Synthetic Vascularized Digital Skin Phantoms for Photoacoustic Simulation and Reconstruction

This dataset contains 60,000 two-dimensional synthetic vascularized digital skin phantoms for photoacoustic simulation, reconstruction and machine-learning research. Each sample provides an optical absorption map (mua), simulated optical fluence map (fluence), initial-pressure distribution (IP) and acoustic sinogram (sinogram). The complete dataset is distributed as 20 losslessly compressed HDF5 shards of 3,000 samples each. A uniformly selected 100-sample HDF5 subset contains all four variables for rapid testing. The record also includes a sample index, README, Python requirements, and the reviewed MATLAB phantom-generation and ValoMC/k-Wave simulation code. The released mua, fluence and IP arrays use aligned depth-by-lateral coordinates. The data are synthetic and contain no identifiable human or animal participant information.