WO2021027152A1 - 基于条件生成对抗网络合成图像的方法及相关设备 - Google Patents

基于条件生成对抗网络合成图像的方法及相关设备 Download PDF

Info

Publication number
WO2021027152A1
WO2021027152A1 PCT/CN2019/117988 CN2019117988W WO2021027152A1 WO 2021027152 A1 WO2021027152 A1 WO 2021027152A1 CN 2019117988 W CN2019117988 W CN 2019117988W WO 2021027152 A1 WO2021027152 A1 WO 2021027152A1
Authority
WO
WIPO (PCT)
Prior art keywords
red blood
blood cell
image
data set
shape data
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.)
Ceased
Application number
PCT/CN2019/117988
Other languages
English (en)
French (fr)
Chinese (zh)
Inventor
王义文
王健宗
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.)
Ping An Technology Shenzhen Co Ltd
Original Assignee
Ping An Technology Shenzhen Co Ltd
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 Ping An Technology Shenzhen Co Ltd filed Critical Ping An Technology Shenzhen Co Ltd
Priority to US17/264,312 priority Critical patent/US11636695B2/en
Priority to JP2021506747A priority patent/JP7329041B2/ja
Priority to SG11202100136SA priority patent/SG11202100136SA/en
Publication of WO2021027152A1 publication Critical patent/WO2021027152A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/60—Type of objects
    • G06V20/69—Microscopic objects, e.g. biological cells or cellular parts
    • G06V20/695—Preprocessing, e.g. image segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2132—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on discrimination criteria, e.g. discriminant analysis
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G06F18/2137—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on criteria of topology preservation, e.g. multidimensional scaling or self-organising maps
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
    • G06F18/2193—Validation; Performance evaluation; Active pattern learning techniques based on specific statistical tests
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
    • G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
    • 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
    • G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/40—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for data related to laboratory analysis, e.g. patient specimen analysis
    • 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
    • G16H30/00—ICT specially adapted for the handling or processing of medical images
    • G16H30/40—ICT specially adapted for the handling or processing of medical images for processing medical images, e.g. editing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20076—Probabilistic image processing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30024—Cell structures in vitro; Tissue sections in vitro

Definitions

  • This application relates to the field of neural networks, and in particular to a method and related equipment for generating an adversarial network synthesis image based on conditions.
  • the present application provides a method and related equipment for generating an adversarial network synthetic image based on conditions, which can solve the problem of cumbersome clinical data processing of different medical institutions in the prior art.
  • this application provides a method for generating an adversarial network synthetic image based on a condition, the method including:
  • each red blood cell image includes multiple red blood cells, and the shape and position of the red blood cells on the same red blood cell image may be the same or different;
  • red blood cell shape data Set is used to provide red blood cell shape data when synthesizing red blood cell images
  • the red blood cell shape data is collected into a plurality of red blood cell images.
  • the present application provides an apparatus for synthesizing images, which has the function of realizing the method corresponding to the condition-based generation of an adversarial network synthetic image provided in the first aspect.
  • the function can be realized by hardware, or by hardware executing corresponding software.
  • the hardware or software includes one or more modules corresponding to the above functions, and the modules may be software and/or hardware.
  • the device includes:
  • the input and output module is used to obtain multiple clinical red blood cell images, divide the red blood cells with different shapes and different positions in each red blood cell image into multiple sub-masks, and multiple sub-masks corresponding to each red blood cell image Perform synthesis processing to generate a mask to obtain multiple masks corresponding to multiple red blood cell images; wherein each red blood cell image includes multiple red blood cells, and the shape and position of the red blood cells on the same red blood cell image can be the same or different;
  • the processing module is used to collect multiple red blood cell shape data from multiple masks through the input and output module to obtain a training data set, calculate the segmentation boundary of each red blood cell in the training data set, and according to the segmentation boundary of each red blood cell
  • the red blood cell shape data set is established; the red blood cell shape data set is used to provide red blood cell shape data when synthesizing red blood cell images; the distribution data of each red blood cell in the red blood cell shape data set is collected through the input and output module; and the red blood cell shape The data is assembled into multiple red blood cell images.
  • the third aspect of the present application provides a computer device, which includes at least one connected processor, a memory, and a transceiver, wherein the memory is used to store program code, and the processor is used to call the program code in the memory To perform the method described in the first aspect above.
  • a fourth aspect of the present application provides a computer storage medium, the storage medium stores computer instructions, which when run on a computer, cause the computer to execute the method described in the first aspect.
  • red blood cell images are acquired, and red blood cells of different shapes and positions in each red blood cell image are divided into multiple sub-masks, and each red blood cell image corresponds to Multiple sub-masks are synthesized to generate a mask to obtain multiple masks corresponding to multiple red blood cell images; multiple red blood cell shape data are collected from multiple masks to obtain a training data set, and each of the training data sets is calculated
  • the red blood cell segmentation boundary is to establish a red blood cell shape data set according to the segmentation boundary of each red blood cell; collect the distribution data of each red blood cell in the red blood cell shape data set; collect the red blood cell shape data into multiple red blood cell images.
  • FIG. 1 is a schematic flowchart of a method for generating an adversarial network synthetic image based on conditions in an embodiment of the application;
  • FIG. 2 is a schematic diagram of a structure of the apparatus 20 for synthesizing images in an embodiment of the application;
  • FIG. 3 is a schematic diagram of a structure of a computer device in an embodiment of the application.
  • a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to the clearly listed Those steps or modules may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or equipment.
  • the division of modules in this application is only a logical division In actual applications, there may be other divisions in implementation. For example, multiple modules may be combined or integrated in another system, or some features may be ignored or not implemented.
  • the present application provides a method, device, equipment and storage medium for generating an adversarial network synthesis image based on conditions, which can be applied to a feature encoder network.
  • this application mainly provides the following technical solutions:
  • CGAM Conditional Generative Adversarial Nets
  • FIG. 1 Please refer to FIG. 1, the following describes a method for generating an adversarial network synthetic image based on conditions in an embodiment of the present application, and the method includes:
  • each red blood cell image includes a plurality of red blood cells, and the shape and position of the red blood cells on the same red blood cell image may be the same or different.
  • the combining multiple sub-masks corresponding to each red blood cell image to generate a mask to obtain multiple masks corresponding to multiple red blood cell images includes:
  • red blood cell shape sampler Calling the red blood cell shape sampler to iteratively select the red blood cell shape s i from the red blood cell data set, where 1 ⁇ i ⁇ n and i is a positive integer;
  • the erythrocytes of different shapes and positions are synthesized into a segmentation mask, that is, the mask, that is, the shape of the red blood cell and the position of the red blood cell in the red blood cell image are obtained, and a representation of the mask obtained after synthesis
  • a segmentation mask that is, the mask, that is, the shape of the red blood cell and the position of the red blood cell in the red blood cell image
  • s 1 , s 2 , ... s n are the shapes of red blood cells
  • l n is the position of red blood cells in a red blood cell image
  • (s n , l n ) is a sub-mask
  • n is a red blood cell image
  • the total number of red blood cells, n is a positive integer
  • background is the background pixel image of the red blood cell image.
  • red blood cells shape of red blood cells, the total number of red blood cells in a red blood cell image, and the position of red blood cells in a red blood cell image are normally distributed.
  • a normal distribution is expressed as n ⁇ Norm( ⁇ n , ⁇ n ), ⁇ n And ⁇ n are determined by the training set.
  • a probability density function to select si , and a set of probabilities can be calculated through the probability density function, and the set of probabilities can be used to enhance the appearance characteristics of red blood cell shape s i , such as enriching the appearance of red blood cells , Including rotation, zoom, horizontal/vertical flip, etc.
  • the red blood cell shape data set is used to provide red blood cell shape data when synthesizing red blood cell images.
  • the multiple red blood cell shape data are collected from the multiple masks to obtain a training data set, the segmentation boundary of each red blood cell in the training data set is calculated, and the red blood cell shape data is established according to the segmentation boundary of each red blood cell Set, including:
  • Edge detection is used to detect the cell membrane of red blood cells in each red blood cell image area to the eight-connected edge of a single pixel in each red blood cell image area;
  • Canny edge detection is performed on the filled binarized image to obtain the segmentation boundary (also called contour) of each red blood cell.
  • T is the segmentation threshold
  • f(i,j) refers to the input red blood cell image
  • g(i,j) refers to the output red blood cell image.
  • the segmentation boundaries of all red blood cells can be extracted and the size is judged. If the segmentation boundary is less than the preset threshold, it is considered that the red blood cells whose segmentation boundary is less than the preset threshold are not red blood cells.
  • filling the inside of the red blood cells in the middle can avoid double edges of the red blood cells inside; filling the inside of each red blood cell in the binarized image can avoid double edges of the red blood cells inside.
  • a distribution estimation algorithm may be used to separately collect the position of each red blood cell in the canvas in a probability density function in a two-dimensional discrete space.
  • the collecting the distribution data of each red blood cell in the red blood cell shape data set includes:
  • the position of each red blood cell and the position of each pixel, the prior probability of each pixel being selected as a red blood cell center is calculated;
  • the distribution data of each red blood cell in the red blood cell shape data set is calculated according to the position of each red blood cell and the position of each pixel point in the true red blood cell adhesion state when the value of each i is simulated.
  • the probability density function is represented by the possibility atlas P(i), and the formula for collecting the position l i of the i-th red blood cell from the possibility atlas P(i) is as follows:
  • the value of each pixel in P(i) refers to the prior probability of being selected as the red blood cell center in the i-th step.
  • P(i) is initially uniform when extracting the first n init cell. When i increases, P(i) will change its shape.
  • formula (2) according to the value of i in descending order, from The position l i where red blood cells of different shapes are extracted in P(i). Therefore, the entire process of extracting red blood cells from P(i) can simulate the real red blood cell adhesion state.
  • a Markov stochastic process can be used to simulate the natural evolution of P(i).
  • One way of expression is as follows:
  • the purpose of this step is to reduce the possibility of red blood cell boundaries that have been allocated to prevent red blood cells from overlapping.
  • red blood cells when locating red blood cells in the synthetic mask, whenever the red blood cells are always on the canvas of the synthetic mask, a color can be given to make the red blood cells in contact with different colors. If this condition cannot be met, then repeat the coordinate collection process. Since the color of each red blood cell is different, the generated synthetic mask can be used as an example segmentation mask with the possibility of extracting each red blood cell.
  • a generator G for generating red blood cell images and two multi-scale distinguishers (referred to as D1 and D2 for short) are provided in the feature encoder network E.
  • the collecting the red blood cell shape data into multiple red blood cell images includes:
  • the generator G converts the segmentation mask in the red blood cell shape data set into multiple red blood cell images, and inputs the converted multiple red blood cell images into the two multi-scale distinguishers D; the conversion obtains The multiple red blood cell images in the forehead are all images that simulate lifelike red blood cells;
  • the two multi-scale distinguishers D perform at least one distinction between the real red blood cell image and the synthesized red blood cell image within a preset time period, so as to train the neural network model;
  • the two multi-scale discriminators D output training results
  • the feature encoder network E combines the training result and the red blood cell shape data set x to obtain a combined result, and the combined result is used to control the style of synthesizing the red blood cell image;
  • the generator synthesizes the multiple masks into the red blood cell image
  • the combination result can be obtained by using a K-means clustering algorithm to generate multiple clusters, such as 10 clusters, from the training result and the red blood cell shape data set, and the style of the red blood cell image obtained by synthesis is determined by the encoder E The random collection characteristics of the multiple clusters are determined.
  • the complete network training goals are as follows:
  • x is the red blood cell shape data set
  • LGAN(G, Dk) is the anti-loss
  • one way of expressing the anti-loss is as follows:
  • LFM (G; Dk) is the feature matching loss.
  • the feature matching loss is used to stabilize the training results and produce better visual results on multiple scales.
  • One way to express feature matching loss is as follows:
  • LPR(G(x);y) is the perceptual reconstruction loss, which is used to further improve the performance of the quality composite image.
  • One way to express the perceptual reconstruction loss is as follows:
  • the above describes a method for generating an adversarial network synthetic image based on conditions in the present application, and the following describes a method for executing the foregoing method for generating an adversarial network synthetic image based on conditions.
  • a schematic structural diagram of a device 20 for synthesizing images can be applied to recognize red blood cell images.
  • the device 20 in the embodiment of the present application can implement the steps corresponding to the method of generating an adversarial network synthesis image based on the conditions executed in the embodiment corresponding to FIG. 1 above.
  • the functions implemented by the device 20 can be implemented by hardware, or implemented by hardware executing corresponding software.
  • the hardware or software includes one or more modules corresponding to the above functions, and the modules may be software and/or hardware.
  • the device 20 may include an input/output module 201 and a processing module 202.
  • the processing module 202 can be used to control the input/output or receiving and sending operations of the input/output module 201.
  • the input and output module 201 can be used to obtain multiple clinical red blood cell images, divide the red blood cells of different shapes and at different positions in each red blood cell image into multiple sub-masks, and divide each red blood cell image.
  • the multiple sub-masks corresponding to the image are synthesized to generate one mask to obtain multiple masks corresponding to multiple red blood cell images; wherein, each red blood cell image includes multiple red blood cells, and the red blood cells on the same red blood cell image
  • the shape and position can be the same or different;
  • the processing module 202 can be used to collect multiple red blood cell shape data from multiple masks through the input and output module to obtain a training data set, calculate the segmentation boundary of each red blood cell in the training data set, and according to the red blood cell
  • the segmentation boundary establishes a red blood cell shape data set;
  • the red blood cell shape data set is used to provide red blood cell shape data when synthesizing red blood cell images;
  • the input and output module collects the distribution data of each red blood cell in the red blood cell shape data set;
  • the red blood cell shape data is collected into multiple red blood cell images.
  • the processing module is specifically used for:
  • red blood cell shape sampler Calling the red blood cell shape sampler to iteratively select red blood cell shape s i from the red blood cell data set, 1 ⁇ i ⁇ n and i is a positive integer, s i refers to red blood cells of different shapes and in different positions;
  • s 1 , s 2 , ... s n are the shapes of red blood cells
  • l n is the position of red blood cells in a red blood cell image
  • (s n , l n ) is a sub-mask
  • n is a red blood cell image
  • the total number of red blood cells, n is a positive integer
  • background is the background pixel image of the red blood cell image.
  • the processing module is specifically used for:
  • Edge detection is used to detect the cell membrane of red blood cells in each red blood cell image area to the eight-connected edge of a single pixel in each red blood cell image area;
  • T is the segmentation threshold
  • f(i,j) refers to the input red blood cell image
  • g(i,j) refers to the output red blood cell image.
  • the processing module is specifically used for:
  • the position of each red blood cell and the position of each pixel, the prior probability of each pixel being selected as a red blood cell center is calculated;
  • the distribution data of each red blood cell in the red blood cell shape data set is calculated according to the position of each red blood cell and the position of each pixel point in the true red blood cell adhesion state when the value of each i is simulated.
  • the processing module is specifically used for:
  • the generator G converts the segmentation mask in the red blood cell shape data set into multiple red blood cell images, and inputs the converted multiple red blood cell images into the two multi-scale distinguishers D; the conversion obtains The multiple red blood cell images in the forehead are all images that simulate lifelike red blood cells;
  • the two multi-scale distinguishers D perform at least one distinction between the real red blood cell image and the synthesized red blood cell image within a preset time period, so as to train the neural network model;
  • the two multi-scale discriminators D output training results
  • the feature encoder network E combines the training result and the red blood cell shape data set x to obtain a combined result; wherein the combined result is used to control the style of synthesizing the red blood cell image; the combined result can be K
  • the mean clustering algorithm generates multiple clusters from the training result and the red blood cell shape data set, and the style of the synthesized red blood cell image is determined by the randomly collected features of the multiple clusters of the encoder E;
  • the generator synthesizes the plurality of masks into the red blood cell image.
  • the physical device corresponding to the input-output module 201 shown in FIG. 2 is the input-output unit shown in FIG. 3.
  • the input-output unit can realize part or all of the functions of the acquisition module 1, or realize the same or similar functions as the input-output module 201 Features.
  • the physical device corresponding to the processing module 202 shown in FIG. 2 is the processor shown in FIG. 3, and the processor can implement part or all of the functions of the processing module 202 or implement the same or similar functions as the processing module 202.
  • the device 20 in the embodiment of the present application is described above from the perspective of modular functional entities.
  • the following describes a computer device from the perspective of hardware, as shown in FIG. 3, which includes: a processor, a memory, a transceiver (or An input and output unit (not identified in FIG. 3) and a computer program stored in the memory and running on the processor.
  • the computer program may be a program corresponding to the method of generating an adversarial network composite image based on conditions in the embodiment corresponding to FIG. 1.
  • the processor executes the computer program to implement the condition-based generation of the adversarial network composite image executed by the apparatus 20 in the embodiment corresponding to FIG.
  • the computer program may be a program corresponding to the method of generating an adversarial network composite image based on conditions in the embodiment corresponding to FIG. 1.
  • the present application also provides a computer-readable storage medium.
  • the computer-readable storage medium may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium.
  • the computer-readable storage medium stores computer instructions, and when the computer instructions are executed on the computer, the computer executes the following steps:
  • each red blood cell image includes multiple red blood cells, and the shape and position of the red blood cells on the same red blood cell image may be the same or different;
  • red blood cell shape data Set is used to provide red blood cell shape data when synthesizing red blood cell images
  • the red blood cell shape data is collected into a plurality of red blood cell images.
  • the method of the above embodiments can be implemented by means of software plus the necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is better. ⁇
  • the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product.
  • the computer software product is stored in a storage medium (such as ROM/RAM), including Several instructions are used to make a terminal (which may be a mobile phone, a computer, a server, or a network device, etc.) execute the method described in each embodiment of the present application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Multimedia (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Computing Systems (AREA)
  • Probability & Statistics with Applications (AREA)
  • Public Health (AREA)
  • Epidemiology (AREA)
  • Primary Health Care (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Quality & Reliability (AREA)
  • Image Analysis (AREA)
  • Investigating Or Analysing Biological Materials (AREA)
  • Apparatus Associated With Microorganisms And Enzymes (AREA)
  • Image Processing (AREA)
PCT/CN2019/117988 2019-08-12 2019-11-13 基于条件生成对抗网络合成图像的方法及相关设备 Ceased WO2021027152A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
US17/264,312 US11636695B2 (en) 2019-08-12 2019-11-13 Method for synthesizing image based on conditional generative adversarial network and related device
JP2021506747A JP7329041B2 (ja) 2019-08-12 2019-11-13 条件付き敵対的生成ネットワークに基づいて画像を合成する方法および関連機器
SG11202100136SA SG11202100136SA (en) 2019-08-12 2019-11-13 Method for synthesizing image based on conditional generative adversarial network and related device

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910741020.7 2019-08-12
CN201910741020.7A CN110648309B (zh) 2019-08-12 2019-08-12 基于条件生成对抗网络合成红细胞图像的方法及相关设备

Publications (1)

Publication Number Publication Date
WO2021027152A1 true WO2021027152A1 (zh) 2021-02-18

Family

ID=69009447

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/117988 Ceased WO2021027152A1 (zh) 2019-08-12 2019-11-13 基于条件生成对抗网络合成图像的方法及相关设备

Country Status (5)

Country Link
US (1) US11636695B2 (https=)
JP (1) JP7329041B2 (https=)
CN (1) CN110648309B (https=)
SG (1) SG11202100136SA (https=)
WO (1) WO2021027152A1 (https=)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111833238B (zh) * 2020-06-01 2023-07-25 北京百度网讯科技有限公司 图像的翻译方法和装置、图像翻译模型的训练方法和装置
CN112102323B (zh) * 2020-09-17 2023-07-07 陕西师范大学 基于生成对抗网络和Caps-Unet网络的粘连细胞核分割方法
CN112435259B (zh) * 2021-01-27 2021-04-02 核工业四一六医院 一种基于单样本学习的细胞分布模型构建及细胞计数方法
CN115424604B (zh) * 2022-07-20 2024-03-15 南京硅基智能科技有限公司 一种基于对抗生成网络的语音合成模型的训练方法
US20250166173A1 (en) * 2023-11-21 2025-05-22 L&T Technology Services Limited Method and system of analysing blood smear image using deep learning model

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107492084A (zh) * 2017-07-06 2017-12-19 哈尔滨理工大学 基于随机性的典型成团细胞核图像合成方法
CN108388925A (zh) * 2018-03-06 2018-08-10 天津工业大学 基于新型条件对抗生成网络的抗模式崩溃鲁棒图像生成方法
CN109493330A (zh) * 2018-11-06 2019-03-19 电子科技大学 一种基于多任务学习的细胞核实例分割方法
CN109843176A (zh) * 2016-07-29 2019-06-04 诺瓦达克技术有限公司 用于利用机器学习来表征受验者的组织的方法和系统
US20190244681A1 (en) * 2018-02-08 2019-08-08 Ohio State Innovation Foundation Synthetic ihc-stained digital sides generated using artificial neural networks

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10176363B2 (en) 2014-06-16 2019-01-08 Siemens Healthcare Diagnostics Inc. Analyzing digital holographic microscopy data for hematology applications
CN107564018A (zh) * 2017-08-30 2018-01-09 北京航空航天大学 一种利用改进迭代算法提取目标图像的方法
EP3685355B1 (en) 2017-09-19 2024-07-10 Siemens Healthcare Diagnostics Inc. Virtual staining of cells in digital holographic microscopy images using general adversarial networks
CN109726644A (zh) * 2018-12-14 2019-05-07 重庆邮电大学 一种基于生成对抗网络的细胞核分割方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109843176A (zh) * 2016-07-29 2019-06-04 诺瓦达克技术有限公司 用于利用机器学习来表征受验者的组织的方法和系统
CN107492084A (zh) * 2017-07-06 2017-12-19 哈尔滨理工大学 基于随机性的典型成团细胞核图像合成方法
US20190244681A1 (en) * 2018-02-08 2019-08-08 Ohio State Innovation Foundation Synthetic ihc-stained digital sides generated using artificial neural networks
CN108388925A (zh) * 2018-03-06 2018-08-10 天津工业大学 基于新型条件对抗生成网络的抗模式崩溃鲁棒图像生成方法
CN109493330A (zh) * 2018-11-06 2019-03-19 电子科技大学 一种基于多任务学习的细胞核实例分割方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LI CHENGCHENG; WANG ZI; QI HAIRONG: "Fast-Converging Conditional Generative Adversarial Networks for Image Synthesis", 2018 25TH IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING (ICIP), IEEE, 7 October 2018 (2018-10-07), pages 2132 - 2136, XP033454734, DOI: 10.1109/ICIP.2018.8451161 *

Also Published As

Publication number Publication date
US20210312243A1 (en) 2021-10-07
SG11202100136SA (en) 2021-03-30
JP7329041B2 (ja) 2023-08-17
JP2022500728A (ja) 2022-01-04
CN110648309B (zh) 2024-05-28
CN110648309A (zh) 2020-01-03
US11636695B2 (en) 2023-04-25

Similar Documents

Publication Publication Date Title
US11861829B2 (en) Deep learning based medical image detection method and related device
WO2021027152A1 (zh) 基于条件生成对抗网络合成图像的方法及相关设备
US20230386048A1 (en) Ultrasound image segmentation method and apparatus, terminal device, and storage medium
CN107730515B (zh) 基于区域增长和眼动模型的全景图像显著性检测方法
CN110335668A (zh) 基于深度学习的甲状腺癌细胞病理图辅助分析方法及系统
CN108257135A (zh) 基于深度学习方法解读医学图像特征的辅助诊断系统
CN111882559B (zh) Ecg信号的获取方法及装置、存储介质、电子装置
CN114332553B (zh) 图像处理方法、装置、设备以及存储介质
WO2019184851A1 (zh) 图像处理方法和装置及神经网络模型的训练方法
CN111415728A (zh) 基于cnn和gan的ct图像数据自动分类方法及设备
JP2022500728A5 (https=)
CN113313680A (zh) 一种结直肠癌病理图像预后辅助预测方法及系统
CN119941731B (zh) 基于大模型的肺结节分析方法、系统、设备及介质
CN111598144A (zh) 图像识别模型的训练方法和装置
AU2019204365C1 (en) Method and System for Image Segmentation and Identification
CN120511037A (zh) 一种基于深度学习实现胃癌患者精准预后预测的系统
Chatterjee et al. A survey on techniques used in medical imaging processing
Barstugan et al. Automatic liver segmentation in abdomen CT images using SLIC and AdaBoost algorithms
CN116168051A (zh) 骨骼的识别方法、装置、电子设备及介质
CN118429369B (zh) 基于自然语言处理技术的医学影像分割方法及系统
Maraci et al. Fisher vector encoding for detecting objects of interest in ultrasound videos
CN117058405B (zh) 一种基于图像的情绪识别方法、系统、存储介质及终端
CN110706209A (zh) 一种网格网络的大脑磁共振图像肿瘤定位方法
Chen et al. Pulmonary nodule segmentation in computed tomography with an encoder-decoder architecture
CN116205913A (zh) 眼科超声图像的处理方法、装置、设备以及存储介质

Legal Events

Date Code Title Description
ENP Entry into the national phase

Ref document number: 2021506747

Country of ref document: JP

Kind code of ref document: A

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19941095

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19941095

Country of ref document: EP

Kind code of ref document: A1

WWG Wipo information: grant in national office

Ref document number: 11202100136S

Country of ref document: SG

WWP Wipo information: published in national office

Ref document number: 11202100136S

Country of ref document: SG