Correcting motion induced fluorescence artifacts in two-channel neural imaging.

Area of Science:

• Neuroscience
• Biophysics
• Computational Biology

Background:

• Motion during animal behavior causes artifacts in fluorescence imaging, obscuring neural signals.
• Simultaneous two-channel imaging (activity-dependent and independent fluorophores) is used to correct motion artifacts.
• Existing ratio-based methods fail to account for channel-independent noise.

Purpose of the Study:

• To develop and validate a novel method, Two-channel Motion Artifact Correction (TMAC), for accurate removal of motion artifacts in fluorescence-based neural recordings.
• To establish a robust metric for evaluating motion correction algorithm performance using behavioral decoding.
• To compare TMAC against existing artifact correction techniques.

Main Methods:

• Developed TMAC, a generative model incorporating motion, neural activity, and noise for two-channel fluorescence data.
• Employed Bayesian inference to estimate latent neural activity and reduce motion artifacts.
• Created a novel ground-truth evaluation method based on comparing behavioral decoding from dual-fluorophore versus control recordings.

Main Results:

• TMAC significantly reduces motion artifacts in fluorescence traces.
• Behavioral decoding (locomotion) from TMAC-corrected GCaMP recordings was 20x more accurate than control.
• TMAC outperformed all tested motion correction methods, with the best alternative achieving only ~8x improvement.

Conclusions:

• TMAC provides a superior approach for motion artifact correction in behaving animal neural imaging.
• The developed evaluation metric offers a reliable benchmark for assessing motion correction algorithm efficacy.
• Accurate motion artifact removal is crucial for reliable neural activity analysis and behavioral decoding.