Abstract

Info

Author(s):: ZHANG Qianwen1; CHEN Ming1; 2; QIN Yufang1; 2; CHEN Xi1; 1. College of Information Technology, Shanghai Ocean University, Shanghai 201306, China; 2. Key Laboratory of Fisheries Information, Ministry of Agriculture, Shanghai 201306, China

Keywords:: pulmonary nodule; segmentation; deep residual structure; recall rate; ResUnet

Abstract:: Objective To propose a novel network ResUnet by combining deep residual structure with U-Net network, and to extract lung nodule region by segmenting the chest CT image with the use of ResUnet deep learning network structure. Methods The CT image data used in the study were derived from LUNA16 dataset. The lung parenchyma was firstly extracted from CT image preprocessing, and then, the stereo image block was intercepted and the simple size was expanded by data enhancement, thereby obtaining the corresponding lung nodule mask image. Finally, the obtained simple were imported into ResUnet model for training. Results The final accuracy and recall rate of the proposed model were 35.02% and 97.68%, respectively. Conclusion The proposed model can automatically learn the characteristics of pulmonary nodules and provide a reliable basis for the subsequent automatic diagnosis of lung cancer, thus reducing the cost of clinical diagnosis and shortening the time for diagnosis.