Intratumoral and peritumoral MRI-based radiomics prediction of histopathological grade in soft tissue sarcomas: a two-center study

ObjectivesThis study aims to develop a model based on intratumoral and peritumoral radiomics from fat-suppressed T2-weighted(FS-T2WI) images to predict the histopathological grade of soft tissue sarcoma (STS).MethodsThis retrospective study included 160 patients with STS from two centers, of which 82 were low-grade and 78were high-grade. Radiomics features were extracted and selected from the region of tumor mass volume (TMV) and peritumoral tumor volume (PTV) respectively. The TMV, PTV, and combined(TM-PTV) radiomics models were established in the training cohort (n = 111)for the prediction of histopathological grade. Finally, a radiomics nomogram was constructed by combining the TM-PTV radiomics signature (Rad-score) and the selected clinical-MRI predictor. The ROC and calibration curves were used to determine the performance of the TMV, PTV, and TM-PTV models in the training and validation cohort (n = 49). The decision curve analysis (DCA) and calibration curves were used to investigate the clinical usefulness and calibration of the nomogram, respectively.ResultsThe TMV model, PTV model, and TM-PTV model had AUCs of 0.835, 0.879, and 0.917 in the training cohort and 0.811, 0.756, 0.896 in the validation cohort. The nomogram, including the TM-PTV signatures and peritumoral hyperintensity, achieved good calibration and discrimination with a C-index of 0.948 (95% CI, 0.906 to 0.990) in the training cohort and 0.921 (95% CI, 0.840 to 0.995) in the validation cohort. Decision curve analysis demonstrated the clinical usefulness of the nomogram.ConclusionThe proposed model based on intratumoral and peritumoral radiomics showed good performance in distinguishing low-grade from high-grade STSs. center dot Peritumoral and intratumoral radiomics models based on FS-T2WI images had a good predictive ability for the histopathological grading of STSs.center dot Peritumoral hyperintensity and radiomic scores were independently associated with histopathological grading of STSs.center dot The nomogram integrating clinic features and radiomics scores demonstrated good performance for predicting the histopathological grading of STSs.

Automated summary

This study developed a model based on intratumoral and peritumoral radiomics to predict the histopathological grade of soft tissue sarcoma (STS). The model showed good performance in distinguishing low-grade from high-grade STSs.