SecUre Privacy-presERving Medical Image CompRessiOn (SUPERMICRO)

Abstract

The privacy and security of biomedical data are important. Ideally, biomedical data should be kept in a secure manner (i.e. encrypted). With the increasing deployment of the electronic health records, it is critical to make protected health information (PHI) available securely to private and public healthcare providers through the National Health Information Network (NHIN). Efficient transmission and storage of these large encrypted biomedical data becomes an important concern. An intuitive way is to compress the encrypted biomedical data directly. Unfortunately, traditional compression algorithms (removing redundancy through exploiting the structure of data) fail to handle encrypted data. The reason is that encrypted data appear to be random and lack the structure in the original data. The "best" practice has been compressing the data before encryption, however, this is not appropriate for privacy related scenarios (e.g., biomedical application), where one wants to process data while keeping them encrypted and safe. In this paper, we develop a Secure Privacy-presERving Medical Image CompRessiOn (SUPERMICRO) framework based on distributed source coding (DSC), which makes the compression of the encrypted data possible without compromising security and compression efficiency. Our approach guarantees the data transmission and storage in a privacy-preserving manner. We tested our proposed framework on two CT image sequences and compared it with the state-of-the-art JPEG 2000 lossless compression. Experimental results demonstrated that the SUPERMICRO framework provides enhanced security and privacy protection, as well as high compression performance.