AN PROFICIENT LS BASED SWITCHED PREDICTOR FOR LOSSLESS CONFINING OF 4-D MEDICAL

Techniques for medical imaging like fMRI, CT, MRI produces large amount of digital data. This paper proposes a context based LS based predictors for lossless compression of such 4-D images. Redundancy inthe form of smoothness and uniform human anatomical structures as well as periodic motion of this structures and presence of high correlation in temporal domain of these 4-D medical image sequences has been exploited. Slope is defined as one of the criteria which predict the level of activity. Based on the estimated slope the current pixel is categorized into one of the seven classification bins. Optimal predictors are assigned to each bin and classification of bin boundaries and estimation of optimal predictors is doneoffline. The proposed method is computationally very simple as it does not require motion estimation which, in general, is a computationally complex process.