Magnetic Resonance Imaging of Human Brain Anatomy and Function

National Institute Of Neurological Disorders And Stroke

Investigators

Linked publications, trials & patents

Abstract

Previous work in the lab has shown that signal variability in fMRI strongly depends on arousal state, and that both neuronal and autonomic activity contribute to this variability. To comprehensively investigate underlying mechanisms, we have been performing overnight fMRI-EEG studies across the sleep-wake cycle, while collecting autonomic indicator signals. After completing our pilot study on 16 subjects, we started a main study under which we now have collected two entire nights of data on 15 subjects. Initial data analysis, completed in previous years, showed a strong correlation between peripheral vascular tone (measured from the finger skin with photoplethysmography) and fMRI global signal (GS), in particular during light (N1 and N2) sleep (and REM), but not N3. We attributed this to sympathetic constriction of both cerebral and peripheral vasculature. During light sleep, these constrictions appeared precipitated by EEG complexes. We developed an explanatory model postulating that joint fMRI and autonomic effects may be caused by autonomic arousals that involve sympathetic activation. Outside N3 sleep, we also found strong correlations between fMRI GS and CSF flow through the cerebral aqueduct. CSF flow may have relevance for brain waste clearance through the glymphatic system as it improves CSF solute mixing and possible a net transport out of the brain. We developed an explanatory model confirming an earlier report that this CSF flow may be generated by the cerebral blood volume changes associated with the fMRI signal. This new type of CSF flow co-occurs with previously reported pulsations associated with cardiac and respiratory cycles but may be of larger amplitude; it therefore may have an unique role in glymphatic clearance. We are currently developing methodology to accurately quantify CSF flow volumes associated with the individual mechanisms. In parallel, we have explored various approaches to separate autonomic and neuronal contributions to the fMRI signal. A preliminary analysis pipeline was developed by which various confounds are regressed out of the fMRI signal in a time-segmented fashion. Removed confounds include cardiac and respiratory cycles, motion, respiratory flow rate, and peripheral vascular tone. To distinguish between the effects of sympathetic cerebral vasoconstriction and CO2-mediated vasoaction secondary to respiratory flow rate changes, dedicated fMRI experiments were performed. Sympathetic changes were elicited by a mental task, which variably co-occurred with respiration changes. Distinct effects of sympathetic effects on the fMRI signal were demonstrated, independently of respiration-related changes. This confirmed the notion that removing of autonomic confounds from the fMRI signal requires consideration of both sympathetic effects and CO2-mediated effects. Consideration of arousal state is also important for the interpretation of fMRI studies of behavior. The dependence of the fMRI signal on arousal state offers opportunity to infer arousal state without the need for EEG, which greatly simplifies the experiment. Previously we demonstrated, in primate, a template correlation method to estimate arousal state from the fMRI signal. More recently, we extended this method to human. After training the method with EEG/fMRI data to derive a generalized template, we have been able to predict reaction time in a button press task. This study was recently published in eLife. It is expected that the template can be used more broadly to account for arousal in fMRI studies of various types of behavior.