Prediction Models for Solitary Pulmonary Nodules Based on Texture of 2D Images in Chinese Han People
Background Lung cancer has been the most common cancer in the world, which usually appears as solitary pulmonary nodules (SPNs), In this paper, three prediction models were used to improve the diagnosis of benign and malignant SPNs using textural features and other parameters.Methods The textural features and medical records from 502 patients (275 men, 227 women) consisting of 4742 CT slices, with a pathological diagnosis of SPNs, were reviewed to use as training test.Combining curvelet-transformation textural features and 11 other characteristics, multilevel model, least absolute shrinkage and selection operator (LASSO) regression method and support vector machine (SVM) prediction model were used to gain initial insights into the data.An additional 18 patients (10 men, 8 women) with pathological diagnosis of SPNs were used to validate the models.Results The least areas under the receiver operating characteristic (ROC) curves were 0.855±0.024 in SVM and the highest value 0.908 ± 0.017 in LASSO, both with P<0.001.The highest Youdens index (75.46%) and positive predictive value (69.12%), were achieved in LASS0, the highest sensitivity (98.96%) and negative predictive value (99.18%) were got in multilevel model, and the highest sensitivity (98.96%) and negative predictive value (99.18%) were resulted from SVM.Conclusions With textural features extracted from a Curvelet transformation and and other features, these prediction models can increase the rate of diagnosis for lung cancer.Which model should be selected depended on what was the point people focus on and further studies were required.
lung cancer solitary pulmonary nodule curvelet transformation
Jingjing Wang Haifeng Wu Tao Sun Wei Wang Xia Li Lixin Tao Da Huo Pingxin Lv Wen He Xiuhua Guo
School of Public Health, Capital Medical University, Beijing, China;Beijing Municipal Key Laboratory School of Public Health, Capital Medical University, Beijing, China;Beijing Municipal Key Laboratory School of Public Health, Capital Medical University, Beijing, China;Beijing Municipal Key Laboratory Department of Radiology, Beijing Chest Hospital, Capital Medical University, Beijing, China Department of Radiology, Friendship Hospital, Capital Medical University, Beijing,China
国际会议
北京
英文
32-42
2013-06-01（万方平台首次上网日期，不代表论文的发表时间）