LiveLattice: Real-time visualization of tilted light-sheet microscopy data using a memory-efficient transformation...

Area of Science:

• Biophysics
• Computational Biology
• Microscopy

Background:

• Light-sheet fluorescence microscopy (LSFM) provides 4D imaging capabilities.
• Tilted-sample-plane LSFM techniques like iSPIM and LLSM simplify sample preparation but require complex post-processing.
• Current deskewing and rotation methods for LSFM data are computationally intensive, limiting real-time analysis.

Purpose of the Study:

• To develop a novel, memory-efficient algorithm for fast preprocessing of tilted-sample-plane LSFM data.
• To enable real-time visualization and analysis of large 4D microscopy datasets.
• To reduce the computational burden and hardware requirements for LSFM data processing.

Main Methods:

• Developed WH-Transform, a GPU-accelerated algorithm integrating deskewing and rotation into a single transformation.
• Implemented a memory-efficient approach to handle large image stacks.
• Benchmarked WH-Transform against conventional methods using large 3D and 4D datasets.

Main Results:

• WH-Transform reduces preprocessing run time by at least 10-fold compared to existing methods.
• The algorithm demonstrates linear scalability, processing 15 GB 3D stacks in under a minute on a single GPU.
• Real-time preprocessing was achieved for 4D datasets of human hepatocytes, lung, and brain organoids.

Conclusions:

• WH-Transform significantly accelerates LSFM data preprocessing, overcoming memory and computational limitations.
• The algorithm enables real-time, on-the-fly data processing and visualization on standard workstations.
• This advancement revolutionizes biological imaging applications by improving the usability of LSFM, SPIM, and similar microscopes.