A contrast-enhanced MRI-based nomogram to identify lung metastasis in soft-tissue sarcoma: A multi-centre study

Area of Science:

• Radiology
• Oncology
• Medical Imaging

Background:

• Lung metastasis (LM) is critical for soft-tissue sarcoma (STS) treatment decisions.
• MRI-based prediction of LM in STSs requires further investigation.

Purpose of the Study:

• To develop and validate MRI-based radiomics models for predicting LM in STSs.
• To assess the performance of a clinical-radiomics nomogram for LM prediction.

Main Methods:

• Extracted radiomics features from T1-weighted contrast-enhanced (T1-CE) MRI scans in 122 STS patients (primary cohort) and 32 (external validation).
• Utilized LASSO for feature selection to build a radiomics signature.
• Constructed a multivariable logistic regression nomogram integrating the radiomics signature and margin status.

Main Results:

• A nomogram integrating radiomics signature and margin achieved superior prediction performance across training, internal, and external validation sets.
• Area Under the Curve (AUC) values for the nomogram were 0.918 (training), 0.864 (internal validation), and 0.843 (external validation).
• The nomogram outperformed the radiomics signature and margin alone in all validation sets.

Conclusions:

• The developed nomogram is a promising tool for preoperative treatment strategy planning in STS patients.
• MRI-based radiomics can significantly improve the prediction of lung metastasis in STSs.