A schizophrenia study based on multi-frequency dynamic functional connectivity analysis of fMRI.

Area of Science:

• Neuroimaging
• Neuroscience
• Computational Psychiatry

Background:

• Current fMRI studies often analyze a broad low-frequency band (0.01-0.08 Hz).
• Neuronal activity is dynamic, suggesting different frequency bands may hold distinct information.
• Schizophrenia research can benefit from more nuanced analysis of brain connectivity.

Purpose of the Study:

• To propose and validate a novel multi-frequency dynamic functional connectivity (dFC) analysis method.
• To investigate if combining different frequency bands enhances the identification of brain alterations in schizophrenia.
• To improve the classification accuracy of schizophrenia using advanced dFC analysis.

Main Methods:

• Utilized Fast Fourier Transform to extract three frequency bands: Conventional (0.01-0.08 Hz), Slow-5 (0.0111-0.0302 Hz), and Slow-4 (0.0302-0.0820 Hz).
• Employed fractional amplitude of low-frequency fluctuations to identify abnormal regions of interest (ROIs) in schizophrenia.
• Applied sliding time window method with varying window-widths for dFC analysis, followed by recursive feature elimination and support vector machine classification.

Main Results:

• The proposed multi-frequency method, combining Slow-5 and Slow-4 bands, demonstrated superior classification performance over the conventional method, particularly with shorter sliding windows.
• Identified variations in dFC among abnormal ROIs across different frequency bands.
• Confirmed that integrating features from multiple frequency bands enhances classification accuracy for schizophrenia.

Conclusions:

• Dynamic functional connectivity patterns differ across frequency bands in schizophrenia.
• Combining multiple frequency bands in dFC analysis offers a promising approach for detecting brain alterations in schizophrenia.
• The novel multi-frequency dFC method improves diagnostic classification accuracy, aiding in schizophrenia research.