Intratumoral and peritumoral CT radiomics-clinical feature fusion model for differentiating pulmonary tuberculosis and pulmonary cryptococcosis nodules

Abstract Objective To investigate the clinical value of a fusion model combining intratumoral and peritumoral CT radiomics features with clinical features for differentiating pulmonary tuberculosis (PTB) from pulmonary cryptococcosis (PC) nodules, and to develop a predictive tool with strong robustness and generalizability. Methods A total of 214 patients with PTB or PC were consecutively enrolled and randomly divided into a training set (n = 149) and a test set (n = 65) in a 7:3 ratio. Radiomics features were systematically extracted from the intratumoral region and from peritumoral regions extending 1 mm, 3 mm, and 5 mm beyond the lesion boundary on CT images. Least absolute shrinkage and selection operator (LASSO) regression was used for optimal feature selection, and the selected features were incorporated into various machine learning algorithms, including Support Vector Machine (SVM), to construct radiomics models. Meanwhile, a clinical model was established using independent clinical risk factors. Clinical and radiomics features were then integrated to construct model. Model performance was comprehensively evaluated using Receiver Operating Characteristic (ROC) curves, calibration curves, and Decision Curve Analysis (DCA). Results The clinical model alone showed limited predictive performance (AUC = 0.693 in the test set) and evidence of overfitting. Among the radiomics models, the model based on 3 mm peritumoral features demonstrated superior performance (AUC = 0.911 in the test set). The combined model, integrating clinical variables with intratumoral and 3 mm peritumoral radiomics features, exhibited the highest diagnostic performance, with AUCs of 0.979 and 0.944 in the training and test sets, respectively, and a sensitivity of 0.933 and a specificity of 0.886. Calibration curves indicated that good accuracy between predicted and observed outcomes, and DeLong’s test confirmed that the combined model outperformed the individual models in most comparisons. Conclusion A combined model integrating intratumoral and peritumoral radiomics features together with clinical variables can compensate for the limitations of the clinical model alone. It significantly improves diagnostic performance and generalizability in differentiating PTB from PC, and may serve as a precise and robust non-invasive tool to support clinical decision-making.