Multi-scale detection of rate changes in spike trains with weak dependencies

Area of Science:

• Computational Neuroscience
• Statistical Signal Processing

Background:

• Neuronal spike train analysis often assumes independent inter-spike intervals (ISIs), which is frequently violated in empirical data.
• Variability in firing rates and dependencies in ISIs pose challenges for change point detection.

Purpose of the Study:

• To extend the Multiple Filter Test (MFT) for analyzing point processes with short-range dependent ISIs.
• To improve the statistical analysis of neuronal spike trains by accounting for ISI correlations.

Main Methods:

• Developed an extended Multiple Filter Test (MFT) that estimates serial dependencies within the test statistic.
• Applied the new MFT to empirical spike train datasets exhibiting positive and negative ISI correlations.

Main Results:

• The enhanced MFT demonstrates applicability to diverse firing patterns, including tonic and bursty firing.
• Simulations show improved performance over methods assuming ISI independence, particularly for positively correlated ISIs where it reduces false positives.
• For negatively correlated ISIs, the new MFT enhances change point detection probability and signal extraction from noisy spike trains.

Conclusions:

• The extended MFT provides a robust statistical framework for analyzing neuronal spike trains with dependent ISIs.
• This method is crucial for accurate change point estimation and reliable signal extraction in realistic neurophysiological recordings.