Medical imaging, such as obtained with Computed Tomography (CT) or Magnetic Resonance Imaging (MRI) has become very common in clinical practice. For example, cancer patients who are eligible for surgery are routinely referred to the radiology department for a CT scan, often multiple times. The images are then visually inspected by both the radiologist and the clinical expert responsible for the patient. The images provide crucial information for diagnosis, treatment and outcome evaluation.
Apart from their clinical usefulness, CT and MR images also contain a wealth of information for research purposes. For example, using the latest radiomics techniques researchers are able to build effective and personalized prediction models that predict surgical outcome or radiotherapy response by analysing the image gray values (pixels) and calculating hundreds or thousands of statistical features from these images.
DataHub already has general-purpose data storage facilities in place but is now working on a storage solution that is dedicated for clinical imaging data and also allows online processing and analysis of this data without the need for downloading it first. The majority of this data is in DICOM format and needs to be anonymized because DICOM files contain privacy-sensitive information about the patient. At DataHub we will setup an environment where anonymization can happen automatically as soon as images are uploaded/ingested into DataHub.
Ralph Brecheisen, member of DataHub and Clinical Data Scientist at the Surgery Department of MUMC.