Research on White Blood Cell Classification Method Based on Convolutional Neural Network
The traditional method of white blood cell classification has the problem of time-consuming and low accuracy in feature extraction.It is proposed to use deep learning methods to achieve automatic feature extraction and use open source BCCD data sets for testing.There are 12 436 white blood cell,images in the data set.The classic convolutional neural networks AlexNet,VGG11,GoogLeNet,ResNet18,ResNet34,DenseNet121,and EfficientNetB0 models are used to train the images,and the model training results are quantitatively evaluated by the confusion matrix.The results show that the EfficientNetB0 network model is generally superior to other networks,with an accuracy rate of 96.02%and a time of only 8.3 s on the test set.In order to improve the interpretability of the model,the heat map is visualized and the experimental results are also confirmed.
while blood celldeep learningimage classificationvisualization
万方数据
维普
