Decontaminate Traces From Fluorescence Calcium Imaging Videos Using Targeted Non-negative Matrix Factorization

Area of Science:

• Neuroscience
• Biophysics
• Computational Biology

Background:

• Fluorescence microscopy and calcium indicators are crucial for studying brain function.
• Analyzing large-scale in vivo imaging data requires accurate extraction of neuronal activity.
• Background noise and signals from adjacent neurons contaminate calcium traces.

Purpose of the Study:

• To develop a novel method for accurate signal unmixing in calcium imaging.
• To improve the analysis of neuronal activity from large-scale in vivo recordings.
• To address challenges posed by non-specific calcium sources and background fluctuations.

Main Methods:

• Developed temporal unmixing of calcium traces (TUnCaT) algorithm.
• Employed background subtraction to eliminate false transients from background.
• Utilized targeted non-negative matrix factorization to resolve overlapping neuronal signals.

Main Results:

• TUnCaT demonstrated superior accuracy compared to existing algorithms on experimental and simulated datasets.
• The algorithm effectively removed false transients from background and neighboring neurons.
• TUnCaT achieved processing speeds faster than or comparable to existing methods.

Conclusions:

• TUnCaT provides a robust solution for accurate calcium signal unmixing.
• The method enhances the reliability of neuronal activity analysis in large-scale brain imaging.
• TUnCaT facilitates more precise understanding of neural dynamics in vivo.