Development and validation of a CT-based comprehensive nomogram for differentiating benign from malignant subcentimeter solid nodules.

Background: Accurate discrimination between benign and malignant subcentimeter solid pulmonary nodules (SSPNs, <1 cm) remains a clinical challenge. This study aims to develop a nomogram integrating intratumoral and peritumoral radiomic features with clinical risk factors to aid clinicians in early diagnosis and precise decision-making for SSPNs. Methods: A total of 415 patients with SSPN who underwent surgical resection in The First Affiliated Hospital of Chongqing Medical University were retrospectively enrolled in this study, and they were divided into a training set and a test set at a ratio of 7:3. Independent risk factors were screened out through univariate and multivariate analyses to construct a clinical model. Radiological features within and around the tumor were extracted and filtered from computed tomography (CT) images, and four machine learning algorithms were used to build radiological models for intratumoral, peritumoral, and combined multi-feature characteristics respectively. Based on the above analyses, an optimal radiological model and independent clinical predictors were integrated to establish a comprehensive nomogram. The area under the curve (AUC) was used to evaluate the model performance, and the calibration curve and decision curve analysis (DCA) were applied to assess its clinical utility. Results: Clinical baseline analysis showed that there were significant differences between the benign and malignant groups in terms of age, lung window-maximum diameter, multiplanar volume rendering (MPVR)-maximum diameter, shape, margin, border, vascular bundle, vacuole, and pleural indentation. Multivariate analysis further confirmed that margin [odds ratio (OR) =0.578; 95% confidence interval (CI): 0.367-0.910] and MPVR-maximum diameter (OR =1.175; 95% CI: 1.103-1.252) were independently associated with the occurrence of malignant tumors. Various CT-based radiological features performed well in distinguishing benign and malignant SSPN, among which the IntraPeri3mm model had the best performance in the test set, with an AUC of 0.891, a sensitivity of 0.764, and a specificity of 0.848. After integrating this model with the maximum diameter based on multiplanar reconstruction (MPVR maximum diameter) and margin features, the comprehensive nomogram achieved the highest AUC values of 0.965 (sensitivity 0.959, specificity 0.842) and 0.966 (sensitivity 0.855, specificity 0.935) in the training set and test set respectively. DCA and calibration curve analysis indicated that this nomogram was superior to other clinical and radiological models in terms of net benefit and calibration ability, providing valuable reference information for treatment decisions. Conclusions: This study confirms that MPVR-maximum diameter is an independent factor for predicting the malignancy of SSPN. By comparing various radiological models constructed based on high-resolution CT (HRCT) images, it is found that the nomogram integrating the IntraPeri3mm model and MPVR-maximum diameter has the optimal predictive efficiency. The above results provide an important scientific basis for the early diagnosis and treatment of SSPN.