3D micro/nano-CT datasets of sandstone rock, lithium-ion battery, and fuel cell used for testing P3T-Net

These are the datasets used for training and testing P3T-Net for 3D unpaired domain transfer in .tif format. Images can be directly opened with ImageJ, Avizo, Python, Matlab, etc.

Dataset contains overall 8 3D images of 4 cases. Each case contains images from the target domain and source domain. 

Case 1. Target domain: micro-CT image of a 9-hour long scan of a sandstone (voxel size: 2.15um); Source domain: micro-CT image of a 7-minute fast scan of a sandstone (voxel size: 2.15um).

Case 2. Target domain: micro-CT image of a 7-hour long scan of a sandstone (voxel size: 3.28um); Source domain: synchrotron micro-CT image of a 24-second fast scan of a sandstone (voxel size: 6.75um).

Case 3. Target domain: nano-CT image of a dual-mode scan of Lithium-ion battery cathode (voxel size: 128nm); Source domain: three nano-CT images of a single-mode scan of lithium-ion battery cathode (two absorption and one phase mode) (voxel size: 128nm).

Case 4. Target domain: micro-CT image of a 3-hour long scan of a fuel cell (voxel size: 2.25um); Source domain: micro-CT image of a 10-minute fast scan of a fuel cell (voxel size: 2.25um).

For using the codes and datasets, please cite the following paper:

Tang, K., Armstrong, R.T., Mostaghimi, P. et al. Scaling deep learning for material imaging with a pseudo 3D model for domain transfer. Nat Commun 16, 11293 (2025). https://doi.org/10.1038/s41467-025-66449-6