CN114511514A - Cell segmentation method and device based on HE staining image - Google Patents

Cell segmentation method and device based on HE staining image Download PDF

Info

Publication number
CN114511514A
CN114511514A CN202210046519.8A CN202210046519A CN114511514A CN 114511514 A CN114511514 A CN 114511514A CN 202210046519 A CN202210046519 A CN 202210046519A CN 114511514 A CN114511514 A CN 114511514A
Authority
CN
China
Prior art keywords
image
binary
system parameters
preset
staining
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
Application number
CN202210046519.8A
Other languages
Chinese (zh)
Other versions
CN114511514B (en
Inventor
孙文灏
马义德
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Lanzhou University
Original Assignee
Lanzhou University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Lanzhou University filed Critical Lanzhou University
Priority to CN202210046519.8A priority Critical patent/CN114511514B/en
Publication of CN114511514A publication Critical patent/CN114511514A/en
Application granted granted Critical
Publication of CN114511514B publication Critical patent/CN114511514B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06T7/136—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
    • G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H70/00—ICT specially adapted for the handling or processing of medical references
    • G16H70/60—ICT specially adapted for the handling or processing of medical references relating to pathologies
    • 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
    • G06T2207/10061—Microscopic image from scanning electron microscope
    • 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Public Health (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Image Processing (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

The invention discloses a cell segmentation method and a cell segmentation device based on an HE staining image, which comprise the following steps: receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a first image; according to a preset calculation method, calculating to obtain a corresponding coupling result image according to the first image and a plurality of different first system parameters; all the first system parameters are obtained by training a plurality of initial system parameters which are randomly generated according to the first image; and receiving segmentation image parameters input by a user, and calculating to obtain a segmentation image corresponding to the HE staining image by combining the coupling result image and the HE staining image. The method comprises the steps of preprocessing an HE dyeing image to obtain a first image, and training system parameters by utilizing the first image to adapt to HE dyeing images with different dyeing styles. Meanwhile, the invention can also receive the image parameters input by the user, adjust the image to be output and obtain and output the segmentation image.

Description

Cell segmentation method and device based on HE staining image
Technical Field
The invention relates to the technical field of image segmentation, in particular to a cell segmentation method and device based on an HE staining image.
Background
The staining has great significance in the diagnosis of pathological tissue morphology, scientific experimental research and teaching work, if the tissue section is not stained, the internal structure of the tissue cannot be seen, and the cell nucleus cannot be distinguished, so that a judgment basis cannot be provided for researchers to judge whether the tissue is abnormal or not. Hematoxylin-eosin staining is the most basic staining method in the preparation of the slices and is extremely widely applied. Hematoxylin is a natural dye extracted from the hematoxylin tree and can stain the cell nucleus purple or blue, and eosin is an acidic cytoplasmic dye and can stain the cell cytoplasm, red cells, muscle tissues, eosin-like particles and the like red or pink to different degrees and is distinguished from the cell nucleus. In scientific research or teaching, tissue sections are stained, so that the contrast between tissues is more obvious, and the subsequent operations such as image segmentation and the like are facilitated, and then different areas of the tissue sections are divided.
In the prior art, there are many ways to segment a stained section image, and a deep learning cell segmentation method is usually adopted to segment cells, however, the method has a large demand on computing resources and high computing cost. In addition, for the conventional image segmentation algorithms, such as a K-means clustering algorithm (K-means), a fuzzy c-means clustering algorithm (FCM) and a maximum inter-class variance method (Otsu), although the problem of calculation cost is avoided to a certain extent, when the composition and the amount of a staining agent and the time and the temperature in the staining process change, the staining style of a hematoxylin-eosin stained section image (namely, an HE staining image) changes greatly, but the conventional image segmentation algorithm cannot solve the HE staining images with different staining styles, and has poor adaptability.
Disclosure of Invention
The invention provides a cell segmentation method and a cell segmentation device based on an HE staining image, and aims to solve the technical problem of how to perform adaptive adjustment on the cell segmentation method when the staining style of the HE staining image changes.
In order to solve the technical problem, an embodiment of the present invention provides a cell segmentation method based on an HE stained image, including:
receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a first image;
according to a preset calculation method, calculating to obtain a corresponding coupling result image according to the first image and a plurality of different first system parameters; all the first system parameters are obtained by training a plurality of initial system parameters which are randomly generated according to the first image;
and receiving segmentation image parameters input by a user, and calculating to obtain a segmentation image corresponding to the HE dye image by combining the coupling result image and the HE dye image.
Further, all the first system parameters are obtained by training a plurality of randomly generated initial system parameters according to the first image, specifically:
responding to the operation of starting system training by a user, and randomly generating a plurality of first parameter vectors; wherein one of said first parameter vectors comprises a plurality of different initial system parameters;
sequentially taking the initial system parameter in each first parameter vector as the first system parameter, and calculating to obtain a corresponding coupling result image;
according to a preset algorithm, calculating and obtaining the fitness corresponding to each first parameter vector by combining the coupling result image and the HE dye image, arranging all the first parameter vectors according to the sequence of the fitness from big to small, and selecting the first M first parameter vectors in the arrangement result as second parameter vectors;
and generating a plurality of third parameter vectors according to all the second parameter vectors in a preset mode, performing iterative optimization by taking the third parameter vectors as the first parameter vectors until the optimization times reach a preset number N, and taking the initial system parameters in the first parameter vectors with the highest current fitness as the first system parameters.
Further, according to a preset calculation method, according to the first image and a plurality of different first system parameters, calculating to obtain a corresponding coupling result image, specifically:
according to a preset sampling method, according to a plurality of different first system parameters, sampling the first image for a plurality of times to obtain a plurality of corresponding sampling result images;
calculating to obtain a plurality of corresponding binary images according to each sampling result image and the first image in sequence and by combining a plurality of different first system parameters; wherein one of the sampling result images corresponds to one of the binary images;
and selecting one binary image from a plurality of binary images as a final coupling result image according to a preset selection rule.
Further, the calculating, according to each of the sampling result images and the first image in sequence and in combination with a plurality of different first system parameters, to obtain a plurality of corresponding binary images specifically includes:
sequentially processing each sampling result image, taking the current sampling result image as a first binary image, and calculating to obtain a corresponding second binary image according to the first binary image and the first image and by combining a plurality of different first system parameters;
calculating a difference value between the first binary image and the second binary image according to a preset calculation method;
if the difference value is larger than a preset value, taking the second binary image as the first binary image, performing iterative processing on the second binary image according to the current first binary image to obtain an updated second binary image, and taking the current second binary image as a final binary image until the difference value between the current first binary image and the current second binary image is smaller than or equal to the preset value;
and if the difference value is smaller than or equal to a preset value, ending the circulation, and taking the current second binary image as the final binary image.
Further, the receiving of the segmented image parameters input by the user and the calculation of the segmented image corresponding to the HE stained image by combining the coupling result image and the HE stained image are specifically as follows:
performing Hadamard product calculation on the coupling result image and the HE dyeing image to obtain an image to be output, wherein the cell nucleus area is in the original color, and the other areas are in black background;
and receiving segmentation image parameters input by a user, and adjusting the image to be output according to the segmentation image parameters to obtain a segmentation image corresponding to the HE dye image.
Further, the method comprises the steps of receiving an HE staining image uploaded by a user, preprocessing the HE staining image, and obtaining a first image, wherein the method specifically comprises the following steps:
receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a hue layer of the HE staining image as the first image;
or receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a gray image corresponding to the HE staining image as the first image.
In order to solve the same technical problem, the present invention also provides a cell segmentation apparatus based on HE stain image, comprising:
the system comprises a preprocessing module, a first image acquisition module and a second image acquisition module, wherein the preprocessing module is used for receiving an HE dyeing image uploaded by a user and preprocessing the HE dyeing image to obtain a first image;
the calculation processing module is used for calculating to obtain a corresponding coupling result image according to the first image and a plurality of different first system parameters according to a preset calculation method; all the first system parameters are obtained by training a plurality of initial system parameters which are randomly generated according to the first image;
and the result output module is used for receiving the segmented image parameters input by the user and calculating to obtain the segmented image corresponding to the HE stained image by combining the coupling result image and the HE stained image.
Further, the calculation processing module further includes:
the training unit is used for responding to the operation of starting system training by a user and randomly generating a plurality of first parameter vectors; wherein one of said first parameter vectors comprises a plurality of different initial system parameters; sequentially taking the initial system parameter in each first parameter vector as the first system parameter, and calculating to obtain a corresponding coupling result image; according to a preset algorithm, calculating and obtaining the fitness corresponding to each first parameter vector by combining the coupling result image and the HE dye image, arranging all the first parameter vectors according to the sequence of the fitness from big to small, and selecting the first M first parameter vectors in the arrangement result as second parameter vectors; generating a plurality of third parameter vectors according to all the second parameter vectors in a preset mode, performing iterative optimization by taking the third parameter vectors as the first parameter vectors until the optimization times reach a preset number N, and taking the initial system parameters in the first parameter vectors with the highest current fitness as the first system parameters;
the sampling unit is used for sampling the first image for a plurality of times according to a preset sampling method and a plurality of different first system parameters to obtain a plurality of corresponding sampling result images;
the cyclic iteration unit is used for sequentially processing each sampling result image, taking the current sampling result image as a first binary image, and calculating to obtain a corresponding second binary image according to the first binary image and the first image and by combining a plurality of different first system parameters; calculating a difference value between the first binary image and the second binary image according to a preset calculation method; if the difference value is larger than a preset value, taking the second binary image as the first binary image, performing iterative processing on the second binary image according to the current first binary image to obtain an updated second binary image, and taking the current second binary image as a final binary image until the difference value between the current first binary image and the current second binary image is smaller than or equal to the preset value; if the difference value is smaller than or equal to a preset value, ending the circulation, and taking the current second binary image as the final binary image; wherein one of the sampling result images corresponds to one of the binary images;
and the selecting unit is used for selecting one binary image from a plurality of binary images as the final coupling result image according to a preset selecting rule.
Further, the result output module further includes:
the arithmetic unit is used for carrying out Hadamard product calculation on the coupling result image and the HE dyeing image to obtain an image to be output, wherein the cell nucleus area is in the original color, and the other areas are in the black background;
and the adjusting unit is used for receiving the segmentation image parameters input by the user and adjusting the image to be output according to the segmentation image parameters to obtain a segmentation image corresponding to the HE dye image.
Further, the preprocessing module further comprises:
the first preprocessing unit is used for receiving the HE dyeing image uploaded by a user and preprocessing the HE dyeing image to obtain a hue layer of the HE dyeing image as the first image;
and the second preprocessing unit is used for receiving the HE staining image uploaded by the user and preprocessing the HE staining image to obtain a gray image corresponding to the HE staining image as the first image.
Compared with the prior art, the embodiment of the invention has the following beneficial effects:
the invention provides a cell segmentation method and a cell segmentation device based on an HE staining image. Meanwhile, the corresponding coupling result image is obtained through calculation by combining the first image and the system parameters obtained through training, the segmentation image parameters input by the user can be received, the segmentation image corresponding to the uploaded HE dye image is adjusted and output, the experience of the user is improved, and the diversified requirements of the user are met.
Further, iterative circulation is carried out on the second binary image according to the first image to obtain a final binary image, and the cell segmentation accuracy of the HE stained image is improved.
Drawings
FIG. 1: the invention provides a flow chart schematic diagram of an embodiment of a cell segmentation method based on an HE staining image;
FIG. 2 is a schematic diagram: the invention provides a structural schematic diagram of a cell segmentation device based on an HE staining image;
FIG. 3: the invention provides a structural schematic diagram of a preprocessing module of a cell segmentation device based on an HE staining image;
FIG. 4: the invention provides a structural schematic diagram of a calculation processing module of a cell segmentation device based on an HE staining image;
FIG. 5: the invention provides a structural schematic diagram of a result output module of a cell segmentation device based on an HE staining image.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The first embodiment is as follows:
referring to fig. 1, a cell segmentation method based on HE stained image according to an embodiment of the present invention includes steps S1 to S3, and the steps are as follows:
step S1: and receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a first image.
Further, step S1 is specifically:
receiving an HE dyeing image uploaded by a user, and preprocessing the HE dyeing image to obtain a hue layer of the HE dyeing image as a first image;
or receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a gray image corresponding to the HE staining image as a first image.
In this embodiment, a user uploads an HE dyed image to be processed to a card computer through upper computer software, and since the HE dyed image is usually a color image in an RGB format and is inconvenient for subsequent processing, the HE dyed image is preprocessed according to actual requirements, converted into an HSV format image, and a hue layer in the HSV format image is acquired, or converted into a corresponding grayscale image.
It should be noted that the card computer is connected to the upper computer software through a local lan network, so as to implement data interaction between the card computer and the upper computer software.
Step S2: according to a preset calculation method, calculating to obtain a corresponding coupling result image according to the first image and a plurality of different first system parameters; all the first system parameters are obtained by training a plurality of randomly generated initial system parameters according to the first image.
Further, all the first system parameters in step S2 are obtained by training several randomly generated initial system parameters according to the first image, and specifically include steps S21 to S24, where the steps are specifically as follows:
step S21: responding to the operation of starting system training by a user, and randomly generating a plurality of first parameter vectors; wherein a first parameter vector comprises a plurality of different initial system parameters.
In this embodiment, at the beginning of the parameter training, 200 first parameter vectors V are randomly generated, each comprising a plurality of different initial system parameters (V), with reference to equation (1)E、VL、σ、n、αE(n), ε (n), P (n)), and initial system parameters (α) in all parameter vectors vE(n)、VE、VLσ, and P) are all constrained by the conditions expressed by the expressions (2) to (6).
Figure BDA0003470929290000071
∑αE(n)＝1 (2)
Smax＜VE≤2*Smax (3)
0＜VL≤10 (4)
0＜α≤2 (5)
0＜P≤1 (6)
Wherein S ismaxRepresents the maximum value of the pixel values of the first image, and S is the color tone layer of the HE dye imagemaxIf the first image is a grayscale image corresponding to the HE stain image, S is 360max＝255。
Step S22: and sequentially taking the initial system parameter in each first parameter vector as a first system parameter, and calculating to obtain a corresponding coupling result image.
Step S23: and according to a preset algorithm, combining the coupling result image and the HE dye image, calculating to obtain the fitness corresponding to each first parameter vector, arranging all the first parameter vectors according to the sequence of the fitness from large to small, and selecting the first M first parameter vectors in the arrangement result as second parameter vectors.
In this embodiment, the fitness f corresponding to each first parameter vector is calculated according to equation (7).
f＝exp(Mean-Dev-Q2) (7)
Wherein Mean is the Mean of the pixel values of the region corresponding to the nucleus region of the coupling result image in the HE staining image, Dev is the variance of the pixel values of the region corresponding to the nucleus region of the coupling result image in the HE staining image, and Q is the difference between the ratio of the number of the nucleus region pixels in the coupling result image to the total number of the nucleus region pixels in the coupling result image and an empirical constant, and the constant can be 0.183 generally.
In this embodiment, all the first parameter vectors are arranged according to the descending order of the fitness, and the first parameter vector with the lowest fitness of 80% is discarded according to the arrangement result, so as to obtain M first parameter vectors as the second parameter vector. The rejection rate of the first parameter vector can be adjusted according to actual requirements.
Step S24: and according to a preset mode, generating a plurality of third parameter vectors according to all the second parameter vectors, performing iterative optimization by taking the third parameter vectors as the first parameter vectors until the optimization times reach a preset number N, and taking the initial system parameters in the first parameter vectors with the highest current fitness as the first system parameters.
In this embodiment, for all the second parameter vectors, 200 third parameter vectors are regenerated in a one-to-one correspondence exchange, averaging and random replacement manner of the initial system parameters between every two second parameter vectors, and are used as the first parameter vectors to perform iterative optimization until the optimization times reach the preset times N, so that the initial system parameters with high fitness are obtained as the first system parameters, and the cell segmentation accuracy of the HE stained image is improved. Wherein, N is the empirical times obtained by a large number of experimental statistics. As an example, N-25.
In this embodiment, when the components of the stain are different, or whether the stain is oxidized, or details of the process such as the staining time and the staining environment temperature are changed, the HE staining image may be changed accordingly, and if the first image corresponding to the HE staining image is sampled and segmented by using the original first system parameters, the segmentation effect may be affected. At the moment, the first system parameters are trained and updated by using the complete image or the local image of the current HE staining image, so that new first system parameters are obtained to adapt to the current staining style, and the segmentation effect of the image is ensured.
Further, in step S2, according to a preset calculation method, according to the first image and a plurality of different first system parameters, calculating to obtain a corresponding coupling result image, specifically including steps S25 to S27, where each step specifically includes:
step S25: according to a preset sampling method, sampling the first image for a plurality of times according to a plurality of different first system parameters to obtain a plurality of corresponding sampling result images.
In this embodiment, the sampling thresholds E (n-2), E (n-1) and the first system parameter (α) are combinedE(n), p (n), sampling the input first image S according to the formula (8) to obtain a corresponding sampling result image y (n).
Figure BDA0003470929290000091
Wherein S represents a tone layer of the HE dye image or a gray image corresponding to the HE dye image, namely S represents a first image; e (n) denotes a sampling threshold; n represents the number of cycles; y (n) represents a sampling result image; alpha is alphaE(n) is the step length of the sampling threshold; p (n) represents a duty ratio coefficient of the sampling pulse; is a multiplication symbol, i.e. P (n) αE(n) represents P (n) and alphaE(n) multiplication. In addition, n has an initial value of 0, and E (-1) and E (-2) are both zero matrices.
In this embodiment, when the first image is sampled for the first time, the input first image S is sampled according to the formula (8) by combining the preset parameters E (-1), E (-2) and the first system parameter, and in the subsequent sampling process, E (0), E (1), …, E (n) are also needed, and may be calculated by the formula (9).
E(n)＝E(n-1)-αE(n)+VE*y(n) (9)
Wherein, VEFor controlling the number of samplings.
Step S26: calculating to obtain a plurality of corresponding binary images according to each sampling result image and each first image and combining a plurality of different first system parameters; wherein, one sampling result image corresponds to one binary image.
In the present embodiment, a white region (pixel value is 1) in the binary image is a cell nucleus region, and a black region (pixel value is 0) is a non-cell nucleus region.
Further, step S26 specifically includes step S261 to step S264, and each step specifically includes the following steps:
step S261: and sequentially processing each sampling result image, taking the current sampling result image as a first binary image, and calculating to obtain a corresponding second binary image according to the first binary image and the first image and by combining a plurality of different first system parameters.
In this embodiment, the Y (n) calculated in step S25 is used as the initial value of Y in equation (10), i.e., the first binary image Y (m-1), the intermediate value l (m) is calculated, and l (m) is substituted for equation (11), and the corresponding second binary image Y (m) is calculated by combining with the sampling threshold E (n-1).
Figure BDA0003470929290000101
Figure BDA0003470929290000102
Wherein the content of the first and second substances,
Figure BDA0003470929290000103
representing convolution calculation, l (m) is a calculation intermediate value, e (n) is a sampling threshold in equation (8), Y (m-1) is a first binary image corresponding to each cycle, Y (m) is a second binary image corresponding to each cycle, and W is used to control the intensity and range of a growing region (i.e., a region of the second binary image). W can be calculated from equation (13). Further, the initial value of m is 0.
Step S262: and calculating a difference value between the first binary image and the second binary image according to a preset calculation method.
In this embodiment, Y (m-1) and Y (m) in step S261 are substituted for equation (12) to calculate the difference value ∈ (m) between the first binary image and the second binary image.
Figure BDA0003470929290000111
Where i and j represent the coordinates of the matrix elements and ε (m) is the loop-stop condition.
Figure BDA0003470929290000112
Wherein, VLAnd σ are used to control the intensity and extent, respectively, of the growing region (i.e., the region of the second binary image).
It should be noted that, if the difference value is greater than the preset value, step S263 is performed, and if the difference value is less than or equal to the preset value, step S264 is performed.
Step S263: and taking the second binary image as a first binary image, performing iterative processing on the second binary image according to the current first binary image to obtain an updated second binary image, and taking the current second binary image as a final binary image until the difference value between the current first binary image and the current second binary image is less than or equal to a preset value.
Step S264: and ending the loop, and taking the current second binary image as a final binary image.
In the present embodiment, the preset value is 10-6And judging whether to end the current cycle or not according to the relation between the difference value epsilon (m) and a preset value. If the current difference value epsilon (m) is more than 10-6And performing next iteration cycle on the second binary image according to the current first binary image until the difference value epsilon (m) between the current first binary image and the current second binary image is less than or equal to 10-6Then, the current second binary image Y (m) is used as the final binary image Y.
Step S27: and selecting a binary image from the plurality of binary images as a final coupling result image according to a preset selection rule.
Step S3: and receiving segmentation image parameters input by a user, and calculating to obtain a segmentation image corresponding to the HE staining image by combining the coupling result image and the HE staining image.
Further, step S3 specifically includes step S31 to step S32, and each step specifically includes the following steps:
step S31: performing Hadamard product calculation on the coupling result image and the HE dyeing image to obtain an image to be output, wherein the cell nucleus area is in the original color, and the other areas are in black background;
step S32: and receiving segmentation image parameters input by a user, and adjusting the image to be output according to the segmentation image parameters to obtain a segmentation image corresponding to the HE dye image.
In this embodiment, the segmented image corresponding to the HE stain image is obtained and output by receiving the segmented image parameters input by the user and adjusting parameters such as the size of the image to be output, the color of the cell nucleus area, the background color and the like.
In order to solve the same technical problem, referring to fig. 2, the present invention further provides a cell segmentation apparatus based on HE stain image, comprising:
the system comprises a preprocessing module 1, a first image acquisition module and a second image acquisition module, wherein the preprocessing module 1 is used for receiving an HE dyeing image uploaded by a user and preprocessing the HE dyeing image to obtain a first image;
the calculation processing module 2 is used for calculating to obtain a corresponding coupling result image according to a preset calculation method and the first image and a plurality of different first system parameters; all the first system parameters are obtained by training a plurality of initial system parameters which are randomly generated according to the first image;
and the result output module 3 is used for receiving the segmented image parameters input by the user, and calculating to obtain the segmented image corresponding to the HE stained image by combining the binary image and the HE stained image.
Further, referring to fig. 3, the preprocessing module 1 further includes:
the first preprocessing unit is used for receiving the HE dyeing image uploaded by a user and preprocessing the HE dyeing image to obtain a tone layer of the HE dyeing image as a first image;
and the second preprocessing unit is used for receiving the HE dyeing image uploaded by the user and preprocessing the HE dyeing image to obtain a gray image corresponding to the HE dyeing image as the first image.
Further, referring to fig. 4, the calculation processing module 2 further includes:
the training unit is used for responding to the operation of starting system training by a user and randomly generating a plurality of first parameter vectors; wherein a first parameter vector comprises a plurality of different initial system parameters; sequentially taking the initial system parameter in each first parameter vector as a first system parameter, and calculating to obtain a corresponding coupling result image; calculating to obtain the fitness corresponding to each first parameter vector according to a preset algorithm by combining the coupling result image and the HE dyeing image, arranging all the first parameter vectors according to the sequence of the fitness from large to small, and selecting the first M first parameter vectors in the arrangement result as second parameter vectors; generating a plurality of third parameter vectors according to all the second parameter vectors in a preset mode, performing iterative optimization by taking the third parameter vectors as the first parameter vectors until the optimization times reach a preset number N, and taking initial system parameters in the first parameter vectors with highest current fitness as first system parameters;
the sampling unit is used for sampling the first image for a plurality of times according to a plurality of different first system parameters according to a preset sampling method to obtain a plurality of corresponding sampling result images;
the loop iteration unit is used for processing each sampling result image in sequence, taking the current sampling result image as a first binary image, and calculating to obtain a corresponding second binary image according to the first binary image and the first image by combining a plurality of different first system parameters; calculating a difference value between the first binary image and the second binary image according to a preset calculation method; if the difference value is larger than the preset value, taking the second binary image as a first binary image, performing iterative processing on the second binary image according to the current first binary image to obtain an updated second binary image, and taking the current second binary image as a final binary image until the difference value between the current first binary image and the current second binary image is smaller than or equal to the preset value; if the difference value is smaller than or equal to the preset value, ending the circulation, and taking the current second binary image as a final binary image; wherein, one sampling result image corresponds to one binary image;
and the selecting unit is used for selecting one binary image from the plurality of binary images as a final coupling result image according to a preset selecting rule.
Further, referring to fig. 5, the result output module 3 further includes:
the arithmetic unit is used for carrying out Hadamard product calculation on the coupling result image and the HE dyeing image to obtain an image to be output, wherein the cell nucleus area is in the original color, and the other areas are in the black background;
and the adjusting unit is used for receiving the segmented image parameters input by the user and adjusting the image to be output according to the segmented image parameters to obtain the segmented image corresponding to the HE dye image.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working process of the apparatus described above may refer to the corresponding process in the foregoing method embodiment, and is not described herein again.
Compared with the prior art, the embodiment of the invention has the following beneficial effects:
the invention provides a cell segmentation method and a cell segmentation device based on an HE staining image. Meanwhile, the corresponding coupling result image is obtained through calculation by combining the first image and the system parameters obtained through training, the segmentation image parameters input by the user can be received, the segmentation image corresponding to the uploaded HE dye image is adjusted and output, the experience of the user is improved, and the diversified requirements of the user are met.
Further, iterative circulation is carried out on the second binary image according to the first image to obtain a final binary image, and the cell segmentation accuracy of the HE stained image is improved.
The above-mentioned embodiments are provided to further explain the objects, technical solutions and advantages of the present invention in detail, and it should be understood that the above-mentioned embodiments are only examples of the present invention and are not intended to limit the scope of the present invention. It should be understood that any modifications, equivalents, improvements and the like, which come within the spirit and principle of the invention, may occur to those skilled in the art and are intended to be included within the scope of the invention.

Claims (10)

1. A cell segmentation method based on an HE staining image is characterized by comprising the following steps:
receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a first image;
according to a preset calculation method, calculating to obtain a corresponding coupling result image according to the first image and a plurality of different first system parameters; all the first system parameters are obtained by training a plurality of initial system parameters which are randomly generated according to the first image;
and receiving segmentation image parameters input by a user, and calculating to obtain a segmentation image corresponding to the HE dye image by combining the coupling result image and the HE dye image.
2. A cell segmentation method based on HE stain images as claimed in claim 1, wherein all the first system parameters are obtained by training several randomly generated initial system parameters from the first image, specifically:
responding to the operation of starting system training by a user, and randomly generating a plurality of first parameter vectors; wherein one of said first parameter vectors comprises a plurality of different initial system parameters;
sequentially taking the initial system parameter in each first parameter vector as the first system parameter, and calculating to obtain a corresponding coupling result image;
according to a preset algorithm, calculating and obtaining the fitness corresponding to each first parameter vector by combining the coupling result image and the HE dye image, arranging all the first parameter vectors according to the sequence of the fitness from big to small, and selecting the first M first parameter vectors in the arrangement result as second parameter vectors;
and generating a plurality of third parameter vectors according to all the second parameter vectors in a preset mode, performing iterative optimization by taking the third parameter vectors as the first parameter vectors until the optimization times reach a preset number N, and taking the initial system parameters in the first parameter vectors with the highest current fitness as the first system parameters.
3. A cell segmentation method based on an HE stain image according to claim 1, wherein the corresponding coupling result image is obtained by calculation according to a preset calculation method based on the first image and a plurality of different first system parameters, specifically:
according to a preset sampling method, according to a plurality of different first system parameters, sampling the first image for a plurality of times to obtain a plurality of corresponding sampling result images;
calculating to obtain a plurality of corresponding binary images according to each sampling result image and the first image in sequence and by combining a plurality of different first system parameters; wherein one of the sampling result images corresponds to one of the binary images;
and selecting one binary image from a plurality of binary images as a final coupling result image according to a preset selection rule.
4. The method for cell segmentation based on the HE stained image according to claim 3, wherein the plurality of corresponding binary images are obtained by calculation according to each of the sampling result image and the first image in sequence and by combining a plurality of different first system parameters, specifically:
sequentially processing each sampling result image, taking the current sampling result image as a first binary image, and calculating to obtain a corresponding second binary image according to the first binary image and the first image and by combining a plurality of different first system parameters;
calculating a difference value between the first binary image and the second binary image according to a preset calculation method;
if the difference value is larger than a preset value, taking the second binary image as the first binary image, performing iterative processing on the second binary image according to the current first binary image to obtain an updated second binary image, and taking the current second binary image as a final binary image until the difference value between the current first binary image and the current second binary image is smaller than or equal to the preset value;
and if the difference value is smaller than or equal to a preset value, ending the circulation, and taking the current second binary image as the final binary image.
5. The method for cell segmentation based on the HE stain image according to claim 1, wherein the segmented image parameters input by the user are received, and the segmented image corresponding to the HE stain image is obtained through calculation by combining the coupling result image and the HE stain image, specifically:
performing Hadamard product calculation on the coupling result image and the HE dyeing image to obtain an image to be output, wherein the cell nucleus area is in the original color, and the other areas are in black background;
and receiving segmentation image parameters input by a user, and adjusting the image to be output according to the segmentation image parameters to obtain a segmentation image corresponding to the HE dye image.
6. The method for cell segmentation based on the HE stain image according to claim 1, wherein the HE stain image uploaded by a user is received and preprocessed to obtain a first image, specifically:
receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a hue layer of the HE staining image as the first image;
or receiving an HE staining image uploaded by a user, and preprocessing the HE staining image to obtain a gray image corresponding to the HE staining image as the first image.
7. A cell segmentation apparatus based on an HE stain image, comprising:
the system comprises a preprocessing module, a first image acquisition module and a second image acquisition module, wherein the preprocessing module is used for receiving an HE dyeing image uploaded by a user and preprocessing the HE dyeing image to obtain a first image;
the calculation processing module is used for calculating to obtain a corresponding coupling result image according to the first image and a plurality of different first system parameters according to a preset calculation method; all the first system parameters are obtained by training a plurality of initial system parameters which are randomly generated according to the first image;
and the result output module is used for receiving the segmented image parameters input by the user and calculating to obtain the segmented image corresponding to the HE stained image by combining the coupling result image and the HE stained image.
8. The HE stain image-based cell segmentation apparatus of claim 7, wherein the computational processing module further comprises:
the training unit is used for responding to the operation of starting system training by a user and randomly generating a plurality of first parameter vectors; wherein one of said first parameter vectors comprises a plurality of different initial system parameters; sequentially taking the initial system parameter in each first parameter vector as the first system parameter, and calculating to obtain a corresponding coupling result image; according to a preset algorithm, calculating and obtaining the fitness corresponding to each first parameter vector by combining the coupling result image and the HE dye image, arranging all the first parameter vectors according to the sequence of the fitness from big to small, and selecting the first M first parameter vectors in the arrangement result as second parameter vectors; generating a plurality of third parameter vectors according to all the second parameter vectors in a preset mode, performing iterative optimization by taking the third parameter vectors as the first parameter vectors until the optimization times reach a preset number N, and taking the initial system parameters in the first parameter vectors with the highest current fitness as the first system parameters;
the sampling unit is used for sampling the first image for a plurality of times according to a preset sampling method and a plurality of different first system parameters to obtain a plurality of corresponding sampling result images;
the cyclic iteration unit is used for sequentially processing each sampling result image, taking the current sampling result image as a first binary image, and calculating to obtain a corresponding second binary image according to the first binary image and the first image and by combining a plurality of different first system parameters; calculating a difference value between the first binary image and the second binary image according to a preset calculation method; if the difference value is larger than a preset value, taking the second binary image as the first binary image, performing iterative processing on the second binary image according to the current first binary image to obtain an updated second binary image, and taking the current second binary image as a final binary image until the difference value between the current first binary image and the current second binary image is smaller than or equal to the preset value; if the difference value is smaller than or equal to a preset value, ending the circulation, and taking the current second binary image as the final binary image; wherein one of the sampling result images corresponds to one of the binary images;
and the selecting unit is used for selecting one binary image from a plurality of binary images as the final coupling result image according to a preset selecting rule.
9. The cell segmentation apparatus based on the HE stain image according to claim 7, wherein the result output module further comprises:
the arithmetic unit is used for carrying out Hadamard product calculation on the coupling result and the HE dyeing image to obtain an image to be output, wherein the cell nucleus area is in the original color, and the other areas are in the black background;
and the adjusting unit is used for receiving the segmented image parameters input by the user and adjusting the image to be output according to the segmented image parameters to obtain the segmented image corresponding to the HE dye image.
10. The HE stain image-based cell segmentation apparatus of claim 7, wherein the pre-processing module further comprises:
the first preprocessing unit is used for receiving the HE dyeing image uploaded by a user and preprocessing the HE dyeing image to obtain a hue layer of the HE dyeing image as the first image;
and the second preprocessing unit is used for receiving the HE staining image uploaded by the user and preprocessing the HE staining image to obtain a gray image corresponding to the HE staining image as the first image.
CN202210046519.8A 2022-01-14 2022-01-14 HE staining image-based cell segmentation method and device Active CN114511514B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210046519.8A CN114511514B (en) 2022-01-14 2022-01-14 HE staining image-based cell segmentation method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210046519.8A CN114511514B (en) 2022-01-14 2022-01-14 HE staining image-based cell segmentation method and device

Publications (2)

Publication Number Publication Date
CN114511514A true CN114511514A (en) 2022-05-17
CN114511514B CN114511514B (en) 2023-07-21

Family

ID=81549537

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210046519.8A Active CN114511514B (en) 2022-01-14 2022-01-14 HE staining image-based cell segmentation method and device

Country Status (1)

Country Link
CN (1) CN114511514B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103473739A (en) * 2013-08-15 2013-12-25 华中科技大学 White blood cell image accurate segmentation method and system based on support vector machine
US20190042826A1 (en) * 2017-08-04 2019-02-07 Oregon Health & Science University Automatic nuclei segmentation in histopathology images
CN110246116A (en) * 2019-04-24 2019-09-17 创新工场(北京)企业管理股份有限公司 Digital pathological section dyes the computer automatic generation method dyed to IHC by HE
CN111210447A (en) * 2020-01-10 2020-05-29 福建师范大学 Method and terminal for hierarchical segmentation of hematoxylin-eosin staining pathological image
CN112132843A (en) * 2020-09-30 2020-12-25 福建师范大学 Hematoxylin-eosin staining pathological image segmentation method based on unsupervised deep learning
CN113237881A (en) * 2021-05-10 2021-08-10 博奥生物集团有限公司 Method and device for detecting specific cells and pathological section detection system
CN113256577A (en) * 2021-05-18 2021-08-13 湖南医药学院 Cancer auxiliary analysis system and device based on HE staining pathological image

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103473739A (en) * 2013-08-15 2013-12-25 华中科技大学 White blood cell image accurate segmentation method and system based on support vector machine
US20190042826A1 (en) * 2017-08-04 2019-02-07 Oregon Health & Science University Automatic nuclei segmentation in histopathology images
CN110246116A (en) * 2019-04-24 2019-09-17 创新工场(北京)企业管理股份有限公司 Digital pathological section dyes the computer automatic generation method dyed to IHC by HE
CN111210447A (en) * 2020-01-10 2020-05-29 福建师范大学 Method and terminal for hierarchical segmentation of hematoxylin-eosin staining pathological image
CN112132843A (en) * 2020-09-30 2020-12-25 福建师范大学 Hematoxylin-eosin staining pathological image segmentation method based on unsupervised deep learning
CN113237881A (en) * 2021-05-10 2021-08-10 博奥生物集团有限公司 Method and device for detecting specific cells and pathological section detection system
CN113256577A (en) * 2021-05-18 2021-08-13 湖南医药学院 Cancer auxiliary analysis system and device based on HE staining pathological image

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
马义德 等: "一种基于脉冲耦合神经网络的植物胚性细胞图像分割", 《科学通报》, vol. 46, no. 21, pages 1781 - 1786 *

Also Published As

Publication number Publication date
CN114511514B (en) 2023-07-21

Similar Documents

Publication Publication Date Title
Lin et al. DeepFL-IQA: Weak supervision for deep IQA feature learning
WO2022199583A1 (en) Image processing method and apparatus, computer device, and storage medium
CN107679466B (en) Information output method and device
CN110163884B (en) Single image segmentation method based on full-connection deep learning neural network
CN109359681A (en) A kind of field crop pest and disease disasters recognition methods based on the full convolutional neural networks of improvement
CN112614060A (en) Method and device for rendering human face image hair, electronic equipment and medium
CN109858480A (en) A kind of digital instrument recognition methods
CN108172278B (en) HE staining pathological image color normalization method
CN109376787B (en) Manifold learning network and computer vision image set classification method based on manifold learning network
CN112750106A (en) Nuclear staining cell counting method based on incomplete marker deep learning, computer equipment and storage medium
CN107609603A (en) A kind of image matching method of multiple color spaces difference fusion
CN113420794B (en) Binaryzation Faster R-CNN citrus disease and pest identification method based on deep learning
CN107516083A (en) A kind of remote facial image Enhancement Method towards identification
CN115526803A (en) Non-uniform illumination image enhancement method, system, storage medium and device
CN114511514A (en) Cell segmentation method and device based on HE staining image
CN113221913A (en) Agriculture and forestry disease and pest fine-grained identification method and device based on Gaussian probability decision-level fusion
CN111429416A (en) Face pigment spot identification method and device and electronic equipment
Zhao et al. Automatic windowing for mri with convolutional neural network
CN112270220B (en) Sewing gesture recognition method based on deep learning
Zhang et al. Single image dehazing via reinforcement learning
CN107729823B (en) Online human-computer interaction method based on E-SOINN network
CN109670508A (en) A kind of cloud atlas segmentation network and its method intensively connecting full convolutional network based on symmetrical expression
CN112037293A (en) Method and device for analyzing color of coating sample plate
CN111680760A (en) Clothing style identification method and device, electronic equipment and storage medium
CN112446321B (en) Hand motion real-time identification method based on frame difference 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
GR01 Patent grant
GR01 Patent grant