Anonymize a folder
Apply one anonymization profile to every study in a folder and write the result to another folder.
Desktop app only, 1.6.0 or newerdcmage pro
Choose Anonymize a folder on the start screen, or Anonymize a Folder… in the session menu. Choose the folder to anonymize and an empty folder for the output, pick a profile, and start. The source folder is only read.
From an archive. To anonymize studies straight from an archive, tick them in the archive browser and choose Anonymize to a folder…. The dialog opens with those studies as the source. See a cohort from an archive.
Profile. Basic, HIPAA, Teaching or your Custom rules. +Pixel is not offered, because it stands for a redaction you draw on an image and nobody is looking at each image in a run.
Patients. Each Patient ID gets one pseudonym for the whole run, so one patient’s studies stay linked. A file with no Patient ID is keyed on its Patient Name, or on its study when the name is missing too.
Subject IDs. To use a trial’s own IDs, choose Use a table… and pick a CSV file with a PatientID column and a SubjectID column. A patient the table does not list is not written. See subject IDs from a table.
Shift dates. Each patient gets their own offset, so the intervals between their studies are kept.
Keep each pseudonym in the vault. Asks for your vault passphrase, twice, and stores each pseudonym with its original ID, so it can be reversed later. If the passphrase does not open what the vault already holds, patients from that point on are not written, and the next run asks for it again.
Burned-in text. A run rewrites tags and does not read images. Files that declare burned-in text are held back, and so are ultrasound, secondary capture and photographic files that do not say either way. Tick the option to write them anyway, and the result tells you which images to check before you share them.
What is written. Files go to pseudonym/study/series/instance.dcm, built from anonymized values, so a folder named after a patient does not follow the data. Before each file is written, dcmage reads it back and looks for the patient’s name, ID and accession number in its text. A file that still holds one is not written, and the result names the tag.
The result. One line per file found, saying what happened to it. A file that could not be anonymized is never copied. Each file also gets a line in the audit log, by number and without its source path, and the run gets one line of totals.
The signed record. After a run that wrote at least one file, Write signed record adds a signed manifest and a readable report to the output folder. See Signed manifest of a batch run and Readable audit report of a batch run. A cancelled or partial run is recorded as incomplete.
The re-link file. After a run, you can save which patient each pseudonym stands for, encrypted under a passphrase you choose, anywhere but the output folder. See the re-link file.
Cancel. The run stops before the next file. A file still being retrieved from an archive is dropped at once. The output folder keeps the files already written, each one complete, and the rest are listed as not processed.
The output folder must be empty and must not overlap the source folder. dcmage refuses the run otherwise, before it reads anything.
From Terminal, on a Mac. The app keeps access to each folder you choose here, so a run can also be started from Terminal on this Mac. It is not for CI, Docker, SSH or Linux, and it reaches no other folder. Forget all folders ends that access for every folder. See running it from Terminal.
From PowerShell or Command Prompt, on Windows. The app keeps no folder access on Windows, because the dcmage command reaches any folder you can. It does share the secret behind its pseudonyms with that command, so a run started there gives each patient the same pseudonym as a run here. The copy is encrypted to your Windows account. Forget the shared secret removes it, and the command then runs nothing until you choose a folder here again.