Research On White Blood Cell Detection Method Of Mask R-CNN Combined With Attention Mechanism

Blood diseases mainly include anemia,leukocyte diseases and hemorrhagic diseases.In recent years,the incidence rate of blood diseases is increasing in China,among the mortality of leukocyte diseases is the highest.In clinical,the number,proportion and morphological changes of different types of leukocytes are observed to diagnose the disease.Therefore,automatic detection and segmentation of leukocytes under microscope have become a hot topic in medical field,which is of great significance for the diagnosis of blood diseases.The accurate detection and segmentation of leukocytes is a challenging task in medical image processing.The white blood cell image obtained under microscope is easily affected by impurities.There are many kinds of white cells,different shapes and small differences among them,and also exiting overlapping and adhesion phenomena,which leads to the inaccuracy of cell edge segmentation.The above problems are always the difficulties of detection and segmentation of white cell image.To solve the above problems,this thesis proposes a leukocyte detection method based on Mask R-CNN and attention mechanism multi-scale feature fusion.Firstly,this thesis uses generative adversarial networks to generate leukocyte images and annotated leukocyte images in pairs,which can effectively expand the dataset and reduce the dependence of leukocyte dataset on manual annotation.Secondly,based on the Mask R-CNN structure,the attention mechanism module is integrated into the FPN(Feature Pyramid Networks)module,and the CSFPN(Channel Spatial Feature Pyramid Networks)structure is proposed.This structure can not only learn the weight of important channel features,but also learn the representation of important feature regions in the feature maps.At the same time,Skip-FPN module is added to the network structure,which fuses more low-level detailed information of leukocytes through short connection,so as to detect and segment leukocytes more accurately.Experimental results show that the average accuracy for white blood cell detection reaches 98.25%,and the average accuracy of segmentation reaches 89.3%,which proves that the method has good detection and segmentation performance.