Abstract

Info

Title:: Application of improved deep learning model in differential diagnosis of benign and malignant breast tumors

Author(s):: DENG Zhuqin1; YU Yongwei2; 1. Department of Obstetrics and Gynecology, The Peoples Liberation Army Joint Service Support Unit No.901 Hospital, Hefei 230031, China 2. Department of General Surgery, Changrong Hospital, Hefei 230001, China

Keywords:: Keywords: Convolutional neural network breast tumor cell recognition image recognition

Abstract:: Abstract: Objective To improve the defects and deficiency of traditional methods in clinical pathological image detection and to solve the problem of misjudgment made by human. Methods The data sets of breast tumor cells are from clinical date. The data sets were enhanced two fold at first and put into the proposed model for training. Results After 100 times of iterations, the accuracy of validation set arrives 97.44%, the test set 96.4%, and the recall rate 95.5%.They are obviously improved compared with the same type of literature. Conclusion The improved convolutional neural network proposed in this paper has advantages of rapid convergence and excellent generalization ability. It can identify and classify the benign and malignant breast tumor cells effectively.