부산대학교 도서관

With recent development of tele-diagnosis and tele-medicine, the amount of medical images transferred through the network has been tremendously increased. These medical images are usually in DICOM (Digital Imaging and Communications in Medicine) format. The need of solutions for authentication and confidentiality of such data is critical. This paper proposes a zero-watermarking approach of DICOM images authentication and identification. The novelty of the proposed approach relies on two main points. The first one is the selection approach of image features, called pertinent features, which are used for identification. These features are selected from a large set of image characteristics, using a statistical analysis approach that aims to select the minimal discriminant set of features with the highest resistance against existing attacks. The second one is the key building process based on the combination of information extracted from the header of the DICOM images, pertinent features and a Jacobian model. This key is the one sent to the receiver for the authentication of the medical image. The experiment results show that the proposed approach has very good performance in terms of authentication and identification of medical images and satisfies timing-constraints of medical applications such as telemedicine. [ABSTRACT FROM AUTHOR]