CN110826483A - Cell identification method for leucorrhea microscopic image - Google Patents
Cell identification method for leucorrhea microscopic image Download PDFInfo
- Publication number
- CN110826483A CN110826483A CN201911071502.2A CN201911071502A CN110826483A CN 110826483 A CN110826483 A CN 110826483A CN 201911071502 A CN201911071502 A CN 201911071502A CN 110826483 A CN110826483 A CN 110826483A
- Authority
- CN
- China
- Prior art keywords
- cells
- microscopic image
- positive
- negative
- leucorrhea
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 25
- 210000004027 cell Anatomy 0.000 claims description 55
- 241000222120 Candida <Saccharomycetales> Species 0.000 claims description 5
- 238000001514 detection method Methods 0.000 claims description 5
- 241000224526 Trichomonas Species 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 210000000265 leukocyte Anatomy 0.000 claims description 4
- 241001502500 Trichomonadida Species 0.000 claims description 3
- 238000013145 classification model Methods 0.000 claims description 3
- 238000013527 convolutional neural network Methods 0.000 claims description 3
- 238000013526 transfer learning Methods 0.000 claims description 3
- 230000008014 freezing Effects 0.000 claims description 2
- 238000007710 freezing Methods 0.000 claims description 2
- 230000005859 cell recognition Effects 0.000 claims 4
- 238000013135 deep learning Methods 0.000 abstract description 2
- 230000007547 defect Effects 0.000 abstract description 2
- 230000000007 visual effect Effects 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 2
- 241000222122 Candida albicans Species 0.000 description 1
- 241000233866 Fungi Species 0.000 description 1
- 206010046914 Vaginal infection Diseases 0.000 description 1
- 201000008100 Vaginitis Diseases 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 210000000601 blood cell Anatomy 0.000 description 1
- 229940095731 candida albicans Drugs 0.000 description 1
- 210000003756 cervix mucus Anatomy 0.000 description 1
- 230000002357 endometrial effect Effects 0.000 description 1
- 229940011871 estrogen Drugs 0.000 description 1
- 239000000262 estrogen Substances 0.000 description 1
- 210000000416 exudates and transudate Anatomy 0.000 description 1
- 230000035784 germination Effects 0.000 description 1
- 210000004907 gland Anatomy 0.000 description 1
- 210000004877 mucosa Anatomy 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000028327 secretion Effects 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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
- G06V20/693—Acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR 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; CALCULATING OR 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
- G06V20/698—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Investigating Or Analysing Biological Materials (AREA)
Abstract
The invention discloses a cell identification method of a leucorrhea microscopic image, which comprises the following steps of S1, shooting 20 microscopic images of a placed leucorrhea sample through a medical microscope; s2, detecting the positions of various cells on each microscopic image and identifying the types of the cells; s2.1, finding out the positions of all suspected cells through an improved YOLOv3-tiny algorithm model; s2.2, classifying the cell areas on each microscopic image by adopting a Resnet50 algorithm model, counting the number of various cells in the cell identification result of each microscopic image, and summarizing the number of various cells on 20 microscopic images; and S3, outputting the positive and negative judgment results of various cells of the leucorrhea sample according to the requirements of the specified negative and positive judgment standards of different cell types on the number of the cells. The invention adopts a deep learning method to identify various cells in the leucorrhea sample and provides the judgment result of the negative and positive of various cells in the leucorrhea sample, thereby reducing the defects of strong subjectivity and severe dependence on experience due to the visual identification of doctors.
Description
Technical Field
The invention relates to a cell identification method, in particular to a cell identification method for a white band microscopic image.
Background
Leucorrhea is a female vaginal secretion and is formed by mixing vaginal mucosa exudate, cervical canal and endometrial gland secretion, and the formation of leucorrhea is related to the action of estrogen. Leukocytes are colorless, spherical, nucleated blood cells; the candida is a fungus, and usually causes vaginitis, the candida albicans in the candida is in an oval shape, and pseudohypha is formed by germination and elongation of spores and cells; trichomonas is a tiny flagellated protozoan organism.
At present, doctors observe leucorrhea samples under an optical microscope for identifying the negative and positive leucorrhea, and the leucorrhea samples are distinguished through experience, so that the subjectivity is strong, and the experience requirement is high. Meanwhile, the optical microscope system is expensive, and because images are required to be left for the doctor to review, dozens of images are required to be shot for each sample, and the requirement on a hard disk of a computer is large.
Disclosure of Invention
The invention aims to provide a cell identification method for a leucorrhea microscopic image, which avoids the influence of experience difference of a doctor on a detection result through manual microscopic examination and improves the detection efficiency.
In order to achieve the purpose, the invention adopts the following technical scheme:
the invention relates to a cell identification method of a white band microscopic image, which comprises the following steps:
s1, taking 20 microscopic images of the placed leucorrhea sample through a medical microscope;
s2, detecting the positions of various cells on each microscopic image and identifying the types of the cells;
s2.1, finding out the positions of all suspected cells through an improved YOLOv3-tiny algorithm model;
s2.2, classifying the cell areas on each microscopic image by adopting a Resnet50 algorithm model, counting the number of various cells in the cell identification result of each microscopic image, and summarizing the number of various cells on 20 microscopic images;
s3, outputting the positive and negative judgment results of various cells of the leucorrhea sample according to the requirements of the specified negative and positive judgment standards of different cell types on the number of the cells; the positive and negative judgment standards of trichomonas and mould are as follows: the total number of cells in 20 microscopic images is more than 1, positive and less than 1, negative; the positive and negative discrimination criteria of the white blood cells are as follows: the total number of cells in 20 images was more than 100 positive and less than 100 negative.
The improved YOLOv3-tiny algorithm model is based on a YOLOv3-tiny algorithm and combines the characteristics of small targets such as cells in a microscopic image, and two groups of multi-scale features of a shallower layer are extracted by adjusting the structure of the model and the output of a feature map so as to adapt to the detection task of the cells.
The Resnet50 algorithm model is a traditional convolutional neural network classification model and consists of a convolutional layer, 4 residual modules and a full connection layer; and (3) introducing transfer learning in the process of training the network, using a Resnet50 model pre-trained on an ImageNet data set provided by a Tensorflow official network, freezing parameters of the convolutional layer and the first three residual modules in the training process, inheriting the bottom layer feature extraction capability of the source model, and setting the parameters of the fourth residual module and the full connection layer to be updatable.
The microscopic image is in a JPG, BMP or PNG image format.
The cells comprise leucocytes, moulds, candida and trichomonads.
The invention adopts a deep learning method to identify various cells in the leucorrhea sample and provides the judgment result of the negativity and the positivity of various cells in the leucorrhea sample, thereby realizing automatic identification and intelligent operation, greatly improving the working efficiency of doctors and reducing the defects of strong subjectivity and severe dependence on experience of the doctors through naked eye identification.
Drawings
FIG. 1 is a block flow diagram of the method of the present invention.
FIG. 2 is a block diagram of the improved YOLOv3-tiny algorithm model flow according to the present invention.
Fig. 3.1 and 3.2 are structural block diagrams of the Resnet50 algorithm model according to the present invention.
Detailed Description
The following describes embodiments of the present invention in detail with reference to the drawings, which are implemented on the premise of the technical solution of the present invention, and detailed embodiments and specific operation procedures are provided, but the scope of the present invention is not limited to the following embodiments.
As shown in fig. 1, the method for identifying cells in a white band microscopic image according to the present invention comprises the following steps:
s1, taking 20 microscopic images of the placed leucorrhea sample through a medical microscope; the microscopic image may be in a JPG, BMP, or PNG image format;
s2, detecting the positions of various cells on each microscopic image and identifying the types of the cells; the cells comprise leucocytes, mould, candida and trichomonad;
s2.1, finding out the positions of all suspected cells through an improved YOLOv3-tiny algorithm model;
s2.2, classifying the cell areas on each microscopic image by adopting a Resnet50 algorithm model, counting the number of various cells in the cell identification result of each microscopic image, and summarizing the number of various cells on 20 microscopic images;
s3, outputting the positive and negative judgment results of various cells of the leucorrhea sample according to the requirements of the specified negative and positive judgment standards of different cell types on the number of the cells; the positive and negative judgment standards of trichomonas and mould are as follows: the total number of cells in 20 microscopic images is more than 1, positive and less than 1, negative; the positive and negative discrimination criteria of the white blood cells are as follows: the total number of cells in 20 images was more than 100 positive and less than 100 negative.
The improved YOLOv3-tiny algorithm model is based on a YOLOv3-tiny algorithm and combines the characteristics of small targets such as cells in a microscopic image, and two groups of multi-scale features of a shallower layer are extracted by adjusting the structure of the model and the output of a feature map so as to adapt to the detection task of the cells.
As shown in fig. 2, an improved YOLOv3 model structure diagram is shown, which is an example of an input picture size of 416 × 416. The improved YOLOv3-tiny model feature extraction network adopts a 7-layer convolution layer and pooling layer network to extract features, and carries out prediction on two dimensions, and 2 feature maps are provided: 52 × 52, 104 × 104, respectively, down-sample the image by 8 times and 4 times. The output at each scale is a 3-dimensional tensor, containing bounding boxes, confidence and prediction classes.
As shown in fig. 3.1 and 3.2, the Resnet50 algorithm model is a traditional convolutional neural network classification model, and is composed of a convolutional layer, 4 residual modules and a full link layer; considering that the image used by the invention is a medical microscopic image, the number of samples is small, and large-scale supplement is difficult, therefore, the method introduces transfer learning in the process of training the network, uses a Resnet50 model pre-trained on an ImageNet data set provided by a Tensorflow official network, freezes the parameters of the convolutional layer and the first three residual modules in the training process, inherits the characteristic extraction capability of the bottom layer of the source model, and sets the parameters of the fourth residual module and the fully-connected layer to be updatable.
Claims (5)
1. A cell identification method for a leucorrhea microscopic image is characterized by comprising the following steps: the method comprises the following steps:
s1, taking 20 microscopic images of the placed leucorrhea sample through a medical microscope;
s2, detecting the positions of various cells on each microscopic image and identifying the types of the cells;
s2.1, finding out the positions of all suspected cells through an improved YOLOv3-tiny algorithm model;
s2.2, classifying the cell areas on each microscopic image by adopting a Resnet50 algorithm model, counting the number of various cells in the cell identification result of each microscopic image, and summarizing the number of various cells on 20 microscopic images;
s3, outputting the positive and negative judgment results of various cells of the leucorrhea sample according to the requirements of the specified negative and positive judgment standards of different cell types on the number of the cells; the positive and negative judgment standards of trichomonas and mould are as follows: the total number of cells in 20 microscopic images is more than 1, positive and less than 1, negative; the positive and negative discrimination criteria of the white blood cells are as follows: the total number of cells in 20 images was more than 100 positive and less than 100 negative.
2. The method for cell recognition in a white band microscopic image according to claim 1, characterized in that: the improved YOLOv3-tiny algorithm model is based on a YOLOv3-tiny algorithm and combines the characteristics of small targets such as cells in a microscopic image, and two groups of multi-scale features of a shallower layer are extracted by adjusting the structure of the model and the output of a feature map so as to adapt to the detection task of the cells.
3. The method for cell recognition in a white band microscopic image according to claim 1, characterized in that: the Resnet50 algorithm model is a traditional convolutional neural network classification model and consists of a convolutional layer, 4 residual modules and a full connection layer; and (3) introducing transfer learning in the process of training the network, using a Resnet50 model pre-trained on an ImageNet data set provided by a Tensorflow official network, freezing parameters of the convolutional layer and the first three residual modules in the training process, inheriting the bottom layer feature extraction capability of the source model, and setting the parameters of the fourth residual module and the full connection layer to be updatable.
4. The method for cell recognition in a white band microscopic image according to claim 1, characterized in that: the microscopic image is in a JPG, BMP or PNG image format.
5. The method for cell recognition in a white band microscopic image according to claim 1, characterized in that: the cells comprise leucocytes, moulds, candida and trichomonads.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911071502.2A CN110826483A (en) | 2019-11-05 | 2019-11-05 | Cell identification method for leucorrhea microscopic image |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911071502.2A CN110826483A (en) | 2019-11-05 | 2019-11-05 | Cell identification method for leucorrhea microscopic image |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110826483A true CN110826483A (en) | 2020-02-21 |
Family
ID=69552454
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911071502.2A Pending CN110826483A (en) | 2019-11-05 | 2019-11-05 | Cell identification method for leucorrhea microscopic image |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110826483A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112528947A (en) * | 2020-12-24 | 2021-03-19 | 山东仕达思生物产业有限公司 | False hypha detection method and device by increasing direction dimension and storage medium |
CN113158979A (en) * | 2021-05-14 | 2021-07-23 | 山东仕达思医疗科技有限公司 | Method for detecting and identifying leucocytes under large visual field of gynecological microscopic image |
CN113205533A (en) * | 2021-05-10 | 2021-08-03 | 江苏硕世生物科技股份有限公司 | Method and system for segmenting gram-stained leucorrhea smear color microscopic image |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108334835A (en) * | 2018-01-29 | 2018-07-27 | 华东师范大学 | Vaginal fluid micro-image visible component detection method based on convolutional neural networks |
CN108364006A (en) * | 2018-01-17 | 2018-08-03 | 超凡影像科技股份有限公司 | Medical Images Classification device and its construction method based on multi-mode deep learning |
CN109034208A (en) * | 2018-07-03 | 2018-12-18 | 怀光智能科技(武汉)有限公司 | A kind of cervical cell pathological section classification method of high-low resolution combination |
CN109598224A (en) * | 2018-11-27 | 2019-04-09 | 微医云(杭州)控股有限公司 | Recommend white blood cell detection method in the Sections of Bone Marrow of convolutional neural networks based on region |
US10332245B1 (en) * | 2018-12-11 | 2019-06-25 | Capital One Services, Llc | Systems and methods for quality assurance of image recognition model |
CN110222769A (en) * | 2019-06-06 | 2019-09-10 | 大连理工大学 | A kind of Further aim detection method based on YOLOV3-tiny |
-
2019
- 2019-11-05 CN CN201911071502.2A patent/CN110826483A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108364006A (en) * | 2018-01-17 | 2018-08-03 | 超凡影像科技股份有限公司 | Medical Images Classification device and its construction method based on multi-mode deep learning |
CN108334835A (en) * | 2018-01-29 | 2018-07-27 | 华东师范大学 | Vaginal fluid micro-image visible component detection method based on convolutional neural networks |
CN109034208A (en) * | 2018-07-03 | 2018-12-18 | 怀光智能科技(武汉)有限公司 | A kind of cervical cell pathological section classification method of high-low resolution combination |
CN109598224A (en) * | 2018-11-27 | 2019-04-09 | 微医云(杭州)控股有限公司 | Recommend white blood cell detection method in the Sections of Bone Marrow of convolutional neural networks based on region |
US10332245B1 (en) * | 2018-12-11 | 2019-06-25 | Capital One Services, Llc | Systems and methods for quality assurance of image recognition model |
CN110222769A (en) * | 2019-06-06 | 2019-09-10 | 大连理工大学 | A kind of Further aim detection method based on YOLOV3-tiny |
Non-Patent Citations (6)
Title |
---|
QIWEI WANG,ET AL.: "Deep learning approach to peripheral leukocyte recognition", 《PLOS ONE》 * |
QIWEI WANG,ET AL.: "Deep learning approach to peripheral leukocyte recognition", 《PLOS ONE》, vol. 14, no. 6, 25 June 2019 (2019-06-25), pages 4 * |
WANGPENG HE,ET AL.: "TF-YOLO:An Improved Incremental Network for Real-Time Object Detection", 《APPLIED SCIENCES》, vol. 9, no. 16, 7 August 2019 (2019-08-07), pages 4, XP055977210, DOI: 10.3390/app9163225 * |
廖子君等: "中华医学百科全书 实验诊断学", vol. 1, 中国协和医科大学出版社, pages: 1173 - 1174 * |
王恒等: "基于ResNet50网络的乳腺癌病理图像分类研究", 《中国计量大学学报》 * |
王恒等: "基于ResNet50网络的乳腺癌病理图像分类研究", 《中国计量大学学报》, vol. 30, no. 1, 31 March 2019 (2019-03-31), pages 73 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112528947A (en) * | 2020-12-24 | 2021-03-19 | 山东仕达思生物产业有限公司 | False hypha detection method and device by increasing direction dimension and storage medium |
CN112528947B (en) * | 2020-12-24 | 2023-05-23 | 山东仕达思生物产业有限公司 | Method, equipment and storage medium for detecting false hyphae by increasing direction dimension |
CN113205533A (en) * | 2021-05-10 | 2021-08-03 | 江苏硕世生物科技股份有限公司 | Method and system for segmenting gram-stained leucorrhea smear color microscopic image |
CN113158979A (en) * | 2021-05-14 | 2021-07-23 | 山东仕达思医疗科技有限公司 | Method for detecting and identifying leucocytes under large visual field of gynecological microscopic image |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11373305B2 (en) | Image processing method and device, computer apparatus, and storage medium | |
CN110826483A (en) | Cell identification method for leucorrhea microscopic image | |
CN110473167B (en) | Deep learning-based urinary sediment image recognition system and method | |
CN108961208A (en) | A kind of aggregation leucocyte segmentation number system and method | |
CN109190441B (en) | Intelligent classification method, diagnostic instrument and storage medium for female genital tract cell pathology | |
CN112085741B (en) | Gastric cancer pathological section segmentation algorithm based on deep learning | |
CN107832801B (en) | Model construction method for cell image classification | |
CN105654141A (en) | Isomap and SVM algorithm-based overlooked herded pig individual recognition method | |
CN104850860A (en) | Cell image recognition method and cell image recognition device | |
CN111079620A (en) | Leukocyte image detection and identification model construction method based on transfer learning and application | |
CN112001315B (en) | Bone marrow cell classification and identification method based on migration learning and image texture characteristics | |
CN115965602A (en) | Abnormal cell detection method based on improved YOLOv7 and Swin-Unet | |
CN115471701A (en) | Lung adenocarcinoma histology subtype classification method based on deep learning and transfer learning | |
CN112712122A (en) | Corneal ulcer classification detection method and system based on neural network model | |
CN115311316A (en) | Small watermelon identification and positioning method in three-dimensional cultivation mode based on deep learning | |
CN112528947A (en) | False hypha detection method and device by increasing direction dimension and storage medium | |
CN116385374A (en) | Cell counting method based on convolutional neural network | |
CN117576103B (en) | Urinary sediment microscopic examination analysis system integrating electric control microscope and deep learning algorithm | |
CN103268494B (en) | Parasite egg recognition methods based on rarefaction representation | |
CN113408480B (en) | Artificial intelligent auxiliary diagnosis system for blood diseases based on bone marrow cell morphology | |
CN114387596A (en) | Automatic interpretation system for cytopathology smear | |
CN110929678B (en) | Method for detecting vulvovaginal candida spores | |
CN111339899B (en) | Catheter feature acquisition method, device, equipment, medium and intelligent microscope | |
CN113065430A (en) | Leukocyte classification method based on fusion of deep learning features and artificial extraction features | |
CN114627123B (en) | Leucocyte detection method integrating double-current weighting network and spatial attention mechanism |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200221 |
|
RJ01 | Rejection of invention patent application after publication |