A Novel Adaptive Lossless Compression Algorithm for Efficient Medical Image Archiving and Transmission

Most medical diagnostic tools generate images that are digitally collected, stored and transmitted. In order to perform these operations efficiently, the images are mathematically compressed before storage or transmission and decompressed after retrieval. Methods which, after decompression, result in a reconstruction that is identical to the original image are referred to as lossless; those whose reconstruction is an approximation of the original image are referred to as lossy. Though lossy methods produce compression factors 20 times greater than lossless methods, in medical applications the need to conserve the diagnostic validity of the image requires the use of lossless compression methods. A medical image is always composed of two summed terms: the first term is the image with its useful structures and information; the second term is useless noise due to the actual experimental measurements. A compression method which is able to compress and reconstruct a medical image by rejecting the information due exclusively to noise has to be considered a lossless method because it conserves all useful information. In this context, a novel lossless compression algorithm, which almost eliminates both redundant information and noise from a medical image while retaining all relevant structures, is proposed. The algorithm is based on the evaluation of the information content of a given image through the calculation of a function that is formally identical to the entropy defined in thermodynamics. The entropy is not calculated directly on the original image but on its projections. Reconstruction from projections is one of the most important acquisition/reconstruction methods used in tomography and projections are the signals that a diagnostic tool actually measures; the image is always reconstructed mathematically, not directly measured. The advantages of this method with respect to other compression algorithms are threefold: first, it compresses the image while eliminating noise, i.e. useless information; second, the resulting compressed image can be elaborated with another standard lossless compression algorithm, further improving the compression ratio; third, in medical tomography it can be applied directly to the measured signals, the projections. Application of the presented algorithm to experimental nuclear magnetic resonance imaging (NMRI) data demonstrates its good performance; the achieved compression factor was about 21.