Study on the Detection of Pulmonary Nodules in CT Images Based on Deep Learning
Gai Li, Wei Zhou, Weibin Chen, Fengtao Sun, Yu Fu, Fengling Gong, Huiying Zhang
Abstract
With the development of medical imaging technology and the introduction of computed tomography (CT), early screening for lung cancer is becoming more and more possible. In this paper, we introduce the method of wavelet dynamic analysis to extract and repair the lung parenchyma, so as to exclude the noise interference outside the lung parenchyma. The algorithm can help us to locate the lung nodules with higher accuracy. Then, the convolution neural network (CNN) optimized by genetic algorithm and the traditional CNN are used to extract the features of CT image of pulmonary nodules. The corresponding features of different images are automatically distinguished. By comparing the accuracy of the two algorithms, it is proved that the CNN optimized by genetic algorithm has higher accuracy. Finally, the CNN optimized by genetic algorithm is used to detect and classify the existing pulmonary nodule images, which provides guidance for CT image detection technology of pulmonary nodule.