SEATTLE—Healthcare costs will not decrease unless radiologists and PACS users stop shuffling paper, stop shuffling film and the information exchange is made electronic between hospitals and regions, according to a presentation during the 2008 Society of Imaging Informatics in Medicine (SIIM) pre-conference symposium held Wednesday for the preparation for Imaging Informatics Professionals (IIP) symposium.
Salvadore Tejada, RT(R)(CT)(MR), director of informatics at Radiology & Imaging Specialists in Lakeland, Fla., divided his IT lecture into six categories: storage and archiving; networking; hardware/software; databases; standards; and hardware replacement.
Of the storage architectures, he explained four:
- Redundant array of independent disks (RAID), which uses multiple hard drives to share or replicate data and can be implemented in different levels with varying degrees of fault tolerance.
- Direct-attached storage (DAS), a system that directly attaches to a server or workstation and is not directly accessible to another one. However, Tejada said that with DAS, “if you attach the storage device to computer A, this device can not be accessed from computer B. With DAS, you are not able to access the storage in another computer,” even if the computers are on the same network.
- Network-attached storage (NAS), dedicated storage connected directly to the network, which does not need a server to access data. Tejada noted that the “file-based storage approach is ideal for file servers.”
- Storage area network (SAN), dedicated storage network that transfers data only between multiple storage devices and file servers, keeping this traffic on a separate high-speed network. Tejada said that the “block-based storage approach ideal for databases.”
In comparing file-based storage to the block-based approach, Tejada said that “if you’re retrieving a saved document or image file, you want to retrieve the entire file.” He also said that “if you’re retrieving information from a database, you want to retrieve just the block of information you’re interested in—not the entire database file.”
Tejada said that if a computer is going to be connected to a network, the administrator needs a network interface card (NIC), a device, which can be wireless or cabled, that is installed on a computer and each one has a unique permanent number—MAC or physical address. The NIC can be assigned an IP or logical address as well, according to Tejada. “The physical address cannot be changed, but the IP address can be changed,” he said.
He listed the typical PACS hardware as storage/archive server; web server; image acquisition gateway; diagnostic workstations; integration servers; and a database server. He also listed the components of PACS software, which he classified as: archive manager; database management system; web server; image viewer; and an interface engine.
Tejada presented several database models. The first of which were relational databases, which provides for linking from any table to another. Some examples of relational databases are Oracle, MS Structured Query Language (SQL), MS Access and MySQL. The second database model was an object database, which links objects and object classes at high speeds. He said object database are “good for individual objects rather than groups.” The final database model that he presented was hierarchical, which he defined as an “upside-down tree structure,” because “every table except root has a single parent table.”
“The concept of database normalization ensures that the structure of the database is as efficient as possible,” Tejada said. It seeks to eliminate redundancy and ensure that data dependencies make sense, he said.
Tejada concluded by discussing the hardware lifecycle. “Depending on your source, your hardware should be replaced every three to five years,” he said. However, he noted that a complete overhaul does not have be done “all at once, but instead, can be staged.” However, he did suggest that PACS administrators should plan to replace 20-25 percent of their workstations every fiscal year.