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

Area of Science:

• Neuroimaging
• Brain Network Analysis
• Medical Physics

Background:

• Standard MRI identifies tumor core and edema but often overlooks signals within these regions.
• The functional and diffusion signals within tumors and their impact on brain connectivity are poorly understood.

Purpose of the Study:

• To explore functional activity and white matter structure considering the whole tumor.
• To investigate the relationship between tumor signals and global connectivity reorganization.
• To predict postsurgical brain network changes using preoperative data.

Main Methods:

• Analysis of resting-state functional signals in the frequency domain.
• Development of a fiber tracking pipeline for tumoral and peritumoral white matter.
• Application of machine learning to predict structural rearrangement based on preoperative brain networks.

Main Results:

• Intertwined alterations found in local and distributed functional signals within the tumor.
• Successful reconstruction of white matter bundles in tumoral and peritumoral tissues.
• Machine learning model accurately predicted postsurgical structural rearrangement and disentangled tumor-specific connectivity patterns.

Conclusions:

• MR signals within damaged brain tissues are critical for understanding brain tumors.
• Tumors exhibit and relate to complex patterns of structural and functional (dis-)connections.
• This approach highlights the importance of comprehensive neuroimaging for surgical planning and outcome prediction.