Detection And Diagnosis Of Pulmonary Nodules Based On Deep Learning

Lung cancer is the most common and one of the most morbid and lethal cancers.Early diagnosis and treatment of lung cancer can significantly improve the patient’s five-year survival rate.Lung cancer is caused mainly by malignant pulmonary nodules.This means that the study of early detection and diagnosis of malignant pulmonary nodule technology has important scientific and clinical significance.In recent years,deep learning technology has developed rapidly.It has made groundbreaking breakthroughs in image,speech,natural language processing and big data feature extraction and so on.This paper mainly studies the detection of lung nodules and the benign and malignant diagnostic algorithms based on deep learning technology.The main contributions are as follows:This paper presents a new lung nodule detection algorithm.The algorithm consists of two parts.The first part generates candidate lung nodules in the lung CT images.The second part filters out false-positive lung nodules in candidate lung nodules and outputs the final test result.Experimental results show that this algorithm has great advantages in detecting pulmonary nodules and filtering out false positives.In this paper,aiming at the benign and malignant diagnosis of pulmonary nodules,we proposes an improved 3DCNN(Three-Dimensional Convolutional Neural Network)diagnostic algorithm.In order to improve the accuracy of the diagnostic algorithm of pulmonary nodules,this paper uses 3DGAN(Three-Dimensional Generative Adversarial Networks)to generate three-dimensional pulmonary nodules.It can increase the number of lung nodule samples.Experimental results show that the accuracy of the algorithm is significantly improved by training data sets of pulmonary nodules.Through the study of end-to-end computer-aided detection and diagnosis of pulmonary nodule system to help doctors diagnose lung cancer and improve patient survival rate.