Digital-SMLM for precisely localizing emitters within the diffraction limit

Area of Science:

• Biomedical imaging
• Computational biology
• Optical microscopy

Background:

• Precisely locating emitters below the diffraction limit is vital for quantitative analysis in biomedical research.
• Existing methods, including single molecule localization microscopy (SMLM), face challenges in achieving high accuracy for closely spaced emitters.

Purpose of the Study:

• To develop a novel deep learning-based approach, Digital-SMLM, for accurate prediction of emitter numbers and positions below the diffraction limit.
• To enhance the capabilities of SMLM for more precise molecular mechanism investigation.

Main Methods:

• Integration of experimental spot datasets with deep learning algorithms.
• Development and validation of the Digital-SMLM model for emitter localization.
• Comparative analysis against existing methods like Deep-STORM.

Main Results:

• Digital-SMLM achieved up to 98% accuracy and a 14 nm root mean square error in predicting emitter positions.
• The method successfully resolved emitters as close as 30 nm apart.
• Digital-SMLM demonstrated superior performance over Deep-STORM in accuracy and ground truth recovery.

Conclusions:

• Digital-SMLM significantly improves the precision of emitter localization below the diffraction limit.
• This approach enhances SMLM's ability to approximate the natural state of high-density cellular structures.
• Digital-SMLM offers a powerful tool for advanced quantitative analysis in biomedical research.