Test-Retest Reliability in Functional MRI Studies

Test–retest reliability in functional MRI (fMRI) studies denotes the degree to which brain activation patterns remain consistent when the same individuals undergo identical scanning protocols on different occasions. It is fundamental for longitudinal research, biomarker development and clinical translation, as unstable measurements can undermine conclusions about treatment effects, developmental trajectories or individual differences. Reliability is influenced by scanner hardware, pulse sequence design, subject physiology, task paradigm and data-processing pipelines. The blood-oxygen-level dependent (BOLD) signal, which underpins most fMRI analyses, is sensitive to haemodynamic fluctuations and non-neuronal artefacts. Metrics such as the intraclass correlation coefficient (ICC) quantify the stability of regional activation, guiding optimisation of acquisition parameters and statistical models. Despite methodological advances—including refined preprocessing, adaptive modelling of the haemodynamic response and correction for physiological noise—reliability varies markedly across brain regions and tasks. Addressing these variations is critical for ensuring robust reproducibility, enabling reliable tracking of disease progression, cognitive training effects and individual neurofunctional profiles in both research and clinical settings.

Research from Nature Portfolio

Studies have highlighted anatomical and physiological confounds that compromise fMRI reliability. One seminal investigation demonstrated that signal changes attributed to amygdala activation are often driven by adjacent venous structures, prompting the adoption of higher temporal-resolution sequences and susceptibility-weighted imaging to disentangle vascular from neural sources. Another contribution introduced threshold-weighted overlap maps as a visual and quantitative tool for assessing spatial consistency across subjects, enabling researchers to identify regions of low reproducibility and refine task designs accordingly. Together, these approaches have advanced understanding of the sources of unreliability and offered practical methods to enhance the robustness of fMRI measurements.

Test-Retest Reliability in Functional MRI Studies publication trend

The graph below shows the total number of articles in test-retest reliability in functional mri studies across all publications each year (not limited to Nature Index journals).

Technical terms

Blood-oxygen-level dependent (BOLD) signal: the MRI contrast mechanism that reflects changes in local blood oxygenation related to neural activity.

Intraclass correlation coefficient (ICC): a statistical index ranging from 0 to 1 that quantifies the consistency of measurements across repeated sessions for the same subjects.

Haemodynamic response function (HRF): a model of the time-course of the BOLD signal following a brief burst of neural activity.

Physiological noise: fluctuations in the fMRI signal arising from cardiac pulsation, respiration and other non-neuronal processes.

Intersubject variability: differences in brain activation patterns across individuals, reflecting anatomical, physiological or cognitive strategy diversity.

References

1. fMRI measurements of amygdala activation are confounded by stimulus correlated signal fluctuation in nearby veins draining distant brain regions. Scientific Reports (2015).
2. Visualising inter-subject variability in fMRI using threshold-weighted overlap maps. Scientific Reports (2016).
3. Model Specification and the Reliability of fMRI Results: Implications for Longitudinal Neuroimaging Studies in Psychiatry. PLOS ONE (2014).
4. Unreliability of putative fMRI biomarkers during emotional face processing. NeuroImage (2017).
5. Understanding the contribution of neural and physiological signal variation to the low repeatability of emotion-induced BOLD responses. NeuroImage (2013).