Bayesian longitudinal spectral estimation with application to resting-state fMRI data analysis

Research output: Contribution to journalArticlepeer-review

1 Scopus citations

Abstract

The amplitude of the oscillatory patterns present in spontaneous fluctuations of brain signals obtained from resting-state functional magnetic resonance imaging (fMRI), measured using an index called the fractional amplitude of low-frequency fluctuation (fALFF), is a well-known measure of brain activity with potential to serve as a marker for brain dysfunction. With the rise of longitudinal neuroimaging studies, there is a great need for methodologies that take advantage of the longitudinal design in modeling the impact of aging or disease progression. Motivated by the longitudinal design of the Alzheimer's Disease Neuroimaging Initiative (ADNI), a novel Bayesian longitudinal model is developed in order to estimate the spectra of resting-state fMRI time courses, from which one can extract estimates of fALFF that are potentially associated with aging. The model incorporates within-subject correlation to improve estimates of the spectra, in addition to the variability that naturally arises between subjects. The model is validated using simulated data to show the gains in performance for estimating fALFF by taking advantage of the longitudinal design. Finally, a longitudinal analysis on fALFF from the resting-state fMRI data from ADNI is conducted, where the impact of both Alzheimer's disease and aging on the spontaneous fluctuations of brain activity is shown.

Original languageEnglish (US)
JournalEconometrics and Statistics
DOIs
StatePublished - Jan 1 2019

Keywords

  • Alzheimer's disease
  • Amplitude of low-frequency fluctuation
  • Longitudinal data analysis
  • Markov chain Monte Carlo
  • Resting-state fMRI
  • Spectral estimation

Fingerprint Dive into the research topics of 'Bayesian longitudinal spectral estimation with application to resting-state fMRI data analysis'. Together they form a unique fingerprint.

Cite this