Enhancing Fluorescence Correlation Spectroscopy with Machine Learning for Advanced Analysis of Anomalous Diffusion

Nathan Quiblier,Jan-Michael Rye,Pierre Leclerc,Henri Truong,Abdelkrim Hannou,Laurent Héliot,Hugues Berry

2024-07-17

Abstract:The random motion of molecules in living cells has consistently been reported to deviate from standard Brownian motion, a behavior coined as ``anomalous diffusion''. Fluorescence Correlation Spectroscopy (FCS) is a powerful method to quantify molecular motions in living cells but its application is limited to a subset of random motions and to long acquisition times. Here, we propose a new analysis approach that frees FCS of these limitations by using machine learning to infer the underlying model of motion and estimate the motion parameters. Using simulated FCS recordings, we show that this approach enlarges the range of anomalous motions available in FCS. We further validate our approach via experimental FCS recordings of calibrated fluorescent beads in increasing concentrations of glycerol in water. Taken together, our approach significantly augments the analysis power of FCS to capacities that are similar to the best-in-class state-of-the-art algorithms for single-particle-tracking experiments.

Quantitative Methods,Biological Physics

What problem does this paper attempt to address?

### Problems the paper attempts to solve

This paper aims to address the limitations of Fluorescence Correlation Spectroscopy (FCS) in analyzing abnormal molecular diffusion in living cells. Specifically, FCS has the following deficiencies:

1. **Limited scope of application**: Traditional FCS methods can only handle a part of random motion and require a long acquisition time.
2. **Challenges in model classification and parameter estimation**: FCS has difficulty in distinguishing standard Brownian Motion (BM) from other types of abnormal diffusion such as fractional Brownian Motion (fBM) and Continuous - Time Random Walk (CTRW).
3. **Low signal - to - noise ratio**: The autocorrelation function signal - to - noise ratio of FCS is usually low, requiring long - term continuous monitoring and averaging a large number of continuous measurements, which limits its ability to track changes in motion parameters in a short time.

### Solutions

To overcome the above problems, the authors propose a new method based on machine learning to perform model classification and parameter estimation tasks using the autocorrelation function. The specific contributions are as follows:

1. **Expand the application range of FCS**: The new method can handle a wider range of abnormal diffusion types, including BM, fBM, and CTRW.
2. **Improve the accuracy of classification and parameter estimation**: Through machine - learning algorithms, high - precision model classification and parameter estimation can be achieved even in short - time and low - signal - to - noise - ratio situations.
3. **Adapt to multiple experimental settings**: This method is robust to different beam widths and light intensities and is suitable for a variety of experimental conditions.

### Experimental verification

The authors verified the effectiveness of this method through simulated data and experimental data:

- **Simulated data**: More than 2.5 million simulated FCS experimental data were generated, covering different motion parameters and experimental settings. The results show that the F1 score of this method in the classification task reaches 0.88 - 0.90, and the mean absolute error (MAE) in the parameter estimation task is also very low.
- **Experimental data**: FCS measurements were carried out using calibrated fluorescent microspheres in glycerol aqueous solutions of different concentrations. The results indicate that this method can accurately classify motion models and estimate the diffusion coefficient, which is consistent with the theoretical value predicted by the Stokes - Einstein law.

### Conclusion

This study shows how to use machine learning to enhance the ability of FCS to analyze abnormal molecular diffusion in living cells, making its performance close to that of the current state - of - the - art Single - Particle Tracking (SPT) algorithms. This method not only expands the application range of FCS but also improves its classification and parameter estimation capabilities in short - time and low - signal - to - noise - ratio conditions.