Event-guided microfluidic hologram deblurring via spatial-frequency learning.

Area of Science:

• Optics and Photonics
• Biomedical Engineering
• Computer Vision

Background:

• Microfluidic holographic imaging is crucial for particle analysis but struggles with motion blur in high-speed applications.
• Existing deblurring methods are ineffective for the high-frequency patterns in fringe-rich holograms.

Purpose of the Study:

• To develop an advanced deblurring technique for microfluidic holography that overcomes limitations of conventional methods.
• To improve the reconstruction quality of holograms from high-speed moving targets.

Main Methods:

• An event-guided spatial-frequency learning approach was developed, leveraging event sensor temporal resolution.
• An event-guided dual-domain adaptive fusion network (EDAF-Net) was designed with integrated spatial and frequency domain branches.
• A dual-domain fusion module was incorporated to maintain spectral fidelity and spatial accuracy.

Main Results:

• EDAF-Net demonstrated superior high-frequency fringe recovery compared to state-of-the-art methods.
• The proposed network achieved the lowest model complexity and computational load among compared methods.
• Experimental validation confirmed effectiveness for standardized microspheres and human red blood cells.

Conclusions:

• The developed EDAF-Net effectively addresses motion blur in microfluidic holographic imaging.
• This method enhances quantitative particle analysis and flow cytometry applications.
• The approach offers a computationally efficient solution for high-speed holographic deblurring.