The discriminative bilateral filter: an enhanced denoising filter for electron microscopy data.

Area of Science:

• Structural Biology
• Biophysics
• Microscopy

Background:

• Three-dimensional electron microscopy (3D EM) offers high resolution for biological structures.
• Specimen beam damage and low signal-to-noise ratios (SNRs) in low-dose cryo-EM limit high-frequency information recovery.
• Existing denoising filters struggle to differentiate noise from fine structural details.

Purpose of the Study:

• To introduce and evaluate a novel discriminative bilateral (DBL) filter for enhancing 3D EM image quality.
• To demonstrate the DBL filter's ability to reduce noise without compromising structural integrity.
• To assess the DBL filter's utility in single particle analysis and cellular tomogram processing.

Main Methods:

• Development of a discriminative bilateral (DBL) filter, an advancement over the original bilateral filter.
• Inclusion of a photometric exclusion function to distinguish noise from object edges.
• Application of the DBL filter to low SNR single particle data and cellular tomograms of stained plastic sections.

Main Results:

• The DBL filter effectively smoothed high-frequency noise pixels while preserving object edge detail.
• Significant noise reduction was achieved in low SNR single particle data.
• The filter demonstrated efficacy in denoising cellular tomograms from stained plastic sections.

Conclusions:

• The DBL filter is a valuable tool for improving image quality in 3D EM.
• It offers enhanced noise reduction capabilities compared to previous methods.
• The DBL filter shows promise for pre-processing data in single particle analysis and cellular tomography, aiding subsequent image segmentation.