Exploratory data analysis of evoked response single trials based on minimal spanning tree

Area of Science:

• Neuroscience
• Data Analysis
• Signal Processing

Background:

• Electrophysiological signals, particularly single trial (ST) data, present challenges in analysis due to inherent noise and variability.
• Understanding evoked responses requires methods that can handle the complexity and trial-to-trial dynamics of neural data.

Purpose of the Study:

• To introduce a novel exploratory data analysis framework for single trial (ST) electrophysiological signals.
• To support the compact description of ST samples using content-dependent ordered lists.
• To enable efficient methods for increasing signal-to-noise ratio (SNR), extracting prototypical responses, visualizing self-organization trends, and tracking evoked responses across trials.

Main Methods:

• The framework utilizes a minimal spanning tree approach for data analysis.
• Magnetoencephalographic (MEG) auditory evoked responses were employed for framework demonstration and validation.
• The method involves creating content-dependent ordered lists to represent ST samples.

Main Results:

• The framework successfully demonstrated benefits in understanding and enhancing evoked signals.
• Evidence supporting the stimulus-induced phase-resetting hypothesis in the 3-20 Hz band was found.
• The analysis identified trials lacking prototypical evoked responses and revealed ordering across trials, suggesting long time-scale underlying processes.

Conclusions:

• The proposed minimal spanning tree framework offers an intelligent approach to manipulating and analyzing ST electrophysiological data.
• The findings provide insights into neural response dynamics, including phase resetting and the presence of non-responsive trials.
• The framework facilitates the discovery of underlying temporal dynamics in neural responses.