Unlock instant, AI-driven research and patent intelligence for your innovation.

Leukocyte classified counting method based on small sample semi-supervised learning

A technique of semi-supervised learning and white blood cell classification, applied in neural learning methods, counting of randomly distributed items, reasoning methods, etc.

Active Publication Date: 2021-05-14
SOUTHEAST UNIV
View PDF3 Cites 15 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The present invention is aimed at the problems existing in the prior art, and provides a leukocyte classification and counting method based on small-sample semi-supervised learning. The problem of classification model accuracy

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Leukocyte classified counting method based on small sample semi-supervised learning
  • Leukocyte classified counting method based on small sample semi-supervised learning
  • Leukocyte classified counting method based on small sample semi-supervised learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0043] Embodiment 1: see Figure 1-Figure 8 , a kind of leukocyte classification counting method based on small sample semi-supervised learning, described method comprises the following steps:

[0044] S1. Use a microscope to take a large number of microscopic images of blood cells from blood smears, and use image processing to locate individual white blood cells;

[0045] S2. For the five types of cells: mononuclear, neutrophil, lymphatic, eosinophilic, and basophilic, mark several images (about 50 to 100 for each type), and the remaining unlabeled images (generally greater than 1000) are used as training samples, and then In addition to the samples in the training set, a number of images (about 100 images in each category) were randomly selected for labeling to test the effect of the model, and there was no intersection between the test set and the training set;

[0046] S3. According to the training samples in step S2, determine the input and output of the semi-supervised ...

specific Embodiment

[0069] Specific examples: refer to figure 1 — Figure 8 , a leukocyte classification and counting method based on small-sample semi-supervised learning, such as figure 1 and Figure 8 , including the following steps,

[0070] S1. Use a microscope to take microscopic images of cells from blood smears, and use image processing to locate individual white blood cells;

[0071] In step S1, a single white blood cell obtained through image processing in the embodiment occupies more than 60% of the entire image, and the cell is relatively complete, and there are background cells such as platelets and red blood cells around it.

[0072] The image processing operation in step S1 is specifically,

[0073]S11, convert the collected color cell image into a grayscale image, utilize the grayscale distribution characteristics of the image (the histogram presents two peaks), and use the Otsu threshold to perform adaptive segmentation, which is not easily affected by image brightness and co...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention relates to a leukocyte classified counting method based on small sample semi-supervised learning. The method comprises the following steps: firstly, clearly positioning a blood cell image shot by a microscope to a single leukocyte in an image processing mode to obtain a to-be-classified cell image; carrying out manual labeling on the selected part of the cell image to respectively obtain a labeled sample and an unlabeled sample, and distributing a training sample and a test sample; determining the input and output of the classification network and the structure of the middle part, and constructing a semi-supervised classification network based on a dual-network structure; training a semi-supervised classification network by using a small number of labeled samples and a large number of unlabeled samples, and storing a model with an optimal training effect; and classifying the positioned single leukocyte images, and outputting classification information of each image according to a semi-supervised classification network, so as to count the number of each class of leukocytes. According to the method, leukocyte image classification statistics can be realized under the condition of less labeling, and the detection efficiency and precision are high.

Description

technical field [0001] The invention relates to a counting method, in particular to a white blood cell classification and counting method based on small-sample semi-supervised learning, and belongs to the technical field of cell classification. Background technique [0002] Peripheral blood leukocyte (white blood cell, WBC) classification is a routine work in clinical examination, which is of great significance to many diseases. At present, laboratories usually use blood cell analyzers for WBC classification and counting. These analyzers generally use physical, cytochemical and other classification techniques, but this method can only be used for counting, and cannot use cell images under a microscope, so that it cannot assist doctors. further analysis. However, the traditional manual microscopic examination requires a large amount of cells, which is time-consuming and labor-intensive. Second, the artificial statistical system has a large deviation. Therefore, improving t...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06K9/62G06N3/04G06N3/08G06N5/04G06M11/02
CPCG06N3/084G06N5/04G06M11/02G06N3/045G06F18/2453
Inventor 胡轶宁陈奕君谢理哲王征
Owner SOUTHEAST UNIV