quickPIV

quickPIV performs three-dimensional particle image velocimetry (3D PIV) analyses to quantify collective cellular migration during embryogenesis.

Key Features:

• Implementation and performance: Implemented in Julia with CPU optimization and reported to be three times faster than the Python-based openPIV for 2D and 3D and faster than the fastest 2D openPIV C++ package.
• Segmentation-free PIV: Applies segmentation-free PIV suited to non-segmentable biological image datasets.
• Cross-correlation methods: Supports zero-normalized cross-correlation and normalized squared error cross-correlation for translation detection.
• Sub-pixel/voxel and multi-pass refinement: Provides sub-pixel/voxel approximation and multi-pass strategies for improved displacement estimation.
• Post-processing metrics: Offers filtering and averaging of vector fields, extraction of velocity, divergence, and collectiveness maps, simulation of pseudo-trajectories, and unit conversion.
• Visualization: Integrates with Paraview for 3D vector field visualization.
• Validation: Evaluated on synthetic data with accuracy consistent with published expectations.
• Scalability: Applicable to large-scale 3D imaging datasets, including terabyte-scale data.

Scientific Applications:

• Developmental biology — embryogenesis: Quantifies collective cellular migration during gastrulation in organisms such as Tribolium castaneum.
• Labeled embryo imaging: Generates vector fields from nuclear- and actin-labeled embryos to characterize morphogenetic movements.

Methodology:

Implemented in Julia with CPU optimization, quickPIV uses segmentation-free PIV employing zero-normalized and normalized squared error cross-correlations with sub-pixel/voxel approximation and multi-pass refinement, and has been evaluated on synthetic data.