Nonlinear biomedical signal processing

Many biomedical signals have properties that are not reflected in their amplitude distribution or spectral properties. Examples include the chaotic oscillations of pupil size (usually becoming more unpredictable under stress or pain), EEG (e.g., becoming more and less chaotic during mental activity and either sleep or pathological deficit of consciousness, respectively) and EMG (with the level of regularity depending on muscle fatigue). Nonlinear processing can extract some of the information that linear analysis cannot retrieve. Some results were obtained in the following applications.

  1. Introduction of a new method for the estimation of entropy from biomedical signals.
  2. Characterization of the pupillogram recorded from healthy subjects under weak stress stimulations and from patients with temporomandibular disorders and with obstructive sleep apnea syndrome (which are affected by a dysregulation of the autonomous nervous system, which controls pupil dynamics).
  3. Analysis of morphology and complexity of EEG of patients with different pathologies: coma, different deficits of consciousness, epilepsy and encephalitis.
  4. Analysis of connectivity of a population of neurons in vitro and of the complexity of its spontaneous activity during maturation.
  5. Study of the fractal dimension of EMG during fatiguing contractions, which correlates with the manifestations of central fatigue, like motor unit synchronism and firing rate modulation.

ERC Sector:

  • PE7_7 Signal processing


  • Biomedical signal processing
  • Nonlinear signal processin

Research groups