生物医学信号的近无损压缩

生物医学信号的压缩在临床中具有广泛的应用，但在实际应用中，医生们普遍认为有损压缩技术会使诊断信息丢失，只有无损压缩才能不失真地保留所有信息，因而临床诊断必须使用无损压缩。然而，事实并非如此，首先，在信号的采集和数字化的过程中就会引入量化误差和其他各种误差，即使用无损压缩，实际也是有损的；其次，无损压缩的压缩比很低，其压缩性能远不能与有损压缩相比，所以，在医学领域真正具有广泛应用需求的还是有损压缩技术，关键是如何控制好误差以确保信号的可信度。而一般的有损压缩技术，由于其对信号失真度缺乏严格的控制，因而不适合应用于医学领域，所以有必要研究特殊的有损压缩技术，这就是近无损压缩技术。为此，我们提出了一个能有效表征信号可信度的指标，在此基础上，结合生物医学信号非线性、非平稳性的特点，设计了一个基于上下文的生物医学信号近无损压缩算法，实验结果表明近无损压缩在保证信号可信度的前提下获得了比无损压缩更好的压缩结果。最后，我们还就近无损压缩技术的研究方向作了有益的探讨。

NEAR-LOSSLESS COMPRESSION OF BIOMEDICAL SIGNALS

The compression of biomedical signals is widely needed in clinic. However, many doctors consider lossy compression techniques not applicable because they may cause loss of diagnostic information signals, therefore lossless compression seems to be the only choice. But this is not the case. Firstly, error and noise may be added to the signal during the process of data acquisition, so even compressed with lossless techniques, the whole process is still lossy. Secondly, the performance of lossless compression is limited, in contrast, lossy compression can yield a much higher compression ratio than the lossless compression. Hence, the lossy compression is applicable in biomedical signals, the key issue is how to ensure the high fidelity of signals. Conventional lossy compression methods would not serve this purpose, so we concentrated on a special lossy technique called near lossless compression. We first introduced a good error criterion for fidelity control; then, based on this criterion, developed a low complexity, context based, near lossless compression algorithm of biomedical signals. Experiments showed that our algorithm not only ensured the high fidelity of signals, but also yielded better compression results than lossless compression did. Finally, we also give our insights into the future research on the near lossless compression.