by Yutong Wu
I am a researcher at the Johann Radon Institute for Computational and Applied Mathematics (RICAM), part of the Austrian Academy of Sciences at Johannes Kepler University (JKU) Linz. My work focuses on single-molecule localization microscopy (SMLM). Together with Simon Hubmer and Ronny Ramlau (JKU Linz), as well as Montserrat López Martínez and Gerhard J. Schütz (TU Vienna), we explore and advance SMLM using a biplane imaging system.
Working Principle of SMLM
Over the past two decades, the development of super-resolution fluorescence microscopy has overcome the classical diffraction limit, enabling visualization of biological structures at the nanoscale. However, conventional fluorescence microscopy suffers from the overlap of emission from densely packed fluorophores, which blurs fine structural details and limits spatial resolution. To overcome this limitation, SMLM was developed, providing a powerful approach to reveal molecular architectures with nanometer precision.
Figure 1: The imaging concept of SMLM. The upper-left panel shows the diffraction-limited image, in which all fluorophores emit simultaneously, resulting in blurred, unresolved spots. The bottom sequence depicts the SMLM acquisition process, where individual fluorophores stochastically emit in separate frames, allowing precise localization of each molecule. By iteratively accumulating these localizations across multiple frames, a super-resolution reconstruction is obtained, as illustrated in the upper-right panel, revealing fine structural details beyond the diffraction limit.
SMLM achieves super-resolution imaging by combining sparse activation of fluorophores, precise localization of single molecules through PSF fitting, and multi-frame accumulation, as illustrated in Fig. 1. In the diffraction-limited image, all molecules emit simultaneously, producing blurred spots that cannot be individually resolved.
In contrast, sparse activation frames ensure that only a few molecules are active per frame, allowing accurate determination of their positions. By accumulating these localizations over many frames, a super-resolution reconstruction is obtained, uncovering structural details far beyond the diffraction limit.
Biplane Imaging System
To estimate the three-dimensional localization of each emitter, we analyze the intensity distribution at the back focal plane, I (x, y; ξ , Θ, Φ), which depends on both the emitter’s spatial position ξ and its orientation angles, the inclination Θ and the azimuth Φ. To fully exploit the information, the molecular orientation must first be determined. Therefore, the localization process is divided into two sequential steps: first, the orientation is estimated, and then the three-dimensional position is calculated. To determine the orientation of single molecules, we employ a biplane imaging system that simultaneously captures two planes separated by a known biplane distance. The entire workflow is visualized in Fig. 2. For each plane, the molecular point spread function (PSF) is fitted with a two-dimensional Gaussian to extract the key PSF centroid position, shown as the red dot in Fig. 2. Using the centroids obtained from Gaussian fitting on the two planes, the orientation can be calculated from the relative positions of the PSF centers and the known biplane distance.
Once the estimated orientation, Θ̂ and Φ̂, is obtained, the corresponding intensity distribution model I (x, y; ξ, Θ̂, Φ̂) is constructed accordingly. The observed photon distribution of each emitter is then compared with the modeled distribution. The three-dimensional location is finally determined as the position that maximizes the likelihood from the observed photon distribution. Based on this approach, we reconstruct emitter positions with high accuracy, enabling in-depth exploration of fine structural details.
Future Work
In the present work, we address the core challenge of single-frame localization and achieve highly accurate and reliable position estimates for individual emitters. Building upon this foundation, our future research will focus on the incorporation of temporal information and emitter dynamics, where multiple fluorophores blink stochastically over time, enabling the joint analysis and reconstruction of all active molecules across frames.