The recent advancement in the research fields of medical imaging and machine learning technologies greatly im- prove the accuracy of the content-based medical image retrie- val(Cbmir). Existing Cbmirmethods, however, often overlooks the effect of image details on retrieval accuracy, which is usually unsatisfactory. The Cbmirof high-definition CT image (CI) based on deep learning technology, combined with patient symptom information, can precisely assist physicians in disease diagnosis and treatment. To address this issue, we take liver CI as an example in this article, and propose a privacy-preservingComputer-AidedDiagnosis(Cad) method of the liver diseases based on the multi-modal clinical information (e.g., patients’ CIs and symptoms, etc). Four enabling techniques such as aWsslnmodel, a unified similarity measure of CIs, a DM-Treeindex and a lightweight privacy-preserving(LPP) scheme are pre- sented to facilitate theCadprocessing of the large liver CIs. Ex- tensive experiments demonstrate that our proposedCadme- thod outperform the state-of-the-art schemes by a large margin in terms of the retrieval accuracy and efficiency, respectively.

This article addresses a multi-query optimization problem for distributed medical image retrieval in mobile wireless networks by exploiting the dependencies in the derivation of a retrieval evaluation plan. To the best of our knowledge, this is the first work investigating batch medical image retrieval (BMIR) processing in a mobile wireless network environment. Four steps are incorporated in our BMIR algorithm. First, when a number of retrieval requests (i.e.,mretrieval images andmradii) are simultaneously submitted by users, then a cost-based dynamic retrieval (CDRS) scheduling procedure is invoked to efficiently and effectively identify the correlation among the retrieval spheres (requests) based on a cost model. Next, an index-based image set reduction (ISR) is performed at the execution-node level in parallel. Then, a refinement processing of the candidate images is conducted to get the answers. Finally, at the transmission-node level, the corresponding image fragment (IF) replicas are chosen based on an adaptive multi-resolution (AMR)-based IF replicas selection scheme, and transmitted to the user-node level by a priority-based IF replicas transmission (PIFT) scheme. The experimental results validate the efficiency and effectiveness of the algorithm in minimizing the response time and increasing the parallelism of I/O and CPU.