Lossless ROI Watermarking of Medical Images

描述(申请人提供)：该项目的目标是开发创新的无损感兴趣区(ROI)水印技术，使医学图像的使用、归档和通信更加安全和易于管理。目前可用的技术仅提供了不适用于大多数医学图像的无处不在的水印手段。该技术通过在不改变感兴趣区域内容的情况下将不可感知的水印嵌入到图像中，为数据认证、安全和版权保护提供了一种有效的手段。这些水印在复制、互联网传输或甚至某些更改后仍然存在，并且可以在需要时检索。重要的信息，如标本标识、实验室签名和规格可以编码到水印中。这一功能将使临床医生和研究人员能够有效地保护其数字图像数据的真实性和完整性。 在第一阶段中，我们将研究在简单通道医学图像上进行无损ROI水印的可行性。我们将开发一个原型系统来实现所提出的水印方法。我们将开发一种基于ROI描述的图像分割方法，该方法处理并跟踪图像区域的划分和位置。结合区域分割，我们将实现空域和小波域的水印方案，并进行改进，以实现对图像中非ROI部分的选择性水印，并使用单帧单色医学图像进行测试。我们将从(1)不可见性、(2)容量和(3)稳健性三个方面对这些方案进行评估。如果建议的技术被证明是可行的，在第二阶段，该技术将扩展到复杂模式医学图像并进行测试。在完全开发后，我们将为这项新技术申请专利，并通过与医疗成像系统制造商以及通用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将其商业化。

DESCRIPTION (provided by applicant): The goal of this project is to develop innovative lossless regions-of-interest (ROI) watermarking techniques to make the usage, archiving, and communication of medical images much more secure and manageable. Currently available techniques only provide means of ubiquitous watermarking that are inapplicable to most medical images. The proposed technology offers an effective means of data authentication, security, and copyright protection by embedding imperceptible watermarks into the images without modifying the content of the ROI. These watermarks remain present after copying, internet transmission, or even some alteration, and can be retrieved when needed. Important information such as specimen identification, laboratory signature and specification can be encoded into the watermarks. This functionality will enable clinicians and researchers to effectively protect the authenticity and integrity of their digital image data. In Phase 1 we will study the feasibility of lossless ROI watermarking on simple modality medical images. We will develop a prototype system to implement the proposed watermarking approach. We will develop an image segmentation method that processes and keeps track of the partition and location of the image regions based on the ROI delineation. In conjunction with the region segmentation, we will implement both a spatial-domain and a wavelet-domain watermarking schemes with the improvement to enable selective watermarking into non-ROI parts of the images, and to test them using single-frame monochrome medical images. We will evaluate these schemes in terms of watermark (1) imperceptibility, (2) capacity, and (3) robustness. If the proposed technique proves to be viable, in Phase 2 the technology will be extended to and tested on complex modality medical images. When fully developed, we will patent the new technology and commercialize it through licensing agreements and partnerships with manufacturers of medical imaging systems and companies of general image processing and publishing software products.