Determining optimal medical image compression: psychometric and image distortion analysis
- Alexander C. Flint
- BMC Medical Imaging
PURPOSE Today healthcare policy is based on effectiveness. Diagnostic imaging became a "pacesetter" due to amazing technical developments (e. g. multislice CT), extensive data volumes, and especially the well defined workflow-orientated scenarios on a local and (inter)national level. To make centralized networks sufficient, image data compression has been regarded as the key to a simple and secure solution. In February 2008 specialized working groups of the DRG held a consensus conference. They designed recommended data compression techniques and ratios. MATERIAL AND METHOD The purpose of our paper is an international review of the literature of compression technologies, different imaging procedures (e. g. DR, CT etc.), and targets (abdomen, etc.) and to combine recommendations for compression ratios and techniques with different workflows. The studies were assigned to 4 different levels (0 - 3) according to the evidence. 51 studies were assigned to the highest level 3. RESULTS We recommend a compression factor of 1 : 8 (excluding cranial scans 1:5). For workflow reasons data compression should be based on the modalities (CT, etc.). PACS-based compression is currently possible but fails to maximize workflow benefits. Only the modality-based scenarios achieve all benefits. CONCLUSION Imaging equipment manufacturers are encouraged to improve the compression technology of their imaging devices (e. g. freely selectable compression ratios in the output filter). Double compression should be strictly avoided. Lossless compression formats should be switched off.