Tumor grading of soft tissue sarcomas using MRI-based radiomics

Area of Science:

• Radiology
• Oncology
• Medical Imaging Analysis

Background:

• Accurate grading of soft tissue sarcoma (STS) is crucial for treatment decisions.
• Differentiating low-grade (G1) from high-grade (G2/G3) STS impacts multimodal therapy planning.
• MRI-based radiomics offers a potential non-invasive method for STS grading.

Purpose of the Study:

• To develop and validate MRI-based radiomics models for differentiating low-grade and high-grade STS.
• To assess the performance of radiomics models using different MRI sequences (T2FS, T1FSGd).
• To evaluate the added value of radiomics models combined with clinical staging for prognosis.

Main Methods:

• Retrospective analysis of two independent STS patient cohorts.
• Extraction of 1394 radiomics features from T2FS and T1FSGd MRI sequences.
• Development of Least Absolute Shrinkage and Selection Operator (LASSO) models using nested cross-validation.
• Assessment of ComBat Harmonization for batch effect correction.

Main Results:

• Radiomics models achieved AUCs of 0.78 (T2FS), 0.69 (T1FSGd), and 0.76 (combined) on the validation set.
• The T2FS-based model demonstrated superior reproducibility.
• Integration of the T2FS radiomics model with clinical staging improved prognostic performance and clinical net benefit.

Conclusions:

• MRI-based radiomics tumor grading models effectively classify low-grade and high-grade soft tissue sarcomas.
• These models can aid in patient stratification and improve prognostic accuracy.
• Radiomics analysis holds promise for non-invasive grading of STS.