[Whitepaper] The Essential Features of a Clinical Data and Diagnostic Imaging De-Identification Process That Can Operate at Enterprise Scale

There is growing industry pressure to more readily share clinical trial data in order to accelerate medical insights.  However, a vast amount of medical image data collected in the course of clinical trials has been unusable for research and innovation. One common reason is the lack of a reliable method of removing personally identifiable information (PII) and personal health information (PHI) from medical image datasets, a process known as de-identification.

De-identification is difficult to perform at enterprise scale for all forms of clinical data, but is particularly difficult for medical images, due to the complexity and lack of standardization in DICOM metadata.

Google Cloud created a de-identification application programming interface (API) as one of the tools available in its Cloud Healthcare API, which are standards-based APIs created to power actionable healthcare insights for security and compliance-focused environments.

Life Image evaluated the de-identification capabilities of the Google Cloud Healthcare API against a number of open source, proprietary and commercial de-identification programs and found that the Google Cloud Healthcare API outperformed other tools in the market in its accuracy at de-identifying PHI contained in imaging pixels, metadata tags and associated notes.

Life Image further augmented the performance of the Google Cloud Healthcare API with a value-add service that uses a combination of machine learning and human validators to achieve accuracy levels consistent with some of the most stringent compliance requirements and beyond those operating in healthcare today. Combined, the Life Image and Google Cloud Healthcare API can help protect patient information and promote accelerated research development.



Sponsored by: