We currently are using DCMTK to

1. read certain meta tags;
2. write anonymised image by replacing PatientBirthDate and PatientName with non-identifiable value.

Current issue is 1+ Gb DICOMs which encode a large sequence of frames. It causes an unreasonable amount of CPU and RAM to read PixelDatas from such DICOMs, simply to write them to an anonymised file without any processing.

This library could be a great fit for anonymisation workflow if, for example, it would allow to It would be great if this library supported returning an io.Reader for the whole PixelData element, so that non-image elements could be interpreted but image elements could be streamed to output file without change.