E2E-BPF microscope: extended depth-of-field microscopy using learning-based implementation of binary phase filter and...

Area of Science:

• Biomedical imaging
• Optical microscopy
• Computational imaging

Background:

• High-resolution imaging is crucial for biomedical diagnoses.
• Conventional microscopes face a trade-off between depth-of-field and spatial resolution, necessitating time-consuming refocusing.
• Large-scale imaging requires extended depth-of-field capabilities.

Purpose of the Study:

• To develop a computational imaging platform for large-area, high-resolution imaging without serial refocusing.
• To overcome the depth-of-field limitation in conventional microscopy.
• To enable rapid, image-based biomedical diagnoses.

Main Methods:

• Developed a physics-incorporated, deep-learned binary phase filter (BPF).
• Employed a jointly optimized deconvolution neural network.
• Integrated BPF design with deconvolution for extended depth-of-field imaging.

Main Results:

• Achieved high-resolution, high-contrast imaging over extended depth ranges.
• Demonstrated a 15.5-fold larger depth-of-field compared to conventional microscopes.
• Validated through numerical simulations and experiments with fluorescent beads, cells, and tissue sections.

Conclusions:

• The E2E-BPF microscope offers a scalable strategy for depth-of-field-extended optical imaging.
• This method enables large-area, high-resolution imaging without the need for refocusing.
• Potential applications include rapid diagnosis, optical vision, and metrology.