Lossless region-of-interest coding for teleradiology with application to brain MRI

A compromise between the conflicting requirements of high compression efficiency and high image quality in medical image compression for teleradiology is selective lossless image compression (SeLIC) where only diagnostically important Region of Interest (RoI) is encoded losslessly. In this paper we present a fully automatic SeLIC system for brain MRI data. We also present a novel lossless image compression technique to encode the RoI. In compressing 8-bit T1-weighted brain MRI volumes from Internet Brain Segmentation Repository (IBSR), the proposed lossless scheme outperformed JPEG-LS in 15 out of 20 volumes and the proposed automatic SeLIC system achieved more than 16:1 compression ratio on average without incurring any loss in diagnostically important area