Okkalides D
Radiation Physics Laboratory, IPPOKRATEIO General Hospital, Thessaloniki, Greece.
Comput Med Imaging Graph. 1998 Jan-Feb;22(1):25-30. doi: 10.1016/s0895-6111(98)00009-3.
The need for diagnostic image compression of the lossy or irreversible type has been declining due to the rapid increase in commercially available formatted hard disk capacity. It is estimated that the latter has increased about three orders of magnitude in the past 14 years while the size of diagnostic image files has, of course, remained constant. During the same period, despite claims for significantly improved performance by vendors, it seems that only small progress has been made in commercial lossless and lossy compression algorithms. There is still no consensus for lossy compression to a level acceptable for diagnosis. This is mostly considered to be around a ratio of 10:1. However, acceptable compression ratios depend heavily on the type of images processed and may be compared with the 3:1 ratio produced by lossless algorithms. This last value was shown to increase to more than 5.5:1 for gamma-camera images when corrected for the noise content of individual bit planes and for the display capabilities of computer monitors. Therefore, any possible benefits of lossy over lossless compression become questionable when the currently available hard disk capacity and network transmission speed are considered against the inevitable loss of information in the lossy type of compression.
由于商用格式化硬盘容量的迅速增加，对有损或不可逆类型的诊断图像压缩的需求一直在下降。据估计，在过去14年中，后者增加了约三个数量级，而诊断图像文件的大小当然保持不变。在同一时期，尽管供应商声称性能有了显著提高，但在商业无损和有损压缩算法方面似乎只取得了很小的进展。对于有损压缩到可接受的诊断水平仍未达成共识。这一水平大多被认为约为10:1。然而，可接受的压缩率在很大程度上取决于所处理图像的类型，并且可以与无损算法产生的3:1的比率进行比较。当针对各个位平面的噪声含量和计算机显示器的显示能力进行校正时，对于γ相机图像，最后这个值显示增加到超过5.5:1。因此，当考虑到当前可用的硬盘容量和网络传输速度，以及有损压缩中不可避免的信息损失时，有损压缩相对于无损压缩的任何可能优势都变得值得怀疑。
