Lossless Watermarking for Data Security and Integrity of Medical Images

DESCRIPTION (provided by applicant): With the rapid development and deployment of high-performance computer and sensor technologies, digital imaging has come to its primetime and has replaced traditional film-based systems as the most widely used technology for medical image acquisition, processing, archiving, and communication. While digital imaging offers many great advantages over its analog counterpart, the success of the Internet and cost-effective digital recording has made it possible to create, replicate, transmit, and distribute digital content in an effortless way. The vulnerability to unauthorized use and alteration associated with easy access to the digital content of medical images now raises serious concerns. Protection of patient privacy and enforcement of data security for digital medical records has thus become increasingly important as the Health Insurance Portability and Accountability Act (HIPAA) Security Standards mandate health institutions to protect health information against unauthorized use or disclosure. The goal of this project is to complete the development of novel lossless digital watermarking techniques for data security and integrity of medical images. Existing digital watermarking products only provide lossy watermarking solutions that are primarily used in the entertainment industry for copyright protection of digital music, pictures and DVDs. These products are inapplicable to medical images because the lossy operation causes permanent alteration to the digital image content after watermarking. On the other hand, current DICOM standards used in clinical applications encode patient information separately and append it to the image as a header. There are reported cases where digital clinical images become separated from their DICOM headers and therefore have HIPAA implications. For clinical research, sometimes images must be anonymized by altering the DICOM headers to lessen the HIPAA impact and to reduce bias from knowing patient information in the analysis of the images. Hence there is clearly a need for an effective tool that would enable medical image watermarking with the patient information and the records such as the institution and accession number to allow tracking the image back to its source. To our knowledge, no commercial product exists at present time that provides such a tool. To address the need and seize the emerging market opportunity, the proposed project seeks to develop innovative techniques and products that will uniquely offer lossless watermarking solutions to medical imaging applications. Our approach is to make the usage, archiving, and communication of medical images much more secure and protectable, by embedding imperceptible watermarks into the images. These watermarks remain present after copying, network/Internet transmission, and even some alteration, and they can be retrieved when needed. Important information, such as patient identification, physician signatures, diagnostic specification and user history, can all be encrypted and encoded into the watermarks along with the image authentication code. Yet the original image can still be perfectly recovered from the watermarked image upon decoding. Hence the new technology will enable clinicians and researchers to protect effectively both patient privacy and the authenticity of their image data. During the Phase 1 project, we developed methods and software to achieve lossless region-of-interest (ROI) watermarking of medical images. We evaluated the new methods using several data sets of monochrome medical images from cytogenetic and radiology applications, based on the criteria of (1) imperceptibility, (2) capacity, and (3) robustness of the embedded watermarks. Test results show that the lossless ROI watermarking approach works very well and clearly exceeds the Phase 1 feasibility criteria. In the Phase 2 project we will complete the development of the proposed technology. We will refine the lossless ROI watermarking by optimizing several system parameters. We will extend the methodology to support lossless ubiquitous watermarking of medical images. We will also incorporate encryption functions and color transformations in the system to provide additional security for patient information and the support for color image watermarking. We will evaluate the resulting algorithms and software for a variety of medical imaging applications including radiology, pathology and cytogenetics. We will beta test the new technology in the clinical environment and optimize the technology as end-user input and feedbacks are gathered. Once fully developed and qualified, this new technology will be patented and incorporated into future IRIS (ADIR's parent company) in-vitro diagnostic imaging products. It also will be made commercially available through licensing agreements and partnerships with manufacturers of medical imaging systems and companies of general image processing and publishing software products. Finally, we will seek DICOM and JPEG2000 adoption and standardization for broader application and dissemination. This project develops innovative software techniques that will offer currently unavailable lossless watermarking solutions to medical imaging applications. The approach is to make the usage, archiving, and communication of medical images much more secure and protectable, by embedding imperceptible watermarks into the images. These watermarks remain present after copying, network/Internet transmission, or even some alteration, and can be retrieved when needed. Important information such as patient identification, physician signatures, diagnostic specification and user history can all be encrypted and encoded into the watermarks along with the image authentication code. Yet the original image can still be perfectly recovered from the watermarked image upon decoding. Hence the new technology will enable clinicians and researchers to effectively protect the patient privacy and integrity of their image data.