Interframe Compression Of Medical Images

Abstract

Today the necessity to compress data in order to improve archiving and transmission of images is a widely held opinion in the medical imaging field. Medical image sequence compression is a real challenge which has not yet received much attention even if the use of 3D orthogonal transforms to achieve interframe compression was considered. We used Factor Analysis of Dynamic Structures (FADS) which allows to process dynamic image studies (nuclear medicine, CT or MRI) to estimate underlying physiological functions and simultaneously to compress data as a first step of an image sequence compression. To compress 3D spatial sequences, the proposed method is based on Correspondence Analysis (CA) of an array obtained after dividing the 3D initial data into cubic subarrays. To improve compression, an adaptive coding is applied to the results obtained after the factor or the correspondence analysis. It achieved high compression ratios as high as 2O:l to 100:l.