Efficient Double Helix Detection with Steerable Filters

Area of Science:

• Biophysics
• Optical Microscopy
• Computational Imaging

Background:

• Accurate 3D single-molecule localization microscopy (SMLM) is crucial for understanding molecular behavior.
• Localization precision is often limited by the accuracy of point-spread function (PSF) detection and fitting.
• Existing deep learning methods for PSF analysis can be computationally intensive.

Purpose of the Study:

• To present an efficient detection scheme for localizing Double Helix point-spread functions (PSFs) in 3D SMLM.
• To develop a computationally inexpensive yet accurate method for determining both lateral and axial positions of single molecules.
• To integrate this scheme into a complete SMLM analysis pipeline.

Main Methods:

• Utilized steerable filters for rapid extraction of 2D position and lobe orientation (axial position) from Double Helix PSFs.
• Employed a double Gaussian model fitter with optimal parameterization for enhanced localization accuracy.
• Implemented the detection and fitting scheme as a plugin for the open-source PYthon Microscopy Environment (PYME).

Main Results:

• Achieved efficient localization using only 7 convolutions, a significant reduction compared to deep learning approaches.
• Demonstrated accurate estimation of both 2D position and axial position from Double Helix PSFs.
• Successfully integrated the method into a functional SMLM analysis pipeline within PYME.

Conclusions:

• The proposed steerable filter-based detection scheme offers a computationally efficient alternative for 3D SMLM.
• This method provides accurate localization of Double Helix PSFs, improving SMLM analysis.
• The PYME plugin facilitates the adoption of this advanced technique in biological research.