Detecting chaotic structures in noisy pulse trains based on interspike interval reconstruction

Area of Science:

• Neuroscience
• Computational Neuroscience
• Signal Processing

Background:

• Accurate analysis of neural firing patterns is crucial for understanding brain function.
• Noisy pulse trains present challenges in distinguishing between periodic and chaotic dynamics.
• Interspike interval (ISI) sequences are key to characterizing neural firing patterns.

Purpose of the Study:

• To evaluate a nonlinear prediction method for detecting deterministic structures in noisy pulse trains.
• To improve the discrimination between noisy periodic and chaotic pulse trains.
• To assess the impact of noise-induced pulses on nonlinear prediction accuracy.

Main Methods:

• Application of a nonlinear prediction method based on interspike interval (ISI) reconstruction.
• Analysis of ISI sequences from noisy pulse trains.
• Utilizing a genetic algorithm for grouping ISI sequences to eliminate noise-induced pulses.

Main Results:

• The nonlinear prediction method initially failed to discriminate between noisy periodic and chaotic pulse trains when noise-induced pulses were present.
• Eliminating noise-induced pulses by grouping ISI sequences with a genetic algorithm clarified the chaotic structure.
• The improved method successfully discriminated noisy chaotic pulse trains from periodic ones.

Conclusions:

• Standard nonlinear prediction methods are insufficient for analyzing noisy pulse trains with extraneous pulses.
• Genetic algorithm-based grouping of ISI sequences is an effective preprocessing step for chaotic dynamics detection.
• This approach enhances the ability to identify deterministic chaos in biological neural signals.