Quality and performance are paramount when it comes to selecting hardware for displaying radiological images. The sophistication of systems on offer continues to improve, with increased resolution and in-built QA control tools. First-time buyers and departments seeking to upgrade their present display systems can be faced with a difficult choice.
Tender and evaluation
By summer 2005, radiologists at the University Hospital of North Norway, based in Tromsø, were only too aware that their image-viewing monitors needed replacing. Many had been acquired back in 1998, when the hospital’s radiology department purchased its picture archiving and communications system (PACS). Not surprisingly, the quality of images viewed on-screen was beginning to suffer. It had got to the stage where any new system would have been an improvement. So the department initiated a scientific selection process in a bid to find the optimum medical display solution.
Several companies responded to the initial call in June 2005 to make a tender, recalls Yngve Skar, radiographer at the University Hospital of North Norway, and the staff member responsible for QA throughout the radiology department. Some product offerings were clearly inappropriate and could be ruled out almost from the start. The remaining bidders were invited to bring their systems to the hospital for a head-to-head comparison with their competitors.
Vendors attending the group evaluation were asked to demonstrate the same selection of test patterns and sample images on their display systems. Radiologists, radiographers, and IT personnel from the Tromsø site then had the chance to compare like with like.
Test patterns were acquired from the American Association of Physicists in Medicine (AAPM) Task Group 18 (TG18), a task force of medical imaging experts and organisational affiliates dedicated to QA evaluation of electronic display devices. The AAPM TG18 provides test patterns as free downloads from its website for initial acceptance testing and ongoing quality control. These patterns are consequently readily available to vendors and prospective purchasers alike.
Sample images were taken from the University Hospital of North Norway’s existing digital archive of CT, MRI and X-ray images. The selection was chosen to represent the wide spectrum of cases that radiologists at the busy 625-bed university hospital might be asked to interpret. The hospital in Tromsø provides specialised medical services for the whole of northern Norway, covering a population of 465,000. It additionally serves as a local hospital for the 130,000 inhabitants of the Tromsø area. Radiologists can consequently expect to see a diverse array of injuries, diseases, and infections on a weekly basis.
Several of the department’s radiologists had been very keen to sit in on the evaluation, Skar says. However, their comments on image quality tended to be quite subjective. The radiologists’ observations often related simply to how a particular case looked on a particular monitor. While these observations were no doubt salient, an overview of the systems’ performance in relation to one another was required for a purchasing decision to be made.
To produce a more objective assessment, Skar rated each display in terms of the more general parameters that make the difference to image-viewing quality; contrast, resolution, and brightness. This assessment was a little more complex than a straightforward 1–5 scoring system, he notes. Yet still a clear winner could not be found. Two or three vendors were deemed to have passed the test almost equally. “They were so close that it was difficult to say that one was better than the others,” he says.
The final selection, then, came down to two additional criteria; price and functionality. Having identified a number of systems that would display medical images with sufficient clarity for radiological diagnosis, attention turned to the questions: “What can the department get for its money?” and “What added benefits do each of these systems bring?”
A decision was eventually taken to place an order with NEC Display Solutions for 22 dual-monitor workstations (44 screens). This comprised 38 19-inch colour monitors (MDView19) with a resolution of 1MP, and six 21-inch monochrome displays (MD21) with a resolution of 2MP.
The decision to purchase three higher-resolution monochrome displays was made with QA in mind. These 2MP workstations are generally reserved for more specialist musculoskeletal applications, where the extra clarity can be particularly helpful. “This was something that I specifically wanted. I wanted to have some displays with better resolution,” Skar says. “The radiologists wanted to have colour displays, but I think they like the resolution of the monochrome displays now that they see it.”
The six paired MD21 displays each have NEC’s X-Light technology. This controls and adjusts luminance and white point levels via an integrated feedback sensor throughout the life of the monitor. In addition to reducing the need for calibration checks, the X-Light technology is designed to alleviate backlight colour shift to the yellow spectrum. Other features incorporated with the MD21 displays include fast-switch gamma curves, black-level adjustment, ultrathin frame design (bezel width 16mm) and an internal power supply. A 10-bit video card provides 1,024 out of a possible 3,061 grey shades, while an 8-bit card offers 256 shades of grey.
Both types of display system come complete with a number of user-friendly features. The display screens are made from “Superfine TFT” glass, which offer a wide viewing angle of 176° and minimal off-angle colour shift. Use of round resin beads in the “Superfine TFT” inherently reduces glare. Reflections are kept to a minimum by the addition of a low-reflection coating. The screens can be positioned in portrait or landscape mode, depending on users’ preference, and tilted to provide an optimum viewing angle. “This makes it more comfortable for whoever is looking at the images,” Skar notes. These ergonomic aspects were one factor that helped clinch the deal for NEC.
Installation of the colour and monochrome workstations began in September 2005. This was very much a “plug-and-play” procedure, and little – if any – technical backup was required. However, communication channels with the supplier and distributor remained open throughout installation in case troubleshooting was required. “If we had anything to ask, it wasn’t a problem,” Skar says. “We had good communications all the way.”
Two of the dual-head workstations were installed in radiologists’ offices. The remainder were set up in communal reporting areas for easy access by the department’s radiology staff. “The radiologists started using them straight away,” Skar says. “They had been working with displays for several years, so soft-copy reading was nothing new to them. But of course the displays were better, so the quality of images they were viewing improved.”
Should problems arise with the new display systems, radiologists should suffer less inconvenience, Skar says. Because NEC monitors have built-in, standalone calibration tools, paired monitors can be swapped in and out, and recalibrated and rematched whenever required. This means that if one half of a dual-head display system fails, it is not necessary to return the entire workstation. “Some vendors pair their displays so if there is a fault with one, you have to change both of them. With NEC, if there is anything wrong with one display, we just change that one and calibrate it again,” he says.
The display systems have also been loaded with NEC’s GammaComp MD calibration and conformance software. This QA tool ensures that individual monitors conform to current DICOM standards. It also offers the option of centralising QA management of multiple display systems. “The networking software makes it possible to sit at one workstation and get information from the displays, wherever they are, instead of going to a workstation and collecting information there. It is a more centralised way to work,” Skar says.
The radiology department at the University Hospital of North Norway is currently the largest networked QA installation for NEC in Europe. As such, NEC regards the Tromsø hospital as somewhat of a pioneer site. Meanwhile, staff members at the Tromsø hospital are only just beginning to appreciate the full capabilities of the QA software. The full functionality of the GammaCompMD Administrator Suite has yet to be exploited, Skar says. He plans to instigate a monthly checkup for each and every display system in the near future.
Prior to the software setup, completed in November 2005, NEC personnel briefed medical technicians, IT staff, and QA management about its use. They additionally proffered advice on the importance of QA policy. This one-off session included information on the AAPM TG18 guidelines for QA of medical imaging displays. For instance, the AAPM TG18 has detailed the tools and procedures required for assessing parameters such as geometric distortions, reflection, luminance response, luminance dependencies and resolution, to ensure they meet defined criteria.
Skar now intends to pass the QA message on to radiologists himself. While an “external voice” can help reinforce the need for QA control, doctors are busy and may resent having to attend such briefings. Inculcation of QA principles probably requires an in-house approach, which can be pursued over a long time period, he says. “This is not something that can be done in a day. It will take a while longer to get the message across.”