Features and Limitations

Features

Since the base data format is standard HDF5, you can add any number of your own additional customized attributes, datasets, and groups without violating the format specification. However, you will need to use your own HDF5 file write/read scripts to save and retrieve these extra attributes. See the Code Examples section on extending the file format.

Furthermore, since the base file format is HDF5, these files are suitable for archival storage. HDF5 is a major scientific data format. HDF5 will continue to be readable long into the future, in all variety of computer programming languages, even without pradformat readers and writers.

Compression of the datasets within the HDF5 files is performed in both the Python writers and in the MATLAB writers. Compression can dramatically reduce file size. Both Python and MATLAB readers can read compressed or uncompressed datasets.

Limitations

For simplicity, the MATLAB and Python readers in this package read all objects within the HDF5 files into memory (uncompressed). This is not a limitation of HDF5. If you run into memory issues while loading large datasets, I suggest writing your own bespoke MATLAB or Python code to read just the data you need using the native HDF5 readers for MATLAB and Python (and thus avoid loading the entire datasets into RAM).

Future Work

For distribution of common benchmark files, I'm imagining that we could create (eventually) a Radiography Commons with test grids of E&M fields and test radiographs. I will plan to host a small set of example files on this website for now.