examples

Examples

Self-contained scripts demonstrating each part of the ffdas pipeline. Each example simulates synthetic data, reconstructs an image, and saves the result as a PNG.

The examples use conservative data sizes by default so they run on most CUDA-capable GPUs. If your GPU has more memory, increase batch_size, grid resolution (nz, ny, nx), and n_scatterers to see what the library can do at scale.

Scripts

• reconstruct.py — Single plane-wave reconstruction. Simulates diverging-wave channel data with greens, then reconstructs a 3D volume with das.
• compound.py — Coherent plane-wave compounding. Transmits at several steering angles and combines the resulting images with das.
• sparse_compounding.py — Sparse compounding with das_sparse. Diverging waves from virtual sources at different angles each illuminate only part of the volume. Each voxel selects the relevant transmit events, avoiding iteration over zero-weight contributions.
• clutter_filter.py — Clutter filtering. Reconstructs a slow-time sequence, then separates stationary tissue from moving flow using truncate_rank.
• interpolation.py — Grid interpolation. Reconstructs on a spherical grid, then interpolates back to Cartesian coordinates.

Requirements

All examples require CuPy and matplotlib:

pip install cupy-cuda12x matplotlib

The interpolation.py example also requires Pillow (pip install Pillow).