Optimal Drift Correction for Superresolution Localization Microscopy with Bayesian Inference.

Area of Science:

• Biophysics
• Microscopy techniques
• Computational biology

Background:

• Superresolution microscopy, specifically single-molecule localization microscopy (SMLM), demands precise sample drift correction for high-resolution imaging.
• Current drift correction methods, such as fiducial markers or image correlation, have limitations including increased experimental complexity or reduced temporal resolution.

Purpose of the Study:

• To develop and present a novel, highly accurate method for sample drift correction in SMLM.
• To leverage a Bayesian statistical framework for direct drift calculation from single-molecule data.

Main Methods:

• A new algorithm based on Bayesian inference was developed to estimate sample drift.
• The method directly utilizes single-molecule coordinates from SMLM data to calculate drift for each image frame.
• Theoretical underpinnings and a practical implementation of the algorithm are presented.

Main Results:

• The proposed Bayesian approach enables drift calculation at the temporal resolution of individual image frames.
• The implemented algorithm demonstrates significantly higher accuracy in drift estimation compared to traditional image-correlation methods.
• The technique directly uses localization data, avoiding the need for additional fiducial markers.

Conclusions:

• The Bayesian statistical framework offers a powerful and accurate solution for sample drift correction in SMLM.
• This method enhances the reliability and resolution achievable in superresolution imaging by providing precise, frame-by-frame drift compensation.
• The approach simplifies experimental workflows by eliminating the need for fiducial markers.