Our pilot study results suggest that features derived from mp-MRI can be valuable for assessing treatment responses in rhabdomyosarcoma (RMS). A pre-surgical model that utilized the high-order radiomic feature CE-SWI GLSZM Large Area High Gray Level Emphasis achieved perfect classification performance (AUC=1.0) in differentiating responders from partial or non-responders. The CE-SWI-based texture radiomic model, and to a lesser degree, the recognition of the CE-SWI Complete Ring, can suggest successful treatment outcomes in RMS with a treatment effect of over 90% PATE. This model appears more effective than RECIST in predicting response. On RMS treatment assessment, CE-SWI outperforms RECIST and the predictive value of DWI/ADC and PWI/DCE. Comparable results have been described in undifferentiated pleomorphic sarcomas (UPS) and leiomyosarcomas (LMS). The sample size limited the present study, hindering the possibility of demonstrating a higher statistical significance. An expansion study with a larger sample will be needed to validate these promising results further. Institutions that have not yet incorporated CE-SWI into their tumor MRI protocols could consider developing and including CE-SWI. The present and previous publications regarding UPS, LMS, and desmoid tumors outline CE-SWI’s utility in assessing soft tissue tumors.

Our pre-surgical multiparametric MRI (Mp-MRI) model using high-order radiomic CE-SWI GLSZM achieved perfect classification performance (AUC=1.0), appearing as valuable in predicting RMS response, outperforming RECIST (AUC=0.67).