Implementation and evaluation of three compression methods for diagnostic images

Abstract

In radiology, as a result of the increased utilization of digital imagingmodalities, such as computed tomography (CT),ultrasonography and magneticresonance imaging (MRI), over a third of the images produced in a typicalradiology department are currently in digital form, and this percentage issteadily increasing. Image compression provides a means for the economicalstorage and efficient transmission of these diagnostic pictures.The aim of this thesis is to present three major data compressionalgorithms and implement them for radiological images on a PC, using the MSDOS (Version 3.3) operating system and the Turbo PASCAL (Version 5.0).The original and reproduced images have been displayed on a TV monitor byusing a graphics display card.Huffman coding and Run-length coding algorithms are discussed aserror-free compression techniques. Huffman coding algorithm is based on anoptimal, variable length code word design method. Run-length coding is basedon the repeatibility of adjacent pixels in an image data. By using thesealgorithms, compression ratios of z 2: 1 have been achieved with 64 KBytes,256 gray level diagnostic images. At the end of decoding process, perfect imagereconstruction has been obtained.An adaptive fast discrete cosine transform coding system is alsointroduced, yielding compression ratios in the range from 4:1 to 16:1. At theend of decoding process, some degradation has been occurred in thereproduced images, depending on the compression ratio and the number ofquantization levels used.