Applied Radiology

Grayscale workstations are an essential element of any PACS or teleradiologysystem. Gray-scale monitors are the electronic display screens of any PACStermed soft-copy display used in these workstations. Image fidelity forgrayscale monitors is measured by the physical characteristics of luminance,dynamic range, distortion, resolution, and noise. Psychophysical techniques,including receiver-operator characteristics analysis using clinical images andtesting with phantom contrast-detail patterns to determine threshold contrast,are used to measure the user's response to the display of images.

Display protocols are the sequence in which images are presented to the useron the grayscale monitors. The two technologies used for displaying digitalclinical images are grayscale workstations and laser film printers.

Grayscale workstations with cathode ray tubes (CRTs) are used in the displayof digital imaging data. Images are initially displayed on a set of CRTs forthe proper selection of desired information; laser film printers are used togenerate the films from which the primary diagnosis is made and to archivethese films into the patient's film jacket.

Research continues on the issue of high-resolution grayscale displays forprimary diagnosis, particularly for the format used in chest radiography (14¥ 17 inch) or full breast digital mammography (8 ¥ 10 inch, 10 ¥12 inch, 4k ¥ 5k ¥ 16 bits). Studies have compared reader responsesfor conventional analog screen-film radiographs, laser printed digital images,and soft-copy readings. Results of such studies indicate that readers find thesystems nearly equivalent, but some users continue to favor films displayed ona view box. Others favor reading from grayscale workstations. Currently, thetrend is towards soft-copy radiology reading due to the advantages ofinteractive grayscale workstation retrieval, display, manipulation, archiving,and rule-driven workflow engines.

The grayscale monitor has a number of deficiencies compared to filmdisplayed on a view box, however. The luminance of the CRT display screen isless than that of the view box by a factor of 10, and the phosphor granularityof a grayscale display limits the delectability of variations in contrast inthe displayed image. Additionally, a grayscale monitor is limited toapproximately 2,400 scan lines because of its electron gun and deflectioncircuits, and the bandwidth of the video amplifiers. Finally, the dynamic rangeof a grayscale display is narrower than the optical density range of the filmdisplayed on a viewer box. The light scattered in the glass faceplate of thegrayscale monitor, a phenomenon known as veiling glare, limits the dynamicrange and modulation transfer function of the monitor.

Independent of these limitations, grayscale monitors are used in PACS; theseare called soft-copy readings. These monitors can be used for reading of CT,MRI, digital fluoroscopy, computed radiography, ultrasound, and nuclearmedicine images.

The single largest concern of radiologists in the use of grayscaleworkstations is the lack of user-friendly display protocols and how they fitinto the workflow of the department. To combat this, vendors and theirengineering staffs have focused on providing workstations with features theybelieve are essential for basic reviews of imaging for a variety of imagingmodalities. Such capabilities include: 1) placement of patients' examinationsin all modalities into a single folder; 2) patient selection mechanisms(allowing the ability to select data, the entire patient folder, specificpatient exams, a specific exam series, or multiple folders for examinations andseries of examinations); (3) study information displays (to display studyinformation and provide access to detailed information); and (4) results reportdisplays (to display transcribed result reports thatare associated with aparticular patient exam. While the radiologist has the choice of the spectrumof these display functions, all are not needed for each examination reading.Rather, the need for throughput and a few of these functions, carefully chosen,is the key to the radiologist's desire for a friendly and rapid displayprotocol.

Modeling of throughput

To avoid the radiologist being disappointed in the grayscale workstation'sthroughput and integration into the department's workflow, it is necessary toconstruct models that will predict how these parameters will behave. One suchmodel that has been developed is called the resource utilization analysis andit is used to determine how the grayscale workstations will fit into theworkflow. table 1 (a typical CT laser film reading protocol) and table 2 (atypical PACS CT grayscale workstation reading) illustrate a resourceutilization model. This model is simple to construct. It will identify anypossible bottleneck in a series of steps that could compromise a complete job.To construct such a table, first select the resources to be modeled (identifiedby the columns labeled tech, modality, printer, resident, radiologist and filmroom personnel). The first column identifies the steps (need not be sequential)to accomplish one complete reading, and the last column identifies the averagetime required to complete each identified step (one could also record thevariance about the average value). Then for each step in the table, theresources used in that step are indicated by a "l" (being used) and a"0" (not being used). For example, the first step "patientexam" uses the resources of the "tech" and the"modality" for a total of 25.653 minutes. Then, from this table thethroughput per minute for each resource is calculated. This is accomplished byassuming that a resource is 100% utilized and hence the maximum throughput perminute for the "tech" is 1/(25.653 + 6.520) or 0.031 jobs/minute. Thesmallest throughput of the resources is the bottleneck (0.015 job/minute forthe "film room personnel." Table 2 is calculated in the same mannerso that the bottleneck is the "tech" (0.031 jobs/minute).

This modeling method is easy to achieve and determines the bottleneck of theresources for integrating the workflow into the PACS and comparing thesoft-copy to the film reading (table 1). The "disruption" step is theamount of time that results from phone calls and other interruptions to thereader.

This process of modeling the workflow allows a method for seeking answers to"what if" questions. For example, in table 1, the question of addinga second technologist would increase throughput and shift the bottleneck to themodality. One could also compute the "cost-per-throughput" to allowfor the cost of resources and the impact on throughput.

State modeling

An alternative means of modeling is with a state diagram (figure 1). Inthis, the states of the model (workstation, window level, sign in, etc.) andthe transactions between each state are identified. The steps used in thismodel are as follows: l) power up; 2) sign in; 3) call up work list; 4) selectcase #l; 5) select display protocol; 6) send case elsewhere 7) select windowlevels; 8) select start mode; 9) read out case; 10) review consultation report;11) sign case; and 12) log out. However, such modeling is difficult for theradiologist, though it is easy for the vendors engineers.

Conclusion

The radiologist can become an integral part of the PACS acquisition processby modeling the desired operation of the grayscale workstation and how it isintegrated into the workflow. Though the models shown in tables 1 and 2 andfigure 1 are simple to construct, they are important in assuring that theworkstation is an integral part of the workflow and that the workflow resultsin an adequate throughput. Such models, when presented to the vendor, willassist in determining how best to achieve a "user friendly"grayscaleworkstation for your particular departmental needs and preferences. AR