Advances in neuroimaging have enabled non-invasive mapping of brain networks in patients with brain tumors. While conventional MRI reliably detects the tumor core and surrounding edema, signals and functions within these regions are often overlooked. As a result, the role of functional and diffusion signals, as well as their contribution to global connectivity reorganization, remains insufficiently understood. In this work, we investigate functional activity and white matter architecture while explicitly accounting for the entire tumor in a surgical context. Our findings reveal intertwined alterations in both local and spatially distributed resting-state functional signals, detectable in the frequency domain and potentially originating from within the tumor. We further introduce a fiber-tracking framework that integrates anatomical priors yet remains capable of reconstructing pathways in tumoral and peritumoral tissue. Finally, by applying machine learning and normative anatomical data, we predict post-surgical structural rearrangements from preoperative networks. This generative approach also disentangles tumor-type–specific patterns of connectivity reorganization. Taken together, our results highlight the necessity of incorporating MR signals from damaged brain regions, as they reflect complex and non-trivial interactions between structural and functional (dis)connectivity.
Functional and structural reorganization in brain tumors: a machine learning approach using desynchronized functional oscillations
Autors: Joan Falcó-Roget, Alberto Cacciola, Fabio Sambataro, Alessandro Crimi
DOI: 10.1038/s42003-024-06119-3
READ HERE