Activation pattern reproducibility: measuring the effects of group size and data analysis models

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Biostatistics

Background:

• Traditional reproducibility assessment in brain activation studies focuses on localized foci above a statistical threshold.
• A comprehensive evaluation of image-wide pattern reproducibility is lacking.
• Developing new methods is crucial for understanding the reliability of neuroimaging findings.

Purpose of the Study:

• To introduce and validate an alternative approach for assessing brain activation reproducibility using image-wide signal patterns.
• To quantitatively measure reproducibility using pattern similarity metrics.
• To investigate the influence of experimental variables, such as group size and analysis models, on reproducibility.

Main Methods:

• Utilized scatter plots to compare signal levels across all Talairach voxels for pairs of functional activation images.
• Employed Pearson product-moment correlation (rho) to quantitatively summarize image-wide, signal-level reproducibility.
• Generated empirical population distributions of rho using statistical resampling techniques for [(15)O]-water PET scans of a simple motor task.

Main Results:

• Demonstrated the utility of empirical rho-histograms for measuring reproducibility.
• Quantified changes in reproducibility as a function of group size.
• Assessed the impact of different data analysis models on reproducibility metrics.

Conclusions:

• The proposed image-wide pattern similarity approach offers a robust method for assessing neuroimaging reproducibility.
• Statistical resampling of correlation coefficients provides valuable insights into the reliability of functional activation data.
• Understanding factors influencing reproducibility is essential for accurate interpretation of brain imaging studies.