Amazon has developed a system that can automatically remove patients' identifying protected health information from medical images used in research, according to a company blog post published by James Wiggins, Amazon Web Services senior healthcare solutions architect.
The system uses Amazon's machine learning, image analysis service Rekognition, which can identify and extract text from images. That text is then run through Amazon Comprehend Medical, the company's natural language processing, EHR-mining software, to help detect PHI.
Manually removing PHI from medical images can take multiple minutes, causing de-identifying large datasets to be costly and time consuming.
To access the full report, click here.