A Medical Video Coding Scheme with Preserved Diagnostic Quality
| Autor: | Carl J. Debono, Daniel Bonanno |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Lossless compression
Structural similarity Computer science business.industry Data_CODINGANDINFORMATIONTHEORY 02 engineering and technology Superresolution 030218 nuclear medicine & medical imaging 03 medical and health sciences 0302 clinical medicine Diagnostic quality 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Computer vision Artificial intelligence business Coding (social sciences) Data compression |
| Zdroj: | GLOBECOM |
| DOI: | 10.1109/globecom38437.2019.9014083 |
| Popis: | Medical video is increasingly playing a vital role in medicine, with its transmission becoming more mainstream. To support this, better compression techniques which can maintain the diagnostic capabilities of the video are required, especially for transmission over bandwidth-limited channels. This work presents a novel medical video compression system that uses Super Resolution (SR) as a processing step and an Enhancement Layer (EL) to enhance the quality in Regions of Interest (ROIs). This system achieves an average gain of 28.25% in Compression Ratio (CR) over full resolution lossless compression, at a Peak Signal-to-Noise Ratio (PSNR) of 38.17dB, a Structure Similarity (SSIM) of 0.9843 and Multi-Scale Structure Similarity (MS-SSIM) of 0.9991. Moreover, subjective evaluations show that the diagnostic quality is retained. A reduction in encoding time is also registered due to the downscaling of the medical video. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|