The original DICOM is stored unchanged. A reader downloads only the de-identified copy, which does not include the site. This page does not display images and it does not produce a read. Upload uses the imaging API, and files stay within the storage quota.