Detection and visualization method of dynamic state transition for biological spatio-temporal imaging data.

Area of Science:

• Neuroscience
• Biophysics
• Statistical Analysis

Background:

• Functional brain imaging analysis commonly uses regression and cross-correlation methods.
• Existing techniques can only detect activation signals matching a predefined reference function.

Purpose of the Study:

• To propose a novel fusion method for detecting brain neural activation.
• To extract activation as a phase transition of system dynamics without external reference information.

Main Methods:

• Applied an innovation approach in time series analysis combined with statistical tests.
• Fitted Autoregressive (AR) models to time series data before or after state transitions.
• Filtered remaining time series data using AR parameters to obtain innovation (filter output).

Main Results:

• Successfully extracted brain neural activation as a phase transition of dynamics.
• Detected activation through temporal transitions of statistical test values.
• Precisely detected spatio-temporal activation profiles in mammalian brain and cardiac sino-atrial node (SAN) optical imaging data.

Conclusions:

• The proposed method effectively detects spatio-temporal activation profiles.
• This approach offers a new way to analyze functional imaging data by identifying dynamic system changes.
• Validated for use in both brain and cardiac imaging applications.