Wavelet entropy of BOLD time series: an application to Rolandic epilepsy

Lalit Gupta, Jacobus F.A. Jansen, Paul A.M. Hofman, René M.H. Besseling, Anton J.A. de Louw, Albert P. Aldenkamp, Walter H. Backes

Research output: Contribution to journalArticleAcademicpeer-review

10 Citations (Scopus)
191 Downloads (Pure)


Purpose: To assess the wavelet entropy for the characterization of intrinsic aberrant temporal irregularities in the time series of resting-state blood-oxygen-level-dependent (BOLD) signal fluctuations. Further, to evaluate the temporal irregularities (disorder/order) on a voxel-by-voxel basis in the brains of children with Rolandic epilepsy. Materials and Methods: The BOLD time series was decomposed using the discrete wavelet transform and the wavelet entropy was calculated. Using a model time series consisting of multiple harmonics and nonstationary components, the wavelet entropy was compared with Shannon and spectral (Fourier-based) entropy. As an application, the wavelet entropy in 22 children with Rolandic epilepsy was compared to 22 age-matched healthy controls. The images were obtained by performing resting-state functional magnetic resonance imaging (fMRI) using a 3T system, an 8-element receive-only head coil, and an echo planar imaging pulse sequence (T2 * -weighted). The wavelet entropy was also compared to spectral entropy, regional homogeneity, and Shannon entropy. Results: Wavelet entropy was found to identify the nonstationary components of the model time series. In Rolandic epilepsy patients, a significantly elevated wavelet entropy was observed relative to controls for the whole cerebrum (P = 0.03). Spectral entropy (P = 0.41), regional homogeneity (P = 0.52), and Shannon entropy (P = 0.32) did not reveal significant differences. Conclusion: The wavelet entropy measure appeared more sensitive to detect abnormalities in cerebral fluctuations represented by nonstationary effects in the BOLD time series than more conventional measures. This effect was observed in the model time series as well as in Rolandic epilepsy. These observations suggest that the brains of children with Rolandic epilepsy exhibit stronger nonstationary temporal signal fluctuations than controls. Level of Evidence: 2. Technical Efficacy: Stage 3. J. Magn. Reson. Imaging 2017;46:1728–1737.

Original languageEnglish
Pages (from-to)1728-1737
Number of pages10
JournalJournal of Magnetic Resonance Imaging
Issue number6
Publication statusPublished - 1 Dec 2017


  • BOLD time series
  • discrete wavelet transform
  • frequency structure
  • Rolandic epilepsy
  • wavelet entropy
  • Epilepsy, Rolandic/diagnostic imaging
  • Brain Mapping/methods
  • Humans
  • Magnetic Resonance Imaging/methods
  • Entropy
  • Brain/diagnostic imaging
  • Female
  • Child


Dive into the research topics of 'Wavelet entropy of BOLD time series: an application to Rolandic epilepsy'. Together they form a unique fingerprint.

Cite this