Invasive Prediction of Ground Glass Nodule Based on Clinical Characteristics and Radiomics Feature.

Area of Science:

• Radiology
• Oncology
• Medical Imaging Analysis

Background:

• Ground-glass nodules (GGNs) in CT scans present diagnostic challenges for invasive lung adenocarcinoma.
• Accurate invasive classification is crucial for effective treatment planning.

Purpose of the Study:

• To evaluate the diagnostic value of CT radiographic images and radiomics features for invasive classification of lung adenocarcinoma manifesting as GGNs.
• To develop and validate a combined model and nomogram for predicting GGN invasiveness.

Main Methods:

• Retrospective analysis of 312 GGNs, divided into training and test sets.
• Development of a clinical model using logistic regression and a radiomics model using mRMR and LASSO feature selection.
• Construction of a combined model and a nomogram for predicting invasiveness.

Main Results:

• Clinical predictors for invasive adenocarcinoma included nodule diameter, lobulation, and vascular changes.
• The combined model demonstrated superior diagnostic performance (AUC: 0.86 training, 0.80 test) compared to individual models.
• The nomogram achieved a C-index of 0.855, effectively quantifying risk.

Conclusions:

• Radiographic and radiomics features offer high accuracy for invasive GGN diagnosis.
• Combined analysis significantly improves diagnostic efficacy for invasive adenocarcinoma.
• The nomogram serves as a noninvasive tool to guide pre-surgical assessment and personalized treatment.