Privacy-preserving personalized similarity retrieval of large CT image sequences

As one of the most important means of the computer-aided diagnosis, the similarity retrieval of large-scale high-resolution computed tomography image(CI) sequences can effectively assist doctors in diagnosing diseases. In this paper, we present an effective and efficient privacy-preserving Personalized Retrieval method for CISequences based on the radiation model, called the Prcs method. To our knowledge, few studies have been touched on the personalized similarity retrieval of large CI sequence(CIS)s. To better facilitate the Prcs processing, three supporting techniques, i.e., 1) KCI-based similarity measure, 2) perturbation-based privacy preserving scheme, and 3) two indexing schemes, are devised. The experimental evaluation reveals that our proposed Prcs approach is more effective and efficient than the state-of-the-art methods while improving the user experiences.