ABSTRACT
BACKGROUND: Efficient handling of both picture archiving and retrieval is a crucial factor when new PACS installations as well as technical upgrades are planned. MATERIALS AND METHODS: For a large PACS installation for 200 actual studies, the number, modality,and body region of available priors were evaluated. In addition, image access time of 100 CT studies from hard disk (RAID), magneto-optic disk (MOD), and tape archives (TAPE) were accessed. RESULTS: For current examinations priors existed in 61.1% with an averaged quantity of 7.7 studies. Thereof 56.3% were within 0-3 months, 84.9% within 12 months, 91.7% within 24 months, and 96.2% within 36 months. On average, access to images from the hard disk cache was more than 100 times faster then from MOD or TAPE. CONCLUSION: Since only PACS RAID provides online image access, at least current imaging of the past 12 months should be available from cache. An accurate prefetching mechanism facilitates effective use of the expensive online cache area. For that, however, close interaction of PACS, RIS, and KIS is an indispensable prerequisite.
Subject(s)
Database Management Systems , Information Dissemination/methods , Information Storage and Retrieval/methods , Medical Records Systems, Computerized , Radiographic Image Enhancement/methods , Radiology Information Systems , Signal Processing, Computer-Assisted , Databases, Factual , Delivery of Health Care, Integrated/methods , Delivery of Health Care, Integrated/organization & administration , Germany , Internet , Radiographic Image Interpretation, Computer-Assisted/methods , Systems Integration , Technology Assessment, Biomedical , User-Computer InterfaceABSTRACT
PURPOSE: Recommendations for archiving digital radiological image data based on the comparison of retrieval times for different PACS archive levels. MATERIALS AND METHODS: For a large PACS installation (Agfa Impax, Release 4.1), image retrieval times for radiological standard examinations (chest radiographs with 2, MRI with 250, CT with 100 and 1000 images; n = 120, each) from hard disk array, magneto-optical disk (MOD), and magnetic tape archives (TAPE) were examined in high and low network traffic load. RESULTS: Even large CT examinations (1000 images) were available from hard disk arrays within 4.0 +/- 0.8 s, smaller studies within 1.8 +/- 0.3 s. Radiographic image retrieval from MOD (30 +/- 4.7 s) was more then 50 % faster than from TAPE. For typical cross-sectional studies, the velocity gain amounted to 19 %. For both technologies, no significant difference was found for large CT examinations (651 +/- 144 s). For high and low network traffic load scenarios, image retrieval times from hard disk, MOD, and TAPE archives increased by 87 %, 7 %, and 22 %, respectively. CONCLUSION: Hard disk arrays are specifically suited as departmental intermediate storage media because they allow fast access to current and previous examinations within a short time. Performance properties enable both MOD and TAPE systems to serve as long-term archives. However, MOD archives are less flexible in the expansion of storage capacity and at present the medium costs per memory unit are about 2 - 3 times higher than for tape archives. The use of existing MOD-archives may be adequate as intermediate archives. For new PACS installations or system expansions, however, it is recommended to combine a sufficiently large local data memory (RAID) with data storage on tape archives outside the radiological unit that can be used by other departments as well. Future development of hard disk prices will show whether archiving for the whole data retention period may be handled by RAID systems. In any case, prefetching problems and waiting periods for demanded pre-studies would not occur any more.
