A wide variety of metrics are available to summarize resting state f-MRI studies. This course describes basic and advanced metrics ranging from seed-based correlation to graph theory and dynamic analysis. Study considerations that impact the choice and interpretation of different metrics will be discussed.
Resting-State fMRI: Which Index is Really Useful?
TARGET AUDIENCE: Clinicians and researchers interested in the analysis and interpretation of resting-state fMRI and functional connectivity studies.
OUTCOME/OBJECTIVES:
· Understand different metrics used to assess resting state f-MRI
· Evaluate the appropriateness of a particular metric for a particular study
· Identify data characteristics that influence different metrics
OVERVIEW: The immense amount of data obtained in resting state f-MRI studies makes it essential to identify metrics that can summarize the features of interest. The number and type of metrics used to summarize resting state f-MRI data have proliferated over the years, and it can be difficult to decide which metric to use for a given study. This discussion will cover widely used analysis techniques as well as more advanced metrics, including seed-based correlation, independent component analysis, graph theory metrics, and dynamic connectivity. Each will be considered in the context of common neuroimaging scenarios:
- The type of study that is being performed. Whole brain or between selected areas? Intra-individual, inter-individual, or across groups? Dynamic or static?
- The information that is needed. Spatial extent of particular networks? Strength of connectivity between regions? Variations in connectivity over time or across groups? More complex measures of network properties?
- Factors that could influence the selected metric. Amplitude of low frequency oscillations, noise from physiological processes or motion, differences that affect registration, thresholds, window length, etc.
- Considerations for statistical analysis. Appropriate comparisons across groups, construction of appropriate null data, multiple comparisons.
CONCLUSION: No single metric can be optimal for all studies. This course aims to describe the strengths and weaknesses of different metrics used to analyze resting state f-MRI and the factors that must be considered when interpreting the metrics. The goal is to motivate careful choice of evaluation metrics and healthy skepticism when it comes to interpretation.
Network modelling methods for FMRI. Smith SM, Miller KL, Salimi-Khorshidi G, Webster M, Beckmann CF, Nichols TE, Ramsey JD, Woolrich MW. Neuroimage. 2011 Jan 15;54(2):875-91. doi: 10.1016/j.neuroimage.2010.08.063. PMID: 20817103
Test-retest reliability of resting-state connectivity network characteristics using fMRI and graph theoretical measures. Braun U, Plichta MM, Esslinger C, Sauer C, Haddad L, Grimm O, Mier D, Mohnke S, Heinz A, Erk S, Walter H, Seiferth N, Kirsch P, Meyer-Lindenberg A. Neuroimage. 2012 Jan 16;59(2):1404-12. doi: 10.1016/j.neuroimage.2011.08.044. PMID: 21888983
Can sliding-window correlations reveal dynamic functional connectivity in resting-state fMRI? Hindriks R, Adhikari MH, Murayama Y, Ganzetti M, Mantini D, Logothetis NK, Deco G. Neuroimage. 2016 Feb 15;127:242-56. doi: 10.1016/j.neuroimage.2015.11.055. PMID: 26631813
Evaluation of sliding window correlation performance for characterizing dynamic functional connectivity and brain states. Shakil S, Lee CH, Keilholz SD. Neuroimage. 2016 Jun;133:111-28. doi: 10.1016/j.neuroimage.2016.02.074. PMID: 2695219