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

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Distinguishing benign from malignant subcentimeter solid pulmonary nodules (SSPNs) is a clinical challenge.
• Accurate early diagnosis is crucial for effective treatment strategies.

Purpose of the Study:

• To develop a predictive nomogram for SSPNs by integrating radiomic features and clinical risk factors.
• To improve diagnostic accuracy and aid clinical decision-making for SSPNs.

Main Methods:

• Retrospective analysis of 415 patients with SSPNs.
• Development of clinical and radiologic models using machine learning algorithms.
• Integration of optimal models into a comprehensive nomogram for performance evaluation using AUC, calibration curves, and DCA.

Main Results:

• Multiplanar volume rendering (MPVR)-maximum diameter and margin were identified as independent predictors of malignancy.
• The comprehensive nomogram achieved high AUC values (0.965 in training, 0.966 in test set) for distinguishing benign from malignant SSPNs.
• The nomogram demonstrated superior clinical utility and calibration compared to other models.

Conclusions:

• The developed nomogram integrating radiomic features (IntraPeri3mm model) and clinical factors (MPVR-maximum diameter) shows optimal predictive efficiency for SSPNs.
• This tool provides a valuable scientific basis for early diagnosis and treatment of SSPNs.
• MPVR-maximum diameter is a significant independent factor in predicting SSPN malignancy.