Leukocyte five-classification method based on an improved attention convolutional neural network

Abstract

The invention belongs to the field of medical microscopic image classification, and provides a leukocyte five-classification method based on an improved attention convolutional neural network, which uses deep learning to identify blood cell images. ResNeXt-50 is used as a backbone network, a residual module uses packet convolution to reduce model parameter quantity, an independent attention module structure is added in parallel at the end of each stage of the network, and for leukocyte feature maps output by the convolutional neural network in different stages, an attention head part of an attention module is used to extract a leukocyte key region. A head part is output through an attention module, a prediction category and a confidence score are output, the output of a final network model is obtained through weighted average of the prediction category and confidence, and leukocyte five-classification is realized based on improvement under an original ResNeXt-50 network framework. The parallel attention module is used for outputting class prediction and confidence scores in different stages of the network, and the accuracy of leukocyte classification is improved.

Description

technical field [0001] The invention belongs to the field of medical microscopic image classification, and relates to a white blood cell classification method embedded in parallel attention module convolutional neural network. Background technique [0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art. [0003] White blood cells are part of the immune system and they are responsible for destroying and removing old or abnormal cells and cell debris, as well as attacking pathogens and foreign bodies. The white blood cells normally found in the blood are mature neutrophils, lymphocytes, and monocytes, with eosinophils and basophils in smaller numbers. An increase or decrease in the number of white blood cells may represent a sign of the onset of certain diseases, and the shape and proportion of various white blood cells can reflect a person's health status. Therefore, the accur...

AI Technical Summary

Problems solved by technology

[0009] In view of the noise influence caused by stacking residual attention modules in the current method and the failure to fully utilize different levels of features, the present invention provides a five-classification method for white blood cells based on an improved attention convolutional neural network. ResNext-50 is a backbone network. The group convolution in the network can greatly reduce the amount of model parameters. An independent attention mechanis

Images