Resumen
In this study, we developed a predictive model that employs data from multiparametric structural MRI to predict local recurrence in glioblastoma, providing a practical solution to an issue clinicians face in our daily practice: discriminating edema from tumor infiltration. Predicting the location of these areas at high risk of recurrence will potentially allow for personalizing and optimizing the local treatment of glioblastomas, creating new surgical resection limits and radiotherapy targets. Our findings could potentially improve the survival rate of these patients and open a new line of research that permits a better understanding of the mechanisms of glioma invasion. In addition, we evaluated our results in an external multicenter cohort of patients, thus demonstrating the applicability of the model despite the MRI acquisition protocols and scanner manufacturers. The model will be publicly available through a repository for its implementation by any institution.