Algorithmic corrections for localization microscopy with sCMOS cameras - characterisation of a computationally...

Area of Science:

• Microscopy
• Optical Imaging
• Biophysics

Background:

• Scientific CMOS (sCMOS) cameras offer high speed for single molecule localization microscopy (SMLM).
• Pixel non-uniformities in sCMOS sensors (read-noise, offset, sensitivity variations) can reduce localization precision and accuracy.
• Existing methods may not fully address sCMOS-specific noise characteristics.

Purpose of the Study:

• To develop a simplified noise model for sCMOS cameras.
• To create a novel weighted least squared (WLS) fitting method to correct for sCMOS pixel non-uniformities.
• To evaluate the necessity and performance of sCMOS-specific corrections in SMLM.

Main Methods:

• Developed a simplified sCMOS non-uniform noise model.
• Implemented a new weighted least squared (WLS) fitting algorithm for localization.
• Conducted simulations to test correction requirements under various conditions.
• Validated the WLS method using experimental data from sub-resolution beads and DNA origami.

Main Results:

• Simulations indicated that pixel-specific offsets should always be corrected.
• Read-noise and sensitivity corrections are often negligible in thick samples but crucial for thin samples or fast imaging.
• The WLS method recovered the performance of an ideal uniform sensor and achieved the Cramer-Rao lower bound.
• Experimental results confirmed the simulation findings.

Conclusions:

• A simplified sCMOS noise model and WLS fitting method effectively correct for pixel non-uniformities in SMLM.
• The WLS approach provides computationally efficient and accurate localization for various SMLM modalities (2D, 3D, multi-emitter).
• This method enhances the reliability of super-resolution microscopy using sCMOS cameras.