Connectome-based markers predict the sub-types of frontotemporal dementia

Area of Science:

• Neuroscience
• Neuroimaging
• Computational Psychiatry

Background:

• Frontotemporal dementia (FTD) is a group of neurodegenerative diseases with diverse clinical presentations.
• Understanding the neurobiological differences between FTD subtypes is crucial for diagnosis and treatment.
• Connectome analysis offers a powerful tool to investigate brain network alterations in FTD.

Purpose of the Study:

• To investigate alterations in functional brain module organization across different FTD subtypes.
• To identify potential neuroimaging biomarkers for FTD subtypes.
• To enhance the prediction accuracy of FTD subtypes using machine learning.

Main Methods:

• Resting-state functional magnetic resonance imaging (fMRI) data from patients with behavioral variant FTD (BV-FTD), semantic variant FTD (SV-FTD), progressive non-fluent aphasia FTD (PNFA-FTD), and healthy controls.
• Voxel-level functional brain network construction and binarization.
• Calculation of modular segregation index (MSI) and participation coefficient (PC) to assess network integrity and nodal properties.

Main Results:

• BV-FTD and SV-FTD showed decreased MSI in subcortical, default mode (DMN), and ventral attention networks (VAN).
• BV-FTD exhibited disrupted frontoparietal network (FPN) integrity compared to other groups.
• All FTD subtypes displayed altered connectivity involving the insular cortex, with significant associations between network changes and clinical variables.

Conclusions:

• FTD subtypes exhibit distinct patterns of brain module organization, highlighting shared and unique neurobiological underpinnings.
• Connectome-based machine learning models demonstrate high performance in differentiating FTD subtypes.
• These findings offer potential biomarkers for improved FTD subtype diagnosis and personalized treatment strategies.