An Automatic Technique for Region- Based Coding of Medical Images

Selective compression of a medical image (MI) becomes an important task in fields of diagnosis, in solving archiving, and transmission problems. In this paper, a proposed bichannel automatic compression technique for medical image processing (MIP) with ROI is introduced using two approaches. The first approach uses two different compression schemes (lossless and lossy) for inside and outside of the ROI. In the second approach only lossy compression is applied progressively using fast and reduced bit embedded zerotree wavelet algorithm (FEZW) [1] with a higher refinement level inside the ROI. Both approaches use the introduced automatic neurodifference fuzzy region of interest (ROI) segmentation technique [2]. Several brain MRI and fluorescene ophthalmic images (FOI) are processed; results are studied and analyzed. The obtained results of the two approaches are presented and compared. Key-words: Selective Image Compression, Hybrid lossless channel, Fast Embedded zero tree wavelet, Modified Arithmetic Coding, Progressive Transmission.