A Hybrid Lossless Compression Technique with Segmentation and Modified Arithmetic Coding
Mazhar Basyouni Tayel, Mohamed A. Abdou · 2006
Abstract:- Compression of medical images (MI) is an important field of study in biomedical engineering. Although lossy compression could solve storage space and transmission bandwidth problems, it is not recommended by most physicians because of data loss. Lossless compression saves all details inside image; however it could not be applied for the whole image area. This paper introduces a hybrid lossless compression channel with two different steps. The first is an automatic segmentation technique, where a region of interest (ROI) is automatically segmented by aid of an artificial neural network (ANN) and an introduced difference fuzzy model (IDFM). The second is a modified arithmetic coding (MAC) lossless compression algorithm. This hybrid channel is to combine in parallel with a lossy compression channel that transmits non important parts of the MI progressively, using the fast algorithm of embedded zerotree wavelet (FEZW) [1]. The proposed technique reduces complexity, storage space, bandwidth, and saves time. Moreover, it is a fully automatic system. Several brain magnetic resonance images (MRI) and fluorescene ophthalmic images are analyzed.