Establishment and validation of a prediction model for high-risk patients with solitary pulmonary nodules based on high-resolution CT scans(PDF)
《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]
- Issue:
- 2023年第2期
- Page:
- 190-195
- Research Field:
- 医学影像物理
- Publishing date:
Info
- Title:
- Establishment and validation of a prediction model for high-risk patients with solitary pulmonary nodules based on high-resolution CT scans
- Author(s):
- ZHANG Mi1; TAN Yanhuan1; QIAN Xin1; WU Zhichao2; GENG Yue1; SHAO Yu1; YU Yang1; LI Junchen1
- 1. Department of Radiology, Changshu Hospital Affiliated to Nanjing University of Chinese Medicine, Changshu 215500, China
2. Department of Thoracic Surgery, Changshu Hospital Affiliated to Nanjing University of Chinese Medicine, Changshu 215500, China
- Keywords:
- Keywords: prediction model solitary pulmonary nodule CT nomogram receiver operating characteristic curve
- PACS:
- R734.2；R814.42
- DOI:
- DOI:10.3969/j.issn.1005-202X.2023.02.011
- Abstract:
- Objective To screen out the high-resolution CT sign of high-risk lesions of solitary pulmonary nodules (SPN) for establishing a prediction model for high-risk SPN, and to conduct an internal validation of the model. Methods The chest imaging signs of 317 patients with SPN were retrospectively analyzed, and the patients were randomly divided into training set and validation set in a ratio of about 6:4. The independent risk factors associated with high-risk SPN were screened out using univariate and multivariate Logistic stepwise regressions, and a prediction model was established. The established model is verified using validation set, and the receiver operating characteristic (ROC) curve is drawn to evaluate the predictive value of the model. Results Univariate analysis showed that there were significant differences between high- and low-risk SPN in the following imaging signs: location, maximum diameter, CT value, spicule sign, vacuole sign, lobulation sign (P<0.05), and that the proportion of clear or unclear boundary in high-risk SPN group was similar to that in low-risk SPN group (P>0.05). Multivariate analysis showed that 4 factors including location, maximum diameter, CT value and lobulation sign were independent predictors for high-risk SPN (P<0.05). The prediction model constructed based on training set was as follow: P=ex/(1+ex), where the e is the natural logarithm, x=-7.767+(2.821×location)+(0.391×maximum diameter)-(0.003×CT value)+(3.576×lobulation). The area under the ROC curve (AUC), 95%CI, the optimal cutoff value, sensitivity, and specificity were 0.932, 0.892-0.973, 0.208, 91.7%, and 85.6% in training set, and 0.911, 0.847-0.975, 0.268, 78.8%, and 93.1% in validation set. Conclusion Location, maximum diameter, CT value, and lobulation sign are independent influencing factors for high-risk SPN, and the mathematical prediction model established by these independent influencing factors has high accuracy and is worth to be popularized in clinical practice.
Last Update: 2023-03-03