| Summary: | As computed tomography image (<italic>CTI</italic>) sequence can be regarded as a medical image set, which contains several neighboring and visually similar <italic>CTI</italic> s with temporal information, conducting an efficient similarity retrieval of the large <italic>CTI</italic> sequences poses great challenge. Finding the similar <italic>CTI</italic> sequences can assist diagnosis and treatment by allowing physicians to quickly locate pathological images of lesion tissues, accurately find the source of disease. The paper proposes a progressive <italic>P</italic>rivacy-preserving similarity <italic>R</italic>etrieval scheme for the <italic>CTI</italic> <italic>S</italic>equences in the mobile telemedicine network (<italic>MTN</italic>) called the <italic>P RS</italic> method. To better facilitate the <italic>P RS</italic> processing, five enabling techniques are devised: 1) <italic>key CTI (KCTI)-based similarity measure</italic>, 2) <italic>local privacy preserving scheme</italic>, 3) <italic>zero-distance</italic> (<italic>ZD)-based distributed storage scheme</italic>, 4) <italic>uniform distributed index framework</italic>, and <italic>5</italic>) <italic>crowd-assisted verification</italic>. Extensive experiments demonstrate that the retrieval efficiency of our proposed <italic>P RS</italic> method is about 90% higher than that of the existing ones.
|
|---|