Related Experiment Video
Updated: May 20, 2025
00:09
Fluorescent Calcium Imaging and Subsequent In Situ Hybridization for Neuronal Precursor Characterization in Xenopus laevis
Published on: February 18, 2020
7.9K
Precise calcium-to-spike inference using biophysical generative models
Gerard Joey Broussard1, Giovanni Diana2, Francisco J Urra Quiroz2
1Neuroscience Institute, Washington Road, Princeton University, Princeton, New Jersey USA 08540.
Research Square
|May 2, 2025
Summary
New algorithms improve neural spike time accuracy. Biophysical models for GCaMP indicators reduce false positives, achieving near-theoretical limits for precise action potential detection.
Area of Science:
- Neuroscience
- Biophysics
- Biochemistry
Background:
- Intramolecular dynamics of neural activity indicators can affect action potential (spike) time estimation.
- Existing inference algorithms may produce inaccuracies due to indicator kinetics.
Purpose of the Study:
- To develop a more accurate spike inference algorithm.
- To characterize the kinetic responses of GCaMP6f, jGCaMP7f, and jGCaMP8f indicators to calcium.
Main Methods:
- In vitro stopped-flow and brain slice recordings were used to analyze indicator kinetics.
- A multistate model of GCaMP was developed based on experimental data.
- Bayesian Sequential Monte Carlo (BiophysSMC) and machine learning (BiophysML) inference methods were created.
Main Results:
- jGCaMP8f exhibited use-dependent slowing, leading to false positives with current methods.
- The new biophysical inference methods significantly reduced false positives.
- Spike time accuracy improved, with a median uncertainty of 4 milliseconds, reaching theoretical limits.
Conclusions:
- Model-based approaches offer advantages over model-free algorithms for spike inference.
- The developed biophysical methods provide a substantial improvement in action potential detection accuracy.
- This framework enhances the precise estimation of neural activity timing.
