Related Experiment Video
Updated: Apr 8, 2026
07:51
Single-Molecule Localization Microscopy of Membrane Proteins using Single-Antibody Labeling
Published on: March 20, 2026
235
IMAGE REGISTRATION ERROR ANALYSIS WITH APPLICATIONS IN SINGLE MOLECULE MICROSCOPY
1Eric Jonsson School of Electrical Engineering and Computer Science, University of Texas at Dallas, Richardson, TX 75083-0688 USA.
Summary
This study addresses localization errors in fluorescence microscopy image registration. Generalized least squares, accounting for measurement errors, provides a more accurate method than traditional linear least squares for precise image analysis.
Area of Science:
- Microscopy
- Image Analysis
- Biophysics
Background:
- Accurate localization is crucial in fluorescence microscopy.
- Image registration aligns multiple images for analysis.
- Traditional methods struggle with inherent measurement errors.
Purpose of the Study:
- To assess localization errors in monochromatic fluorescence image registration.
- To identify appropriate statistical methods for handling errors-in-variables in image registration.
- To develop a framework for deriving localization errors based on experimental parameters.
Main Methods:
- Framing image registration as an errors-in-variables problem.
- Applying multivariate generalized least squares (GLS) for accurate regression.
- Utilizing an extension of GLS to accommodate non-independent and identically distributed (non-iid) noise.
Main Results:
- Demonstrated that linear least squares is inappropriate due to measurement errors.
- Established generalized least squares as the correct approach for this problem.
- Derived localization errors based on photon counts and experimental parameters using advanced GLS techniques.
Conclusions:
- Accurate image registration in fluorescence microscopy requires advanced statistical methods.
- The proposed generalized least squares approach minimizes localization errors.
- Understanding these errors is key to improving quantitative analysis in microscopy.
