A statistical framework for analysis of trial-level temporal dynamics in fiber photometry experiments

Area of Science:

• Neuroscience
• Computational Neuroscience
• Biostatistics

Background:

• Fiber photometry is widely used for in vivo neural activity measurement.
• Current analysis methods often reduce signal detection by averaging across trials, losing valuable trial-level information.

Purpose of the Study:

• To develop a novel statistical framework for fiber photometry data analysis.
• To enable hypothesis testing at every trial time-point without averaging trial-level signals.
• To improve the detection of event-related signal changes and compare signal timing and magnitude across conditions.

Main Methods:

• Functional linear mixed modeling applied to fiber photometry data.
• Utilizing trial-level signals and exploiting signal autocorrelation for joint confidence intervals.
• Reanalysis of existing mesolimbic dopamine reward learning data and simulation experiments.

Main Results:

• The framework allows hypothesis testing at each trial time-point, accounting for between-animal variability.
• Identified two distinct dopamine components with unique temporal dynamics in response to reward delivery.
• Demonstrated improved statistical power compared to common analysis approaches in simulations.

Conclusions:

• The proposed framework offers a more sensitive approach to analyzing fiber photometry data.
• It enables a more nuanced understanding of neural dynamics, particularly in reward learning.
• An open-source package and guide are provided for practical application.