Lung Nodule Recognition And Detection Based On Deep Learning

Deep Learning has been widely applied in natural language processing,image recognition,target detection and other fields.With the development of deep learning,medical imaging diagnosis has become an important direction of artificial intelligence in the medical field.Deep learning algorithm can identify and detect lung nodules.Diagnosis of lung nodules using machines can prevent doctors from misdiagnosing due to fatigue.At the same time,radiologists need to diagnose a large number of medical images every day,which is inefficient.Using machine to assist doctors to diagnose can greatly improve the diagnostic speed.At present,deep learning has made some progress in the identification and detection of pulmonary nodules,but there are still many problems.First of all,it is difficult to distinguish between benign and malignant pulmonary nodules,the diagnostic accuracy of convolutional neural networks does not meet the actual application requirements.Secondly,the number of labeled lung nodules in chest radiographs is too small to reach the amount of data required for convolution neural network training.Finally,there are some missed diagnosis and misdiagnosis by using computer to diagnose pulmonary nodules.In view of the above problems,this article studied the benign and malignant diagnosis of pulmonary nodules and the direction of pulmonary nodule detection on X-ray chest radiographs.1.A new activation function and reconfiguration network is proposed for capsule network(CapsNet).The improved capsule network reduces the number of neural network parameters and improves the accuracy of identification.The automatic diagnosis of benign and malignant pulmonary nodules is achieved.2.Deconvolution has the characteristic of restoring images,and a reconstruction network using deconvolution is proposed to improve the structure of traditional convolutional neural networks.By minimizing the variance between the reconstructed network output and the input of a conventional convolutional neural network(CNN),the recognition accuracy of convolutional neural network models on benign and malignant pulmonary nodules is improved.3.In view of the fact that the number of public medical image data sets is pool and the number of detected lesions is less,etc.The medical image data acquisition and lesion labeling methods are introduced.The location of lung nodules was marked by ChestXray8.In this paper,we propose Xray-yolov3 model to achieve high accuracy automatic detection of pulmonary nodules in Chest X-ray.Realizing high-precision automatic detection of pulmonary nodules in X-ray chest radiographs.This paper improves the capsule network and convolutional neural network.The improved neural network has improved recognition accuracy on benign and malignant pulmonary nodule datasets.This article presents the Xray-yolov3 model to achieve highprecision detection of X-ray chest radiographs.It is an important applied research.