Dataset of Fluorescent Nuclear Track Detector images for neutron dosimetry

This dataset was originally generated in a previous publication:
Schmidt, S., Christensen, J.B., Lutz, B., Stabilini, A., Yukihara, E.G., Vedelago, J., 2025. Sensitivity analysis of fluorescent nuclear track detectors for fast and high-energy mono-energetic neutron dosimetry. Medical Physics 52, e17799. https://doi.org/10.1002/mp.17799.

This dataset was also used in:
Thai, L.-Y. J., Schmidt, S., Walter, A., Häcker, R.V. , Giske, K., Vedelago, J., 2026. Capability of deep learning to predict recoil protons for neutron dosimetry with Fluorescent Nuclear Track Detectors. Radiation Measurements, 107662. https://doi.org/10.1016/j.radmeas.2026.107662.

This dataset contains raw images and binary reference label masks for the training and testing.

The training dataset comprises of all six mono-energetic neutron cases:
- 1800 raw images can be found in "imagesTr",
- 1800 binary reference label masks can be found in "labelsTr".

The test dataset comprises of all seven ambient dose equivalent H*(10) values for the 241Am-Be neutron source:
- 2100 raw images can be found in "imagesTs",
- 2100 binary reference label masks can be found in "labelsTs".
The seven H*(10) values were separated into subfolders:
- "ss0643_44_45" -> 0 mSv,
- "ss0610_11_12" -> 1 mSv,
- "ss0613_14_15" -> 5 mSv,
- "ss0616_17_18" -> 10 mSv,
- "ss0637_38_39" -> 15 mSv,
- "ss0601_02_03" -> 50 mSv,
- "ss0604_05_06" -> 100 mSv.

Note: In order to properly visualize the 16-bit images, intensity-scaling (e.g. with ImageJ/FIJI) might be required. Otherwise, the images might appear black.