Comparative Analysis of Different 2D Predictors for Lossless Compression of Biomedical Images

Hospitals and clinical environments are moving towards computerization and digitization and the volume of digital images generated for medical diagnosis is extremely large, dealing with images of increasing resolution. The increase in resolution requires a growing amount of medical image data to be stored, processed and transmitted through network efficiently. Therefore, compression techniques are indispensible for archival and communication of medical images. Predictive based coding techniques are explored in this paper for better compression and reconstruction as it performs well for lossless compression. This paper gives a comparative analysis of predictor coding efficiency and complexity on medical images of different modalities. It was observed that among different predictors Gradient Edge Detection (GED) predictor gave better results than Median Edge Detector (MED). It also achieves approximately the same root mean square error (RMSE) and entropy as Gradient Adaptive Predictor (GAP) with less complexity.