Model-based decoupling of evoked and spontaneous neural activity in calcium imaging data

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Neural activity patterns are influenced by both external stimuli and internal spontaneous activity.
• Calcium imaging data presents challenges for analysis due to the slow dynamics of calcium indicators.
• Differentiating stimulus-evoked from spontaneous neural activity is crucial for accurate interpretation.

Purpose of the Study:

• To develop a statistical model for decoupling stimulus-driven neural activity from low-dimensional spontaneous activity.
• To identify hidden factors contributing to spontaneous activity and their confounding effects on stimulus tuning.
• To quantitatively measure the contributions of evoked activity, spontaneous activity, and their interaction to neural responses.

Main Methods:

• Development of a novel statistical model.
• Joint estimation of stimulus tuning properties and hidden factors of spontaneous activity.
• Application of the model to calcium imaging data from zebrafish optic tectum and mouse visual cortex.

Main Results:

• The model successfully decouples stimulus-driven and spontaneous neural activity.
• Quantitative measurements reveal the distinct and interactive roles of evoked and spontaneous activity in neural responses.
• The approach accounts for confounding effects of spontaneous activity on stimulus tuning.

Conclusions:

• The presented statistical model offers a broadly applicable method for analyzing neural population activity.
• By preserving information within spontaneous activity, the model provides novel insights into single-trial neural dynamics.
• This work enhances the understanding of how neural circuits process information in vivo.