if the images are publicly available nothing prevents to make them available also on another repository such as OAR. If there are concerns for their use for commercial purposes, we can associate to them the CC-BY-NC-ND (Creative Commons by Attribution, Non Commercial, No Derivatives) licence. An advantage of putting them on the OAR is the possibility to assign DOIs to them (if they don't already have any) and make them processable by portals and Science Gateways, such as MIPAR.
The only question is: should we download the images from the UK BioBank one by one manually or are there APIs to let a compute do the job in one go? Could you please check on their website?