CT Radiomic Nomogram Using Optimal Volume of Interest for Preoperatively Predicting Invasive Mucinous Adenocarcinomas...

Area of Science:

• Radiology
• Oncology
• Medical Imaging Analysis

Background:

• Differentiating invasive mucinous adenocarcinoma (IMA) from non-IMA in incidental pulmonary nodules (IPNs) is crucial for treatment planning.
• Computed tomography (CT) imaging is essential for characterizing pulmonary nodules.
• Radiomic analysis offers potential for non-invasive preoperative differentiation.

Purpose of the Study:

• To evaluate the efficacy of radiomic analysis with optimal volumes of interest (VOIs) in differentiating IMA from non-IMA in patients with IPNs.
• To develop and validate a predictive model integrating radiomic features and image findings.

Main Methods:

• A multicenter retrospective study of 1383 patients with IPNs, including 110 with IMA.
• Extraction of radiomic features from multi-scale VOI subgroups (VOI-2 mm, VOIentire, VOI+2 mm, VOI+4 mm).
• Application of synthetic minority oversampling technique for class imbalance, least absolute shrinkage and selection operator for feature selection, and development of a combined nomogram.

Main Results:

• An image-finding classifier (bubble lucency, lower lobe predominance) achieved an AUC of 0.684.
• The VOI+2 mm-based radiomic model showed the highest performance with an AUC of 0.832.
• The combined nomogram integrating radiomics and image findings achieved the best performance with an AUC of 0.850.

Conclusions:

• A nomogram combining a classifier with an optimal VOI-based radiomic model effectively predicts IMA in IPNs.
• This approach aids physicians in preoperative differentiation and development of tailored treatment strategies.
• Radiomics holds significant promise for improving the diagnostic accuracy of pulmonary nodule characterization.