Owing to factors as diverse as budget constraints, personnel turnover, and the availability of technologies, PACS networks are often made up of equipment from several different vendors. This creates integration problems -- nearly every PACS administrator can tell horror stories about DICOM incompatibility, hardware interface glitches, or communications breakdowns that have plagued even the best multivendor systems.
PACS users, on the other hand, couldn’t care less about an administrator’s headaches; they just want their images on demand. But at least one group of researchers took an analytical approach. When performance issues cropped up in a multivendor PACS at Tokushima University Hospital in Tokushima, Japan, they set out to find the reasons and solve the problems.
In a presentation at the RSNA conference in Chicago last year, Dr. Hiroshi Kondoh described the challenges and opportunities he and his colleagues uncovered during a user-performance evaluation of their multivendor PACS.
The Tokushima PACS serves all the wards in the hospital and is comprised of equipment from three vendors: Hitachi, Toshiba, and NEC. Each vendor’s products are linked to a representative DICOM server, and each server is sub-netted via a router for both network and modality security.
A single DICOM server in the PACS acts as a store-and-forward device for radiological images. This server keeps a database of all images in the system, as well as addresses on a hospital-wide basis, so that physicians requesting an image do not have to access each modality’s DICOM server. In addition, the server has JPEG compression-on-the-fly capabilities, so reference images can be accessed faster than diagnostic images.
The PACS contains images from CR systems from two vendors, one direct radiography (DR) device, one CT, one MR, and one fluoroscopy system. Modalities were added onto the system over time, and software upgrades were installed on an "as-needed" basis.
Although there are many criteria on which PACS performance can be based, the researchers decided to define their study by measuring three user-centric parameters: image display time; monitor luminance, and physician acceptance of the images.
Monitor luminance (the amount of visible light leaving a point on a surface in a given direction) was measured as a value of candelas per square meter (Cd/m2). The group chose monitor luminance as a user-satisfaction determinant because its PACS has several kinds of monitors, each of which has different display characteristics.
Dr. Kondoh and his colleagues found that image display time varied depending on the modality and size of the image. For example, compressed chest images took five seconds to display, while CT images took less than one second. Their investigation revealed that the institution’s network routing structure did not affect display time.
The greatest disparity was seen among monitors from various vendors. Monochrome CRTs in the PACS displayed at 700 Cd/m2, general-purpose SVGA color CRTs displayed at 80 Cd/m2, and color liquid-crystal monitors had a luminance of 150 Cd/m2.
However, the physicians surveyed in the study found the image quality to be acceptable as a reference image on all measured monitors.
The doctors were also uniformly satisfied with the PACS response and display time for images. The PACS is routinely used by all wards in the hospital, and most heavily by physicians in the ICU. The authors noted that ICU staff looked at the images only in default display mode, and did not use the available magnification or image processing look-up tables (LUTs).
Although the researchers rated user satisfaction as "good," for the multivendor PACS, they also reported a clutch of problems with them. For example, they found that each vendor in the PACS had a different database schema. This required registration of new clients to be performed on each DICOM server, instead of being managed from a central server and migrated to the sub-systems.
In addition, some equipment had different JPEG implementations that required all images to be sent uncompressed -- needlessly adding to network congestion and image transmission time.
The Tokushima team is looking forward to integrating the PACS as part of a RIS so the institution can manage its total image throughput. They noted that the multivendor PACS environment makes implementing HIS a challenge -- and that display parameters could be problematic under more demanding conditions.
They also found that continuing education of PACS users is crucial to their adoption of features such as image processing and tone-scale control through window/level adjustments.
By Jonathan S. BatchelorAuntMinnie.com staff writer
January 25, 2001
Related Reading
Baltimore VA survey says human factors count in soft-copy reading, August 29, 2000
Post your comments about this story in our PACS Digital Community. Please include the headline of the article in your message.
Copyright © 2001 AuntMinnie.com
