Test-retest reliability of fMRI brain activity during memory encoding

The mechanisms underlying hemispheric specialization of memory are not completely understood. Functional magnetic resonance imaging (fMRI) can be used to develop and test models of hemispheric specialization. In particular for memory tasks however, the interpretation of fMRI results is often hampered by the low reliability of the data. In the present study we therefore analyzed the test-retest reliability of fMRI brain activation related to an implicit memory encoding task, with a particular focus on brain activity of the medial temporal lobe (MTL). Fifteen healthy subjects were scanned with fMRI on two sessions (average retest interval 35 days) using a commonly applied novelty encoding paradigm contrasting known and unknown stimuli. To assess brain lateralization, we used three different stimuli classes that differed in their verbalizability (words, scenes, fractals). Test-retest reliability of fMRI brain activation was assessed by an intraclass-correlation coefficient (ICC), describing the stability of inter-individual differences in the brain activation magnitude over time. We found as expected a left-lateralized brain activation network for the words paradigm, a bilateral network for the scenes paradigm, and predominantly right-hemispheric brain activation for the fractals paradigm. Although these networks were consistently activated in both sessions on the group level, across-subject reliabilities were only poor to fair (ICCs ≤ 0.45). Overall, the highest ICC values were obtained for the scenes paradigm, but only in strongly activated brain regions. In particular the reliability of brain activity of the MTL was poor for all paradigms. In conclusion, for novelty encoding paradigms the interpretation of fMRI results on a single subject level is hampered by its low reliability. More studies are needed to optimize the retest reliability of fMRI activation for memory tasks.