Lossless Watermarking for Data Security and Integrity of Medical Images

描述（由申请人提供）：随着高性能计算机和传感器技术的快速发展和部署，数字成像进入了鼎盛时期，并取代了传统的基于胶片的系统，成为医学图像采集、处理、存档和通信中应用最广泛的技术。虽然数字成像比模拟成像提供了许多巨大的优势，但互联网的成功和经济高效的数字记录使得轻松地创建、复制、传输和分发数字内容成为可能。医学图像的数字内容容易被未经授权的使用和修改，这引起了人们的严重关注。因此，随着《健康保险流通与责任法案》（HIPAA）安全标准要求医疗机构保护健康信息免遭未经授权的使用或披露，保护患者隐私和实施数字医疗记录的数据安全变得越来越重要。本计划的目标是完成新型无损数字水印技术的发展，以确保医学图像的数据安全和完整性。现有的数字水印产品只提供有损水印解决方案，主要用于娱乐行业，用于数字音乐、图片和dvd的版权保护。这些产品不适用于医学图像，因为有损操作会导致数字图像内容在水印后发生永久性改变。另一方面，目前临床应用中使用的DICOM标准对患者信息进行单独编码，并将其作为标头附加到图像中。有报道的情况下，数字临床图像与他们的DICOM头分离，因此有HIPAA的影响。对于临床研究，有时必须通过更改DICOM标头对图像进行匿名处理，以减轻HIPAA的影响，并减少在分析图像时因了解患者信息而产生的偏差。因此，显然需要一种有效的工具，能够对患者信息和记录（如机构和登记号码）进行医学图像水印，以便跟踪图像回到其来源。据我们所知，目前还没有商业产品提供这样的工具。为了满足需求和抓住新兴市场机遇，拟议项目寻求开发创新技术和产品，为医学成像应用提供独特的无损水印解决方案。我们的方法是通过在图像中嵌入难以察觉的水印，使医学图像的使用、存档和通信更加安全和可保护。这些水印经过复制、网络/互联网传输，甚至一些修改后仍然存在，并且在需要时可以检索。重要信息，如患者身份、医生签名、诊断规范和用户历史，都可以与图像身份验证码一起加密并编码到水印中。然而，经过解码后的水印图像仍然可以很好地恢复原始图像。因此，这项新技术将使临床医生和研究人员能够有效地保护患者的隐私和图像数据的真实性。在一期项目中，我们开发了医学图像无损感兴趣区域（ROI）水印的方法和软件。我们使用来自细胞遗传学和放射学应用的单色医学图像的几个数据集来评估新方法，基于(1)不可感知性，(2)容量和(3)嵌入水印的鲁棒性标准。测试结果表明，该无损ROI水印方法效果良好，明显超过了第一阶段的可行性标准。在第二期项目中，我们将完成拟议技术的开发。我们将通过优化几个系统参数来改进无损ROI水印。我们将扩展该方法以支持医学图像的无损泛在水印。我们还将在系统中加入加密功能和颜色转换，为患者信息提供额外的安全性，并支持彩色图像水印。我们将评估各种医学成像应用的算法和软件，包括放射学、病理学和细胞遗传学。我们将在临床环境中对新技术进行beta测试，并根据最终用户的输入和反馈进行优化。一旦完全开发和合格，这项新技术将获得专利，并纳入IRIS （ADIR的母公司）未来的体外诊断成像产品。它还将通过与医疗成像系统制造商和一般图像处理和出版软件产品公司的许可协议和合作伙伴关系商业化。最后，我们将寻求DICOM和JPEG2000的采用和标准化，以实现更广泛的应用和传播。该项目开发了创新的软件技术，将为医学成像应用提供目前无法获得的无损水印解决方案。该方法是通过在图像中嵌入难以察觉的水印，使医学图像的使用、存档和通信更加安全和可保护。这些水印经过复制、网络/互联网传输甚至一些修改后仍然存在，并且可以在需要时检索。诸如患者身份、医生签名、诊断规范和用户历史等重要信息都可以与图像身份验证码一起加密并编码到水印中。然而，经过解码后的水印图像仍然可以很好地恢复原始图像。因此，这项新技术将使临床医生和研究人员能够有效地保护患者的隐私和图像数据的完整性。

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.