A Comparative Study of Correlation Methods in Functional Connectivity Analysis Using fMRI Data of Alzheimer's...

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Medical Diagnostics

Background:

• Functional Magnetic Resonance Imaging (fMRI) is a non-invasive technique for brain function research, detecting brain activation via oxygen consumption.
• Brain functional connectivity analysis explores relationships between brain regions using time-series data.

Purpose of the Study:

• To compare the efficacy of different correlation methods (Pearson, Kendall, Spearman) for analyzing brain functional connectivity in Alzheimer's Disease (AD).
• To assess global and nodal network measures derived from fMRI data in AD patients versus healthy controls.

Main Methods:

• fMRI data from Alzheimer's Disease (AD) patients and healthy individuals were obtained from the ADNI database.
• Brain functional networks were constructed using Pearson, Kendall, and Spearman correlation coefficients.
• Global and nodal network metrics were calculated for the whole brain and the Default Mode Network (DMN).
• Statistical significance was determined using non-parametric permutation tests.

Main Results:

• While nodal analysis showed similar performance across correlation methods, Spearman and Kendall outperformed Pearson in identifying global differences in AD patients.
• Nodal analysis indicated greater functional connectivity disruption in posterior brain regions compared to frontal areas in AD.
• The dominant hemisphere exhibited more pronounced disruptions in functional connectivity due to AD.

Conclusions:

• Non-linear correlation methods like Spearman and Kendall offer advantages over the commonly used Pearson method for detecting global brain connectivity alterations in AD.
• Pearson correlation's limitations in capturing non-linear relationships necessitate exploring advanced methods like distance correlation for comprehensive fMRI analysis.