Functional and structural reorganization in brain tumors: a machine learning approach using desynchronized functional oscillations

This study investigates functional and structural brain network alterations in brain tumor patients using a machine learning approach. The researchers analyze fMRI data to identify changes in the frequency domain of resting-state functional signals, potentially originating within the tumor itself. They also develop a novel fiber tracking pipeline to reconstruct white matter fiber bundles in tumoral and peritumoral tissue, combining anatomical information with diffusion MRI data. Finally, they employ a machine learning model to predict structural connectivity changes after surgery, considering preoperative brain network information and disentangling reorganization patterns for different tumor types. The findings highlight the importance of analyzing MR signals within damaged brain tissue due to the presence of non-trivial structural and functional (dis)connections and activity patterns.

Joan Falcó-Roget, Alberto Cacciola, Fabio Sambataro, Alessandro Crimi