The Application of a New Predictive Technique for Lossless Compression of Medical Images

Abstract

A new predictive technique for lossless compression is applied to CT and MR images in order to achieve high compression in the lossless mode. To accomplish this, a prediction value was obtained by assuming the horizontal and vertical components of each pixel. Wide and narrow predictive coding modes were adaptively selected to achieve an accurate prediction. The performance of the predictive technique was compared with the performance of four representative compression techniques in the lossless mode, using different types of CT and MR image sets with different bit rates. Overall, the proposed predictive technique yielded the best compression, particularly with the lower bit-rate images tested. While the compression results are largely dependent on the images used, the results confirm the strong potential of the proposed predictive technique as a strategy for obtaining high lossless compression in medical images.