GPU accelerated real-time confocal fluorescence lifetime imaging microscopy (FLIM) based on the analog mean-delay...

Area of Science:

• Biophotonics and Imaging
• Microscopy Technologies
• Computational Imaging

Background:

• Confocal fluorescence lifetime imaging microscopy (FLIM) provides critical functional information about biological samples.
• Real-time FLIM is essential for observing dynamic biological processes but is computationally demanding.
• Existing FLIM methods often face limitations in speed and processing efficiency.

Purpose of the Study:

• To develop and validate a GPU-accelerated real-time confocal FLIM system.
• To enhance the speed and efficiency of FLIM data acquisition and processing.
• To enable high-frame-rate imaging for dynamic biological and industrial applications.

Main Methods:

• Implementation of a GPU-accelerated algorithm based on the analog mean-delay (AMD) method for FLIM.
• Verification of the algorithm across diverse fluorescence lifetimes and photon counts.
• System performance evaluation against physical scanning and single-core CPU processing.

Main Results:

• The GPU-accelerated FLIM system achieved processing speeds exceeding physical scanning for images up to 800x800 pixels.
• Processing was over 149 times faster than a single-core CPU.
• A frame rate of 13 fps was demonstrated for a 200x200 pixel image of maize vascular tissue.

Conclusions:

• The developed GPU-accelerated FLIM system offers significant speed improvements for real-time imaging.
• This technology is suitable for observing dynamic biological reactions, aiding medical diagnosis, and enabling real-time industrial inspection.
• The AMD method, when GPU-accelerated, provides a powerful tool for advanced microscopy applications.