Dataset Open Access
Observation data for the LoDoPaB-CT challenge, which asks to reconstruct CT images of the human lung from (simulated) low photon count measurements.
The setting is identical to the one of the LoDoPaB-CT dataset (documented in an arXiv preprint), which is supposed to be employed for training learned methods. This challenge set contains observations for a separate set of patients.
Python utilities for accessing this challenge set and creating the submission file are available at github.com/jleuschn/lodopab_challenge. The LoDoPaB-CT dataset for training can be accessed using the DIVal python library (github.com/jleuschn/dival).
Like for the LoDoPaB-CT dataset, reconstructions from the LIDC/IDRI dataset are used as a basis for this challenge set.
The ZIP file contains multiple HDF5 files. Each HDF5 file contains one HDF5 dataset named
"data", that provides a number of samples (128 except for the last file). For example, the
n-th observation sample is stored in the file
floor(n / 128), at row
(n mod 128) of
For this challenge set no patient IDs are provided (in contrast to the fully public parts of the LoDoPaB-CT dataset), since the reconstruction algorithm should not rely on this information.