into Lab color space from RGB color by the shallower principle of color, L (brightness) layer compared with
Big region is background, while being identified as background in order to avoid dyeing shallower part in cell, will using super-pixel segmentation
With similar luminance, the adjacent pixel of texture and color is divided into irregular block of pixels；
Using the object that L (brightness) layer is clustered as kmeans in the Lab color space of irregular block of pixels, in irregular picture
The position of original image is averaged where plain block, is divided into two classes after doing kmeans cluster to mean value result, so that the point of two classes is from its institute
The gap and minimum of the class of category, the biggish one kind of L (brightness) layer are removed as background, retain the lesser cell portion of L (brightness) layer
Divide and carry out next step segmentation, calculation formula is as follows:
Wherein, E indicates least squares error, and x indicates each irregular block of pixels, and u indicates mass center, g presentation class
Cluster, when E minimum, then kmeans cluster is completed；
S3: morphological feature primary segmentation is split using the essential characteristic of cell；
S31: carrying out mask process to the image and DAB dyeing channel that remove background, and the DAB dyeing for obtaining removal background is logical
Road image the case where for linking together there are cell dyeing, is first gone out using watershed segmentation and links together but have
The cell image on obvious boundary, wherein watershed algorithm be it is a kind of using image as learn topological diagram method, the gray scale of pixel
Height above sea level of the size as the point overflows water upwards at each local minimum, divides water in different basin intersection formation
Ridge, as the part to be divided；
S32: by the good every piece of region of image zooming-out of watershed segmentation in 40X amplification factor (0.2520 micron/pixel)
Feature such as gray average, contrast, compactness and elemental area etc., in the present embodiment, pixel region gray scale summation is divided by picture
Prime number measures gray average, and the number of pixels of each disconnected pixel region forms elemental area, in gray scale 0 to 255
Section is set as contrast, and the distance of each disconnected each pixel of nuclear area is calculated compactness, removes doubtful
The contrast of dust less than 0.04 or gray average be greater than 200 image-region, retain 40X amplification factor (0.2520 micron/
Pixel) elemental area less than 2000 and density be greater than 0.93 image-region, nucleus firstth area good as primary segmentation
Area image L1, elemental area carry out next step segmentation as the first image C1 to be processed greater than 2000 remainder；
S4: local threshold bernsen segmentation, removal dyeing difference interference；
S41: the first image C1 to be processed may influence since dyeing is uneven, cause color partially deep or partially shallow, use
Local threshold segmentation excludes dyeing difference interference, and Local threshold segmentation is movable under 40X amplification factor (0.2520 micron/pixel)
Window is 77*77 pixel size, i.e., automatically determines threshold value by system in the window of every 77*77 pixel size and image is carried out two
The binary image that the first image C1 to be processed is obtained after Local threshold segmentation and DAB dyeing channel are carried out exposure mask by value
Processing obtains the first nuclear area image C1 area image of the processing of DAB dyeing channel, retains local threshold bernsen segmentation
The DAB of the first image C1 image remaining area to be processed dyes picture afterwards, carries out opening operation, that is, uses the disk of 5 pixel sizes
The operation of corrosion reflation is carried out to picture, removal partial noise makes the nuclear area of segmentation more round while interference
It is sliding, carry out watershed segmentation again later；
S42: after eliminating dyeing difference interference, the figure handled by local threshold and watershed segmentation is carried out again
Morphological segment, extracts each isolated area in the elemental area of 40X amplification factor (0.2520 micron/pixel), close
Degree, contrast and the gray scale features such as averagely, retain 40X amplification factor (0.2520 micron/pixel) elemental area less than 2000 and
Density is used as nucleus second area image L2 greater than 0.93, and remainder of the elemental area greater than 2000 is as to be processed
Second image C2 carries out next step segmentation；
S5: prospect marks watershed segmentation to be overlapped cell；
S51: by the segmentation of above-mentioned steps, for the second part image C2 to be processed, remaining many overlaps
Nucleus, the second image C2 to be processed is carried out to carry out mask process, gained portion with DAB dyeing channel after binary conversion treatment
It point is the cell being closer, color is connected together to get to DAB staining cell cut zone to be processed when dyeing,
Using the opening and closing operations based on reconstruction and local maximum is taken to do prospect label, obtains the foreground image of cell, more accurately go
Further divided except background parts, then to overlapping cell, the reconstruction opening and closing operations in the present embodiment use matlab function
It can be realized；
S52: the foreground object of image is marked, with DAB staining cell cut zone to be processed in step S51 into
Row superposition is set as prospect label to give image nuclear centers one local minimum, watershed algorithm will be remained
Remaining overlapping cell segmentation, which comes out, obtains nucleus third area image L3；
S6: as shown in Fig. 2, by nucleus first area image L1, nucleus second area image L2 and nucleus third
Area image L3 and immunohistochemistry pathological image original image carry out mask process, and gained is the figure for the nucleus divided
Picture is finally added contour line to cell nuclear periphery and facilitates observation, and each of the good disconnected nuclear area of above-mentioned segmentation is given birth to
At boundary line.
The present embodiment also provides a kind of immunohistochemistry pathological image CD3 positive nucleus segmenting system, comprising: dyeing channel
Separation module, background removal module, image primary segmentation module, local threshold bernsen segmentation module and watershed segmentation mould
Block；
In the present embodiment, dyeing channel separation module be equipped with color warp product unit, color warp product unit for pair
Immunohistochemistry pathological image carries out color deconvolution, and dyeing channel is separated；Background removal module includes super-pixel segmentation unit
With kmeans cluster cell, super-pixel segmentation unit divides the image into irregular block of pixels, and kmeans cluster cell is used for
Distinguish background image；Image primary segmentation module is used for morphological feature image segmentation, obtains the good nucleus of primary segmentation the
One area image L1 and the first image C1 to be processed；Local threshold bernsen is divided module and is used for be processed first
Image C1 carries out local threshold bernsen segmentation, obtains nucleus second area image L2 and the second image C2 to be processed；Point
Water ridge segmentation module is used to the second image C2 to be processed carrying out prospect label and watershed algorithm is partitioned into nucleus third
Area image L3, nucleus first area image L1, nucleus second area image L2 and nucleus third area image L3 with
After immunohistochemistry pathological image original image carries out mask process, immunohistochemistry pathological image CD3 positive nucleus segmentation result is obtained
Image.
In the present embodiment, cpu parallel processing function is opened when handling pathological image, between each pathology figure mutually solely
Whether vertical, image processing speed is very fast, and establishes under data source directory and save and check file, locate for detection data
It managed while then last time processed progress continues.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, other any changes, modifications, substitutions, combinations, simplifications made without departing from the spirit and principles of the present invention,
It should be equivalent substitute mode, be included within the scope of the present invention.
Claims (10)
1. a kind of immunohistochemistry pathological image CD3 positive nucleus dividing method, which is characterized in that include the following steps:
S1: color deconvolution is carried out to immunohistochemistry pathological image, separates dyeing channel；
S2: being transformed into Lab color space from RGB color for preexisting immunity group pathological image, using super-pixel segmentation at
Irregular block of pixels, the irregular block of pixels carries out kmeans cluster and distinguishes image background, and image background is removed；
S3: based on morphological feature carry out image segmentation, obtain the good nucleus first area image L1 of primary segmentation and to
First image C1 of processing；
S4: the first image C1 to be processed is subjected to local threshold bernsen segmentation and morphological feature is divided, obtains dividing
Nucleus second area image L2 and the second image C2 to be processed；
S5: the second image C2 to be processed is subjected to prospect label, and nucleus third region is partitioned into using watershed algorithm
Image L3；
S6: it by nucleus first area image L1, nucleus second area image L2, nucleus third area image L3 and is immunized
Groupization pathological image original image carries out mask process, obtains nucleus segmented image.
2. immunohistochemistry pathological image CD3 positive nucleus dividing method according to claim 1, which is characterized in that step
Color deconvolution, calculation formula are carried out to immunohistochemistry pathological image described in rapid S1 are as follows:
C=M-1[y]
Wherein, C indicates that isolated H and DAB dyeing channel, M indicate that different dyeing parameter matrixes, the row of parameter matrix respectively indicate
H, eosin and DAB dyeing, the column of parameter matrix indicate RGB Staining Protocol parameter size, and y indicates the optical density of each pixel.
3. immunohistochemistry pathological image CD3 positive nucleus dividing method according to claim 1, which is characterized in that step
Irregular block of pixels described in rapid S2 carries out kmeans cluster and distinguishes image background, and the Lab color of irregular block of pixels is empty
Between the middle L layers object as kmeans cluster, specific calculation formula is as described below:
Wherein, E indicates least squares error, and x indicates each irregular block of pixels, and u indicates mass center, the cluster of g presentation class, when
Then kmeans cluster is completed when E minimum.
4. immunohistochemistry pathological image CD3 positive nucleus dividing method according to claim 1, which is characterized in that step
Image segmentation, specific steps are carried out based on morphological feature described in rapid S3 are as follows:
S31: by the image and DAB dyeing channel progress mask process after removal background, the DAB dyeing channel of removal background is obtained
Image, and use watershed algorithm segmented image；
S32: extracting the characteristics of image after watershed algorithm is divided, and removes the region of doubtful dust, retains the firstth area of nucleus
Area image L1 and the first image C1 to be processed.
5. immunohistochemistry pathological image CD3 positive nucleus dividing method according to claim 4, which is characterized in that institute
Stating characteristics of image includes gray average, contrast, compactness and elemental area, and it is small that the region of the doubtful dust is set as contrast
It is greater than 200 image-region in 0.04 or gray average, the nucleus first area image L1 is set as 40X amplification factor
Elemental area is less than 2000 and density is greater than 0.93 image-region, and the first image C1 to be processed is set as 40X times magnification
Several elemental areas is greater than 2000 image-region.
6. immunohistochemistry pathological image CD3 positive nucleus dividing method according to claim 1, which is characterized in that step
The first image C1 to be processed is subjected to local threshold bernsen segmentation described in rapid S4 and morphological feature is divided, specific steps
Are as follows:
The active window of S41: local threshold bernsen segmentation is set as 77*77 pixel size, after local threshold bernsen segmentation
Obtain the binary image of the first image C1 to be processed, and carry out exposure mask with DAB dyeing channel, obtain DAB dyeing channel to
The first nuclear area image C1 area image is handled, carries out watershed segmentation again after executing opening operation；
S42: the image handled by local threshold and watershed segmentation carries out the image segmentation based on morphological feature, extracts
Characteristics of image retains nucleus second area image L2 and the second image C2 to be processed.
7. immunohistochemistry pathological image CD3 positive nucleus dividing method according to claim 6, which is characterized in that institute
State nucleus second area image L2 be set as the elemental area of 40X amplification factor less than 2000 and density be greater than 0.93 image district
Domain, the elemental area that the second image C2 to be processed is set as 40X amplification factor are greater than 2000 image-region.
8. immunohistochemistry pathological image CD3 positive nucleus dividing method according to claim 1, which is characterized in that step
The second image C2 to be processed is subjected to prospect label described in rapid S5, and nucleus third area is partitioned into using watershed algorithm
Area image L3, specific steps are as follows:
S51: exposure mask is carried out with DAB dyeing channel after the second image C2 progress binary conversion treatment to be processed, is obtained to be processed
DAB staining cell cut zone using the opening and closing operations based on reconstruction and takes local maximum to carry out prospect label；
S52: the image after prospect label is overlapped with DAB staining cell cut zone to be processed in step S51, by prospect
Label is set as the local minimum of image nuclear centers, carries out image segmentation using fractional spins and obtains nucleus
Third area image L3.
9. immunohistochemistry pathological image CD3 positive nucleus dividing method according to claim 1, which is characterized in that also
Include the steps that edge line is arranged in nucleus, obtains the nuclear area edge setting in nucleus segmented image in step s 6
Boundary line, multiple summits borderline linking generate edge line.
10. a kind of immunohistochemistry pathological image CD3 positive nucleus segmenting system characterized by comprising dyeing channel separation
Module, background removal module, image primary segmentation module, local threshold bernsen segmentation module and watershed segmentation module；
The dyeing channel separation module is equipped with color warp product unit, and color warp product unit is used for immunohistochemistry pathology figure
As carrying out color deconvolution, dyeing channel is separated；
The background removal module includes super-pixel segmentation unit and kmeans cluster cell, and the super-pixel segmentation unit will scheme
As being divided into irregular block of pixels, the kmeans cluster cell is for distinguishing background image；
Described image primary segmentation module is used for morphological feature image segmentation, obtains the good nucleus first area of primary segmentation
Image L1 and the first image C1 to be processed；
The local threshold bernsen segmentation module is used to the first image C1 to be processed carrying out local threshold bernsen points
It cuts, obtains nucleus second area image L2 and the second image C2 to be processed；
The watershed segmentation module is used to the second image C2 to be processed carrying out prospect label and watershed algorithm is partitioned into
Nucleus third area image L3,
The nucleus first area image L1, nucleus second area image L2 and nucleus third area image L3 and immune
After groupization pathological image original image carries out mask process, immunohistochemistry pathological image CD3 positive nucleus segmentation result image is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910517113.1A CN110415255B (en) | 2019-06-14 | 2019-06-14 | Immunohistochemical pathological image CD3 positive cell nucleus segmentation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910517113.1A CN110415255B (en) | 2019-06-14 | 2019-06-14 | Immunohistochemical pathological image CD3 positive cell nucleus segmentation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110415255A true CN110415255A (en) | 2019-11-05 |
| CN110415255B CN110415255B (en) | 2021-08-20 |
Family
ID=68359139
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910517113.1A Active CN110415255B (en) | 2019-06-14 | 2019-06-14 | Immunohistochemical pathological image CD3 positive cell nucleus segmentation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110415255B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111275727A (en) * | 2020-01-19 | 2020-06-12 | 北方工业大学 | Cell segmentation method, cell segmentation apparatus, and computer-readable storage medium |
| CN111402267A (en) * | 2020-03-13 | 2020-07-10 | 中山大学孙逸仙纪念医院 | Segmentation method, device and terminal for epithelial cell nucleus in prostate cancer pathological image |
| CN111583185A (en) * | 2020-04-14 | 2020-08-25 | 山东省千佛山医院 | Ki67 cell nucleus counting method and system based on pathological immunohistochemistry |
| CN112614142A (en) * | 2020-12-25 | 2021-04-06 | 华侨大学 | Cell weak label manufacturing method and system based on multi-channel image fusion |
| CN113223097A (en) * | 2021-04-29 | 2021-08-06 | 武汉工程大学 | Image preprocessing method for improving density counting precision |
| CN113436206A (en) * | 2021-06-17 | 2021-09-24 | 易普森智慧健康科技（深圳）有限公司 | Pathological tissue section scanning area positioning method based on cluster segmentation |
| CN116645390A (en) * | 2023-07-27 | 2023-08-25 | 吉林省星博医疗器械有限公司 | Fluorescent image cell rapid segmentation method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104484877A (en) * | 2014-12-12 | 2015-04-01 | 山东大学 | AML cell segmentation method based on Meanshift cluster and morphological operations |
| CN105741266A (en) * | 2016-01-22 | 2016-07-06 | 北京航空航天大学 | Pathological image cell nucleus quick location method |
-
2019
- 2019-06-14 CN CN201910517113.1A patent/CN110415255B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104484877A (en) * | 2014-12-12 | 2015-04-01 | 山东大学 | AML cell segmentation method based on Meanshift cluster and morphological operations |
| CN105741266A (en) * | 2016-01-22 | 2016-07-06 | 北京航空航天大学 | Pathological image cell nucleus quick location method |
Non-Patent Citations (1)
| Title |
|---|
| WOLFGANG AICHINGER等: ""Automated cancer stem cell recognition in H and E stained tissue using convolutional neural networks and color deconvolution"", 《PROCEEDINGS OF SPIE》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111275727A (en) * | 2020-01-19 | 2020-06-12 | 北方工业大学 | Cell segmentation method, cell segmentation apparatus, and computer-readable storage medium |
| CN111402267A (en) * | 2020-03-13 | 2020-07-10 | 中山大学孙逸仙纪念医院 | Segmentation method, device and terminal for epithelial cell nucleus in prostate cancer pathological image |
| CN111402267B (en) * | 2020-03-13 | 2023-06-16 | 中山大学孙逸仙纪念医院 | Segmentation method, device and terminal of epithelial cell nuclei in prostate cancer pathological image |
| CN111583185A (en) * | 2020-04-14 | 2020-08-25 | 山东省千佛山医院 | Ki67 cell nucleus counting method and system based on pathological immunohistochemistry |
| CN112614142A (en) * | 2020-12-25 | 2021-04-06 | 华侨大学 | Cell weak label manufacturing method and system based on multi-channel image fusion |
| CN112614142B (en) * | 2020-12-25 | 2023-05-30 | 华侨大学 | Cell weak label manufacturing method and system based on multichannel image fusion |
| CN113223097A (en) * | 2021-04-29 | 2021-08-06 | 武汉工程大学 | Image preprocessing method for improving density counting precision |
| CN113436206A (en) * | 2021-06-17 | 2021-09-24 | 易普森智慧健康科技（深圳）有限公司 | Pathological tissue section scanning area positioning method based on cluster segmentation |
| CN113436206B (en) * | 2021-06-17 | 2022-03-15 | 易普森智慧健康科技（深圳）有限公司 | Pathological tissue section scanning area positioning method based on cluster segmentation |
| CN116645390A (en) * | 2023-07-27 | 2023-08-25 | 吉林省星博医疗器械有限公司 | Fluorescent image cell rapid segmentation method and system |
| CN116645390B (en) * | 2023-07-27 | 2023-10-03 | 吉林省星博医疗器械有限公司 | Fluorescent image cell rapid segmentation method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110415255B (en) | 2021-08-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110415255A (en) | A kind of immunohistochemistry pathological image CD3 positive nucleus dividing method and system | |
| CN109034208B (en) | High-low resolution combined cervical cell slice image classification system | |
| CN112070772B (en) | Blood leukocyte image segmentation method based on UNet++ and ResNet | |
| US11669971B2 (en) | Colony contrast gathering | |
| CN106056118B (en) | A kind of identification method of counting for cell | |
| CN109145941B (en) | Irregular cervical cell mass image classification method and system | |
| Dzyubachyk et al. | Advanced level-set-based cell tracking in time-lapse fluorescence microscopy | |
| CN105122308B (en) | System and method for using the multichannel biological marker of the structural unicellular division of continuous dyeing quantitative | |
| Shu et al. | Segmenting overlapping cell nuclei in digital histopathology images | |
| CN110400318B (en) | Immunohistochemical pathological image CD4 positive cell nucleus segmentation method and system | |
| CN109447998B (en) | Automatic segmentation method based on PCANet deep learning model | |
| CN109035269A (en) | A kind of cervical cell pathological section sick cell dividing method and system | |
| Gudla et al. | A high‐throughput system for segmenting nuclei using multiscale techniques | |
| US20200193139A1 (en) | Systems and methods for automated cell segmentation and labeling in immunofluorescence microscopy | |
| CN112991263B (en) | Method and equipment for improving TPS (tissue specific differentiation) calculation accuracy of PD-L1 immunohistochemical pathological section | |
| CN103679184A (en) | Method for leukocyte automatic identification based on relevant vector machine | |
| CN104992435A (en) | Cervix uteri single cell image segmentation algorithm | |
| CN108305253A (en) | A kind of pathology full slice diagnostic method based on more multiplying power deep learnings | |
| CN107730499A (en) | A kind of leucocyte classification method based on nu SVMs | |
| CN104732229A (en) | Segmentation method for overlapping cells in cervical smear image | |
| US20230273188A1 (en) | Method and system for detecting cell-killing efficacy and/or immune activity, and application thereof | |
| Jayapriya et al. | Comparative analysis of image segmentation techniques and its algorithm | |
| CN117670895B (en) | Immunohistochemical pathological image cell segmentation method based on section re-staining technology | |
| CN113450363B (en) | Meta-learning cell nucleus segmentation system and method based on label correction | |
| Lal et al. | A robust method for nuclei segmentation of H&E stained histopathology images |
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 |