Deep learning-based aberration compensation improves contrast and resolution in fluorescence microscopy.

Area of Science:

• Biophysics
• Computational Biology
• Microscopy

Background:

• Optical aberrations degrade fluorescence microscopy images of thick biological samples, limiting signal, contrast, and resolution.
• Existing aberration correction methods often require specialized hardware or increase imaging time and light dose.

Purpose of the Study:

• To develop a deep learning-based method for efficient aberration compensation in fluorescence microscopy.
• To improve image quality and downstream quantitative analysis without altering the imaging setup or acquisition parameters.

Main Methods:

• A deep learning strategy was developed to introduce synthetic aberrations into shallow image planes.
• Neural networks were trained to reverse these synthetic aberrations, effectively correcting images acquired deeper within samples.
• The method was validated using simulations and experimental data across various microscopy techniques.

Main Results:

• The deep learning 'de-aberration' networks significantly improved image quality, matching the performance of adaptive optics.
• Restored images facilitated better qualitative inspection and quantitative analysis in diverse microscopy datasets.
• Specific improvements included enhanced blood vessel orientation analysis in mouse tissue and better cellular segmentation in C. elegans embryos.

Conclusions:

• Deep learning offers a powerful, non-invasive approach to correct optical aberrations in fluorescence microscopy.
• This method enhances the utility of standard microscopy techniques for imaging thick biological specimens.
• The developed strategy improves both the visual quality and quantitative accuracy of microscopy data for biological research.