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

Urine red blood cell classification method based on supervised comparative learning

A classification method and red blood cell technology, applied in the field of medical image processing, can solve the problems of difficulty in meeting the physical examination needs of kidney disease patients and long examination cycle

Pending Publication Date: 2022-08-05
HARBIN ENG UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Therefore, on the one hand, the traditional manual inspection method has a long inspection cycle, and on the other hand, it is difficult to meet the physical examination needs of kidney disease patients in areas where professional doctors are lacking.

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
  • Urine red blood cell classification method based on supervised comparative learning
  • Urine red blood cell classification method based on supervised comparative learning
  • Urine red blood cell classification method based on supervised comparative learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0062] A method for classifying urinary red blood cells based on supervised contrastive learning, comprising the following steps:

[0063] Step 1: Collect and expand the data set; collect urine erythrocyte images, perform category labeling, and divide the collected images into a training set and a test set according to a certain proportion; the collected urine erythrocyte images are divided into normal erythrocytes, Ring-shaped red blood cells, shadow-shaped red blood cells, crescent-shaped red blood cells, bagel-shaped red blood cells, shriveled red blood cells, acanthous red blood cells, lateral red blood cells, and erect red blood cells. Preprocessing for data enhancement is performed on the categories with less data in the training set Mainly including rotation and flipping to get a larger number of training images;

[0064] Step 2: Build a feature extraction network model; on the basis of the ResNet-50 convolutional neural network, in view of the low resolution of the uri...

Embodiment 2

[0078] The concrete steps of the method for classifying urine erythrocytes based on supervised contrastive learning of the present invention are:

[0079] Step 1: Use a microscope to collect the initial image of the urine sediment, collect the urine red blood cell data set, label the data category information according to the cell morphological characteristics, and divide the training set and the test set. The collected urine red blood cells are divided into nine categories: normal red blood cells, annular red blood cells, shadow red blood cells, crescent red blood cells, doughnut red blood cells, crumpled red blood cells, acanthocytes, lateral red blood cells, and erect red blood cells. like figure 1 shown. Due to the small number of crescent-shaped red blood cells, crescent-shaped red blood cells, and acanthocytes, data enhancement methods were used for these three types of red blood cells to expand the data set. Among them, horizontal flipping of crescent-shaped red bloo...

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 provides a urine red blood cell classification method based on supervised comparative learning. The method comprises the following steps: 1, collecting and expanding a data set; step 2, constructing an SRRN feature extraction network model based on a ResNet-50 convolutional neural network; 3, constructing a loss function; 4, training the SRRN feature extraction network; step 5, training a classifier; 6, testing the model; according to the invention, after the SRRN network model is used for training, the accuracy on a test set is improved by 0.36%; the accuracy of the SRRN network model trained by adding a comparison loss function based on Euclidean distance on a test set is improved by 0.49%; the accuracy of the model after the classifier is finely adjusted by a weight penalty mechanism on a test set is improved by 0.71%; the TTA strategy is used in the model test stage, the accuracy of the model on a test set is improved by 0.13%, and finally the accuracy of the method reaches 92.39%.

Description

technical field [0001] The invention belongs to the field of medical image processing, in particular to a method for classifying urine red blood cells based on supervised contrast learning. Background technique [0002] On December 1, 2020, Kidney International Supplements (IF12.818), the official journal of the International Society of Nephrology, published the latest research results of the China Kidney Disease Data Network (CK-NET). The research results show that the composition of major chronic non-communicable diseases in my country is changing, especially chronic kidney disease showing a rapid upward trend. There is a huge imbalance between supply and demand in the prevention, control and diagnosis and treatment of chronic kidney disease. The pathogenesis of kidney-related diseases is complex, the doctor's consultation is difficult, and the treatment methods are lacking, which is a worldwide problem. The country and society spend a lot of medical expenses every year, ...

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): G06V20/69G06V10/82G06V10/764G06V10/774G06V10/40G06N3/04G06N3/08
CPCG06V20/698G06V20/69G06V10/40G06V10/82G06V10/764G06V10/774G06N3/08G06N3/045
Inventor 汲清波刘清泉姜月王玲婕章强陈奎丞
Owner HARBIN ENG UNIV