This study investigates the impact of the number of brain parcels (or scale) on the results of a mass univariate general linear model (GLM) in connectome analysis using functional magnetic resonance imaging (fMRI). The brain parcels used as nodes in the connectome analysis were functionally defined by a group cluster analysis. The researchers first validate that a classic Benjamini-Hochberg procedure with parametric GLM tests appropriately controls the false-discovery rate (FDR) at a given scale. They then observe through realistic simulations that there is no substantial inflation of the FDR across scales, as long as it is controlled independently within each scale. Furthermore, both simulations and analyses on three real resting-state fMRI datasets (schizophrenia, congenital blindness, motor practice) reveal that the rate of discovery varies significantly depending on scales, with relatively higher rates for low scales below 25. Despite these differences, the statistical maps derived at different scales show consistent results across all three datasets. Some specific effects are better observed around scale 50, highlighting the potential benefits of multiscale analysis. The statistical maps obtained from real data align well with existing literature, supporting the validity of multiscale GLM connectome analysis with FDR in capturing biologically meaningful effects in various experimental conditions. Overall, this study provides valuable insights into how altering brain parcel scale impacts connectome analysis results and highlights the importance of controlling FDR independently within each scale to identify significant associations in fMRI studies across different scales and experimental conditions.
In connectome analysis using fMRI, the number of brain parcels (or scale) affects the results. Brain parcels are groups of brain regions that are defined based on how they function together. A classic statistical test called GLM can control for false discoveries at a given scale. Simulations show that controlling for false discoveries independently within each scale is important. The rate of discovering meaningful effects varies depending on the scale used."
Definitions
- Brain parcels: Groups of brain regions that work together.
- Connectome analysis: Studying how different parts of the brain are connected and communicate with each other.
- fMRI: A type of imaging technology that shows brain activity by measuring blood flow changes.
- GLM: General Linear Model, a statistical method used to analyze data and test hypotheses.
- False discoveries: Incorrect findings or associations made in a study.
- Scale: The number or size of brain parcels used in the analysis.
Exploring the Impact of Brain Parcel Scale on Connectome Analysis Results
Connectome analysis is a powerful tool for studying functional magnetic resonance imaging (fMRI) data. It involves mapping the connections between different brain regions, or “parcels”, to better understand how they interact and affect behavior. However, it can be difficult to determine what scale of parcels should be used in connectome analysis.
Methodology
The researchers first validated that a classic Benjamini-Hochberg procedure with parametric GLM tests appropriately controls the false-discovery rate (FDR) at a given scale. They then observed through realistic simulations that there was no substantial inflation of the FDR across scales as long as it was controlled independently within each scale. Furthermore, both simulations and analyses on three real resting-state fMRI datasets (schizophrenia, congenital blindness, motor practice) revealed that the rate of discovery varied significantly depending on scales, with relatively higher rates for low scales below 25.
Results
Despite these differences in discovery rates across scales, the statistical maps derived at different scales showed consistent results across all three datasets. Some specific effects were better observed around scale 50, highlighting the potential benefits of multiscale analysis. The statistical maps obtained from real data aligned well with existing literature supporting its validity in capturing biologically meaningful effects in various experimental conditions.
Conclusion
Overall, this study provides valuable insights into how altering brain parcel scale impacts connectome analysis results and highlights the importance of controlling FDR independently within each scale to identify significant associations in fMRI studies across different scales and experimental conditions.
Alterations in brain connectivity have been associated with a variety of clinical disorders using functional magnetic resonance imaging (fMRI). We investigated empirically how the number of brain parcels (or scale) impacted the results of a mass univariate general linear model (GLM) on connectomes. The brain parcels used as nodes in the connectome analysis were functionnally defined by a group cluster analysis. We first validated that a classic Benjamini-Hochberg procedure with parametric GLM tests did control appropriately the false-discovery rate (FDR) at a given scale. We then observed on realistic simulations that there was no substantial inflation of the FDR across scales, as long as the FDR was controlled independently within each scale, and the presence of true associations could be established using an omnibus permutation test combining all scales. Second, we observed both on simulations and on three real resting-state fMRI datasets (schizophrenia, congenital blindness, motor practice) that the rate of discovery varied markedly as a function of scales, and was relatively higher for low scales, below 25. Despite the differences in discovery rate, the statistical maps derived at different scales were generally very consistent in the three real datasets. Some seeds still showed effects better observed around 50, illustrating the potential benefits of multiscale analysis. On real data, the statistical maps agreed well with the existing literature. Overall, our results support that the multiscale GLM connectome analysis with FDR is statistically valid and can capture biologically meaningful effects in a variety of experimental conditions.
