A recurring question is how to make sure that images of human tissue you decide to use in My Pathomation are compliant with privacy regulations.

These simple rules will help you decide when anonymization of identifiable information for whole slide images of human material becomes necessary. You can always ask advice to your Data Protection Officer.

If you are sharing images with medical professionals that have a therapeutic relation with the patient or you have a research project that has the necessary approval the measures you have already implemented are sufficient and no additional steps need to be taken. My Pathomation runs on Azure cloud services that guarantee security for medical information and you can control who you grant access to the content you have published. When you share human tissue images in public like at a conference or to teach pathology to students, privacy becomes a requisite and all identifiable data must be removed before sharing these images.

Be aware that an accession number (e.g. B22-12345) is pseudonymization not true anonymization, the GDPR regulations still consider this to be identifiable data.

There are 3 places where identifiable data are typically found in whole slide images:

• The name of the file, as it can contain a biopsy number or even a patient name
• The label of the glass slide, as it can have a biopsy number, dates, hospital or patient names
• The metadata embedded with the image in the file, this metadata can be diverse and extensive as it can collect data from the slide label and the laboratory information system.

To assure full compliance with privacy regulations, perform the following steps prior to publishing online:

1. Change the names of your images to be non-identifiable, eg B22-12345 to Case 1 or Hepatocellular carcinoma,..
2. Run the Pathomation DICOMizer tool on your image. The DICOMizer tool converts your images to the DICOM standard and removes all the metadata from your original image during the conversion process. The resulting images files will have a thumbnail file for the label that can then be deleted without consequence for the integrity of the image.