CBV-informed MR Fingerprinting for Improved Classification of Intratumoral Heterogeneity in Post-Treatment Glioblastomas

Motivation: Post-treatment glioblastoma lesions often present with both local tumor recurrence and radiation necrosis. Segmenting within-lesion heterogeneity is clinically challenging. Cerebral blood volume (CBV) maps can inform high-risk areas, but with limited specificity. Goal(s): We aim to extend the use of MR Fingerprinting (MRF) from whole-lesion classification to better characterize within-lesion heterogeneity and local recurrence, with histopathological validation. Approach: We propose 1) a retrospective cohort (n=97) that develops MRF-CBV TR signatures from CBV-defined high-risk regions, and 2)a prospective validation cohort (n=14) that correlates the developed metrics with multi-location histopathological samples within lesion, and compares performance of CBV-defined versus biopsy-defined MRF signatures. Impact: The MRF/CBV-informed local tumor recurrance signatures can improve characterization of intratumoral heterogeneity beyond CBV in recurrent glioblastomas. This pipeline will lead to more precise tumor sampling and better treatment response evaluation.