Random access to compressed volumetric data
Ayberk Ozturk (author) · University of Southern California Digital Library · 2015
Volumetric medical data is a three dimensional view of the object being captured, examples include imaging modalities such as magnetic resonance (MR), computer tomography (CT) and positron emission tomography (PET). For compression of these data, a new 3-D wavelet-based lossless coding scheme, which is an improved version of 2-D PROGRES, is developed. In this compression method, a significance map, which consists of dynamic range values of coefficients of a slice, is designed for reconstructing individual voxels. Significance maps allow any region of interest to be decoded by retrieving only necessary wavelet coefficients. This reduces the number of decoded voxels and the decoding time very significantly. A key novelty of the proposed compression scheme is that it achieves random access decoding of diagonal cuts through the compressed data. This allows flexible display, which can be useful for diagnostic, while not requiring manipulation of very large uncompressed data sets.