Reversible compression of medical images using decomposition and decorrelation methods

A new reversible image compression technique is presented for compressing and reconstructing various types of medical images. This technique involves a decomposition method to split the image into the most, intermediate, and the least significant bit images in order to reduce the noise effect, followed by decorrelation methods which involve differential pulse coded modulation (DPCM) and a novel interpolative technique. Three types of medical images including 60 MR, 40 CT, and 5 chest X-ray images are considered in the experiment. Depending upon the image quality and characteristics, this error-free compression technique has improved the compression efficiency by approximate 30-50% over the conventional DPCM/Huffman coding. The practical compression ratios vary from 3.2:1 to 4:1 and the true storage saving ratio reduces up to 5.35:1.<<ETX>>

[1]  Abraham Lempel,et al.  Compression of individual sequences via variable-rate coding , 1978, IEEE Trans. Inf. Theory.

[2]  M A Viergever,et al.  Reversible interframe compression of medical images: a comparison of decorrelation methods. , 1991, IEEE transactions on medical imaging.

[3]  Tenkasi V. Ramabadran,et al.  The use of contextual information in the reversible compression of medical images , 1992, IEEE Trans. Medical Imaging.

[4]  S C Lo,et al.  A method for splitting digital value in radiological image compression. , 1991, Medical physics.

[5]  Abraham Lempel,et al.  Compression of two-dimensional data , 1986, IEEE Trans. Inf. Theory.

[6]  Max A. Viergever,et al.  Reversible image data compression based on HINT (hierarchical interpolation) decorrelation and arithmetic coding , 1991, Medical Imaging.

[7]  Ian H. Witten,et al.  Arithmetic coding for data compression , 1987, CACM.

[8]  Ian H. Witten,et al.  Text Compression , 1990, 125 Problems in Text Algorithms.

[9]  Nassrin Tavakoli,et al.  Lossless compression of medical images , 1991, [1991] Computer-Based Medical Systems@m_Proceedings of the Fourth Annual IEEE Symposium.