Background removal in fluorescence microscopy images based on baseline estimation.

Area of Science:

• Biomedical Imaging
• Computational Microscopy
• Optical Physics

Background:

• Wide-field fluorescence microscopy suffers from out-of-focus background haze, degrading image contrast and obscuring cellular details.
• Existing methods often require complex hardware or lack real-time applicability.

Purpose of the Study:

• To develop a high-performance computational sectioning framework for fluorescence microscopy.
• To improve image contrast and subcellular detail without specialized equipment.
• To provide a robust, interpretable, and cost-effective solution for deep-tissue imaging.

Main Methods:

• A physics-informed variational model was developed to separate in-focus signals from defocused background.
• Penalised least-squares optimisation was employed for precise signal-background separation.
• A Gaussian-equivalent regularisation strategy was introduced to enhance computational efficiency for real-time imaging.

Main Results:

• The computational framework significantly increased the Resolution Scale Pearson (RSP) coefficient by 1.5-fold.
• The method achieved a six-fold reduction in the Resolution Scale Error (RSE) compared to confocal microscopy.
• Performance was benchmarked favorably against DeepMRA and HiLo techniques.

Conclusions:

• The proposed framework offers a low-cost, robust, and physically interpretable solution for high-contrast deep-tissue fluorescence imaging.
• This computational approach overcomes fundamental limitations of wide-field microscopy.
• The method enhances image quality and enables clearer visualization of subcellular structures.