The NiftyPET software, thanks to the high-throughput processing using dedicated parallel computing devices, enables processing of bigger datasets for large clinical trials. It facilitates fast reconstruction of dynamic four-dimensional images and thus allows more robust and faster research on disease modelling as well as novel imaging methods dedicated to dementia. The paper relates to a software package that can used in the reconstruction brain images acquired from MR-PET scanners. The software is an open-source Python package and is currently being extended to support all the MR-PET scanners deployed in the UK within the Dementias Platform UK network (DPUK) for harmonised image reconstruction and analysis in multi-centre clinical trials.