An Accurate Urine Erythrocytes Detection Model Coupled Faster RCNN with VggNet
Keshu Li, Ming Li, Yongfei Wu, Xinyu Li, Xiaoshuang Zhou · 2020
In the process of kidney disease diagnosis, it is particularly important to screen the morphology of erythrocytes in urine. Because of the small and similar morphology of erythrocytes in urine, manual labeling and observation is time-consuming and labor-consuming. Deep learning can not only quickly and accurately locate urine erythrocytes, but also effectively identify the types of cells. In this paper, we employ the Faster RCNN combined with VggNet to detect urine erythrocytes. We collected images of urine erythrocytes under optical microscope and used data enhancement method to obtain total 3969 images of urine erythrocytes, which filled in the blank of domestic urine erythrocytes data sets. Experimental results show that the presented method can identify five kinds of urine erythrocytes and the recall rate achieve up to 99.8%, which greatly helped doctors better identify and judge urine erythrocytes.