English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Remote medical diagnosis has emerged as a critical and indispensabletechnique in practical medical systems, where medical data are required to beefficiently compressed and transmitted for diagnosis by either professionaldoctors or intelligent diagnosis devices. In this process, a large amount ofredundant content irrelevant to the diagnosis is subjected to high-fidelitycoding, leading to unnecessary transmission costs. To mitigate this, we proposediagnosis-oriented medical image compression, a special semantic compressiontask designed for medical scenarios, targeting to reduce the compression costwithout compromising the diagnosis accuracy. However, collecting sufficientmedical data to optimize such a compression system is significantly expensiveand challenging due to privacy issues and the lack of professional annotation.In this study, we propose DMIC, the first efficient transfer learning-basedcodec, for diagnosis-oriented medical image compression, which can beeffectively optimized with only few-shot annotated medical examples, by reusingthe knowledge in the existing reinforcement learning-based task-driven semanticcoding framework, i.e., HRLVSC [1]. Concretely, we focus on tuning only thepartial parameters of the policy network for bit allocation within HRLVSC,which enables it to adapt to the medical images. In this work, we validate ourDMIC with the typical medical task, Coronary Artery Segmentation. Extensiveexperiments have demonstrated that our DMIC can achieve 47.594%BD-Rate savingscompared to the HEVC anchor, by tuning only the A2C module (2.7% parameters) ofthe policy network with only 1 medical sample.

Diagnosis-oriented Medical Image Compression with Efficient Transfer Learning