Accurate Estimation of Particle Dynamics Bypassing Substrate Drift Bias: Application to Cell Nucleus Motion

In microscopic imaging, the movement of a living substrate can be caused by its own displacement (e.g., cell motion/migration) or other technical factors such as microscope stage drift. This drifting motion is one of the main biases resulting in poor estimation of particle dynamics since it seriously affects the estimation of the biophysical parameters (the diffusion constant D and anomalous exponent α), especially when performed on the basis of mean squared displacement (MSD) analysis. In this paper, we compare a few substrate drift correction/registration methods based on the use of additional fluorescent spots (landmarks). In the particular case of cell nucleus motion, we labeled telomeres spreading throughout the cell nucleus. We show that compared to the MSD analysis, the use of Gaussian processes is an effective and more accurate way to estimate the substrate drift, and major biophysical parameters of particle dynamics.