The following steps are automatically performed in all executions of the Medical Image Data analysis:
- Virus scanning: A check for viruses within the uploaded ZIP file. If a virus is present, all the data within the ZIP file is immediately deleted and you will be warned via the QC Status as explained in this article.
- De-identification: Only applicable to DICOM files. The PHI present within the DICOM headers is removed following the criteria explained in this article. Skipping the deidentification step is also possible per QMENTA Project upon request. Note that PHI included in PDF Reports, directory names, file names, or text within the images themselves are not removed. It is your responsibility to remove this PHI.
- File type recognition: Recognition of the individual files included in the ZIP file. An extensive list of supported file types is included in this article. Non-supported files will not be classified.
- Metadata extraction: Extraction of extensive sets of metadata from the header of the uploaded files.
- Harmonization: Only applicable to DICOM files. The metadata extracted from the DICOM headers of different manufacturers is harmonized.
- Classification: Performs the classification of each medical imaging file. This classification is primarily based on the metadata extracted in previous steps, on the name of the file, and on a deep learning classifier based on the image intensities. This classification results in the addition of modalities and tags to all the identified files.