医学图像感兴趣区的内嵌小波编码新方法

提出了一种新颖的适合医学图像的压缩算法AR-EWC(embedded wavelet coding of arbitrary ROI).该算法保证了感兴趣区域边界与背景光滑过渡,支持任意形状ROI的无失真编码,生成内嵌有损到无损图像质量渐进优化的码流,并且可以对ROI独立随机存取.针对医学图像的特点,对包含重要临床诊断信息区域采用完全无失真的无损压缩,而对背景采用高压缩比的有损压缩,既保证了医学图像对高质量的要求,相对于传统的无损方案又大大提高了图像的整体压缩比.在临床头部MR图像数据集上的实验表明,该算法在保证了ROI区域无损压缩的前提下,达到了与经典的有损压缩算法相当的压缩比率.

Novel method of embedded wavelet coding of region of interest for medical images

This paper proposed a novel medical image compression scheme, AR EWC. The key advantages included the following four points: smooth transition between ROI and background without border distortion, the support of arbitrary shape ROI, embedded lossy to lossless progressive refining code stream and random access to any ROI. By using lossless compression on ROI containing clinical details and high ratio lossy compression on background, AR EWC algorithm could achieve the goal of preserving critical clinical information of medical images as well as gains more compression ratio improvement than the traditional lossless scheme. The experiments on MR head images show the algorithm can get compression ratio comparable to classic lossy algorithms while preserving lossless information in ROI.