Research On The Classification Method Of White Blood Cells Based On Deep Learning

With the rapid development of computer technology,the use of artificial intelligence and image processing technology to assist medical diagnosis has attracted increasing attention.In the medical clinical examination,the examination of white blood cells has important value for the diagnosis of many diseases.At present,the detection methods used in hospitals are mainly blood cell analyzers and artificial microscopy,that is,preliminary screening with a blood cell analyzer and judging whether there are quantitative abnormalities,if abnormalities are found,artificial microscopy is performed.Due to the low efficiency of artificial microscopy and the slow classification speed,the automatic classification and recognition technology of peripheral blood leukocyte images has high practical value.Based on deep learning method,the thesis designs a scheme for automatic classification and recognition of peripheral blood leukocyte images.Firstly,observe the smear of peripheral blood cells with a high-power microscope,and use the camera to take a picture of blood cells containing a large number of white blood cells.Median filtering and bilateral filtering are used to preprocess the image,and the RGB and-HLS color spaces and common threshold segmentation methods are analyzed.Then,the L component image is segmented using the improved maximum between-class variance method,and the final binarization.The post-images are processed using morphological methods to obtain complete single leukocyte images.Secondly,sort and analyze the obtained leukocyte image set to eliminate the leukocyte images that have been stained incorrectly or failed.In order to solve the problem of im balance between the number of white blood cell images,the translation method is used to oversample the white blood cell with a small number of original cells to solve this problem.Then divide the white blood cell images into a training set and a test set according to a certain ratio,and use rotation,contrast enhancement and other data enhancement methods to amplify the two types of data set of this subject.Finally,the classification model of white blood cells is built according to the existing deep neural network architecture.The model consists of six convolutional layers,three pooling layers and three fully connected layers.The prepared training data set and verification data set are used for training verification,and the performance of the network model is evaluated through the test set to obtain white blood cell images.Then,the experimental results are visualized.The test results showed that the average recognition rate of white blood cell images is 92.87%,which meets the target.Using the method of convolutional neural network to classify and identify white blood cells not only avoids the complexity of accurate segmentation of white blood cell nuclei and cytoplasm,but also the individualized differences in artificially selected features.People are satisfied with the accuracy rate to achieve true end-to-end classification and recognition.