Multi-class leukocyte automatic identification method based on deep convolutional neural network

Abstract

The invention discloses a multi-class leukocyte automatic identification method based on a deep convolutional neural network, and belongs to automatic identification of a medical microscope image by using a deep learning machine vision scheme. The method comprises the following steps: firstly, carrying out arrangement and data enhancement operation on a data set of multiple types of leukocytes; then, using an inception module to perform cascading to form a multi-scale feature fusion convolutional neural network model; during cascade combination, establishing the mutual dependence relationshipbetween the characteristic channels through continuous squeeze operation and excitation operation, so that the network performance is improved, and the accuracy of leukocyte classification is improved.

Description

technical field [0001] The invention relates to the field of automatic recognition and classification of medical microscope images by using a computer vision scheme, in particular to a method for automatic recognition of multiple types of white blood cells based on a deep convolutional neural network. Background technique [0002] In recent years, artificial intelligence and deep learning have become technologies that all walks of life urgently need to learn and integrate. Therefore, computer vision has also undergone creative development, and has been applied to many medical image processing methods, and the efficiency has been greatly improved. improvement. The identification and counting of white blood cells provides valuable information for the diagnosis of human health and diseases. According to the classification results and white blood cell counts, infections, leukemia and certain types of cancers can be diagnosed. Highly accurate classification of biomedical images ...

AI Technical Summary

Problems solved by technology

[0005] Disadvantages: Several groups of artificially extracted features are used. Although these features have a certain degree of discrimination and representativeness, it is not guaranteed to obtain all or most of the distinguishable effective features, so this method has certain limitations.

[0007] Disadvantages: only morphological parameter analysis is used to distinguish several types of white blood cells, and the characteristics are relatively single, which has a certain impact on the accuracy of classification; the method is divided into two stages to divide ideas from coarse to fine, but it may appear that the first stage The error is brought into the second stage, which greatly increases the error rate of classification

Images