Dr. Kyung Won Lee and colleagues moved the thin-slice volumetric data set from 16-slice CT scanners to a miniPACS with 271-GB online and 680-GB nearline storage capacities. A thicker slice data set was stored in the main PACS.

The researchers studied a two-week period to determine the impact of storage needs of each data set type: volumetric, thick axial, standardized 3D images routinely produced by technologists, 3D images added by radiologists, and scan planning. They also analyzed the storage need of each PACS over a five-month period.

For the 867 CT exams performed during the two-week period, the percentage of data volumes for volumetric, thick axial, standardized 3D, additional 3D, and scan planning data sets decreased linearly: 74.4%, 15.9%, 7%, 2.3%, and 0.5%, respectively.

Over the five-month period, 278 GB of CT data (8976 exams) were stored in the main PACS, and 738 GB of volumetric data sets (6193 exams) were stored in the miniPACS. The volumetric data sets formed 33% of total data for all modalities (2.2 TB) in both the main PACS and the miniPACS. At the end of this period, volumetric data sets of 1892 and 5162 exams were kept online and nearline, respectively.

"Using a miniPACS is an effective method, at least during any transitional period until a more ideal system becomes commercially available and affordable," Lee said.

In another study, Dr. Stefan Wirth and colleagues at Ludwig-Maximilians University in Munich determined that the best workflow requires access to every relevant prior from a fast PACS archive cache. Eighty-five percent of priors in this study were accessed within a year; nearly all within four years.

Researchers included 1200 consecutive radiography, CT, and MR studies of inpatients and outpatients. They accessed all priors and analyzed age, modality, and body region, as well as relevance for the interpretation of the current examination. Priors existed in 60% of cases, and the mean number of prior exams was nearly eight. The mean age of relevant priors was 203 days, distributed as follows: 56% within three months, 72% within six months, 85% within 12 months, 91% within 24 months, and 99.5% within 60 months.

While this research demonstrated a minimum need for a cache size covering actual images of at least 12 months, the falling price of hard disks will make prefetching superfluous, Wirth said.