3D radiation field dataset for interventional radiology: Dynamic x-ray tube

This repository contains a dataset of radiation fields stored in the RadFiled3D file format.It contains synthetically generated radiation fields using the Geant4 based RadField3D dataset generator.This dataset is meant to be used for the training of neural networks for spatially resolved doserate prediction. It can be directly used with the pytorch dataloaders and dataset classes of the RadFiled3D python module. This dataset should represent an advanced view on the problem of estimating scattered radiation fields that occur during interventional radiology procedures like angiography or cardiology. Therefore, it contains radiation fields created by an x-ray tube with various output photon spectra, that is centered onto the torso of an headless alderson phantom at a certain distance. The x-ray tubes direction as well as opening angle and distance are randomly sampled for each field. The phantom is surrounded by air. Each radiation field has an extend of 1m x 1m x 1m with a voxel resolution of 2cm. More statistics on the parameters inside the dataset can be found in the statistics.json at the root of the dataset.