| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Computers in Radiology |
1
Radiology Service (114), Veterans Affairs Boston Health Care System, 150 S.
Huntington Ave. Boston, MA 02130.
2
Boston University School of Medicine, Boston, MA 02130.
Received April 26, 1999;
accepted after revision June 14, 1999.
Address correspondence to M.E. Gale
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CONCLUSION. Automated image display increases efficiency in image interpretation and remedies the normally cluttered presentation environment. At our institution, acceptance of automated image display has been overwhelmingly positive. In fact, automated image display has improved radiologist productivity.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
When redesigning our PACS interface, we worked with the vendor to optimize the program to enhance work flow at the PACS server and at the PACS workstations. The original interface was a typical work list display. The radiologist opened the work list by clicking the appropriate desktop icon and then choosing the patient's folder. When the user chooses the study and series from both current and prior examinations, and the chooses the display screen for each set of images, the selected images appear on the four workstation screens. Before the radiologists could start interpretation, they had to click an icon to notify the server and other workstations that the study was opened for interpretation. After interpretation, the radiologists could mark the study as read by clicking the appropriate icon.
To optimize the PACS interface, we redesigned the interface to include a small dialog box called Autoread. The Autoread box appears on the screen throughout an interpretation session and enables users to perform previously multistepped functions with a single mouse click (Fig. 1). The Autoread box contains five function buttons: Read, Cancel, Opposite Read, Opposite Cancel, and Close. The Read button marks the case as interpreted and instructs the server to display the oldest uninterpreted studies with comparison images in the automated display queue. The Cancel button skips the case without marking it as read and brings up the oldest uninterpreted studies. The Opposite Read and Opposite Cancel buttons look similar to the Read and Cancel buttons with opposite facing arrows. These buttons instruct the workstation to display the newest studies rather than the oldest studies in the automated display queue. Finally, the Close button enables the interpreter to exit the automated display session.
|
Each time the Read or Cancel button is clicked, images from the current study are cleared from the screens and a new study is displayed with comparison images. The images are displayed according to screen rules that are configured for each user. For example, a single anteroposterior chest radiograph is typically placed on the far right screen and the comparison images are placed in reverse chronologic order on adjacent screens. As another example, posteroanterior and lateral chest radiographs are displayed in the center two screens with the comparison posteroanterior and lateral radiographs placed on the outer two screens. Each image is automatically loaded onto each screen to facilitate selection for easy side-by-side comparisons. Earlier comparison images are also automatically loaded in video memory behind the visible images on each of the four screens. These images can be selected by clicking on a single icon. In essence, Autoread was designed to emulate a properly prehung film alternator.
The status of each examination is tracked by the server as read, unread, or in process of being read. The server dynamically updates Autoread and work lists at the workstations. Cases that are marked as read are removed from the automated display queue at each workstation to which they have been routed. The Digital Imaging and COmmunications in Medicine (DICOM) interface with the hospital information system provides PACS with the report text for each verified report. Even if a case has not been marked as read on the workstation, the server treats the presence of a verified report as an indication that the study has been read and removes the case from the automated display queue. Such cases arise most commonly from failure to click the required icons during a work list-based interpretation, from using film-based interpretation (which typically occurs during the transition to softcopy interpretation), or from erroneously selecting the Cancel button during automated display.
Similar to a typical work list display, when a case is opened for interpretation using automated image display, additional users may view cases as view-only files. View-only files may not be changed and when an additional user attempts to read a view-only file, that user will receive a message indicating the file is being interpreted elsewhere.
Previous reports have suggested a slight time advantage or equivalence using hard copy compared with soft copy for interpretation [5, 6, 7, 8]. The purpose of this study was to evaluate the efficiency of automated image display compared with both hard-copy film hanging and PACS work list methods.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Because our institution does not regularly print computed radiography studies, all examinations used for this study were manually printed on laser film and placed in a standard film jacket. Comparison images were printed before the start of the study to provide comparison case material for film hanging during the study period.
In this study, examinations were limited to ICU films to maintain consistency and to assure the presence of old films. Patients in the ICU typically have one or more examinations per day, and therefore have at least two or more comparison examinations readily available. Radiologists were instructed to use as many comparison images as they would during a typical interpretation session. In all cases, the radiologists used at least two of the three comparison studies (automated image display, work list display, and manual film hanging). When interpreting with the work list method, the work list display was typically left open on the fourth screen instead of being used to display a comparison image (removing the necessity to click on the work list Open and Close icons between each case). For film interpretation, the current examination was most often hung with two comparison films, but in some cases, either one or three comparison studies were used. The automated display mode used all four available screens when at least three comparison studies existed.
The trial was conducted over a period of 12 days. On each day, three radiologists interpreted the same 10-16 ICU chest films using one of the three methods on a rotating basis. Each radiologist had more than 1 year's experience using the PACS workstations. Each case was manipulated, interpreted, and dictated using techniques appropriate to the method. The radiologists were blinded to which of the three dictations would be preserved in the medical record.
For each case, the inter-interpretation time was measured. For the PACS-based interpretations using automated display and work list display, the inter-interpretation time was assessed from the time the case was marked as read until every image of the next case was displayed either automatically or manually on the screen. For film-based interpretation, the inter-interpretation time was measured from the time the first film was removed until the last film of the subsequent case was hung on the viewboxes. The interpretation time was calculated as the difference between the measured total time and the measured inter-interpretation time.
Statistical analysis was performed using a mixed design analysis of variants. Pairwise comparisons of means were conducted using the Newman-Keuls test.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
Inter-Interpretation Time
For each radiologist, a significant difference was noted in
inter-interpretation time between automated display and work list display and
between the automated display and manual film hanging (p 
0.0001). For two of the three radiologists, inter-interpretation time was
faster using work list display than manual film hanging, whereas the converse
was true for one radiologist. Combining all results show that no reliable
difference existed in the inter-interpretation time between work list display
and manual film hanging.
Interpretation Time
All three radiologists performed consistently in interpretation time.
Automated display was significantly faster than with the other two methods
(p 0.0001). Also, each radiologist was significantly faster
using the work list display than using manual film hanging (p
0.0001).
Total Interpretation Time
For total interpretation time, the pattern across interpreters for
automated display, work list display, and manual film hanging was consistent.
All three radiologists were significantly faster using automated display than
using work list display and manual film hanging (p 0.0001). The
radiologists were also faster using the work list display than using manual
film hanging, but the results only approached significance (p <<
0.01).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A portion of the savings in total interpretation time found with automated display resulted from the radiologist's ability to use the automated image display inter-interpretation time to begin the process of dictation. After the single click to mark a study as read, the radiologist could begin entering the next patient's identity electronically and verbally in the dictation station while waiting for the PACS to serve as a film-hanging clerk.
Interpretation time was also significantly shorter using the automated display rather than the work list display or manual film hanging. The additional savings in interpretation time may be attributable to the automated display's ability to display uninterpreted images first and then comparison studies sequentially in reverse temporal order. As a result, during the inter-interpretation time, the radiologist could begin studying the newest examination and start the report dictation before all the screens finished displaying images and the interpretation time officially began.
We did not study the use of a prehung multiple-image film viewer that is currently available in some practice settings. The inter-interpretation time for this method might be less than that of manual film hanging; however, radiologists typically spend time rearranging the order of films and retrieving the needed comparison images from the master folder, diminishing the potential timesaving advantage of this technique. Further, prehanging film-based studies requires additional time and labor.
Time saved using both the automated display and work list methods was facilitated by an extensive CPT code-based, prefetch algorithm. In our system, it is unusual to need to fetch additional images from the server or the archive at the time of interpretation. For example, a new posteroanterior and lateral chest examination triggers a prefetch from the server or archive that includes the three most recent posteroanterior and lateral images, the three most recent anteroposterior portable chest examinations, the most recent chest and abdominal CT scans, the most recent abdominal studies, the most recent thoracic and cervical spine and shoulder examinations, and the oldest available posteroanterior and lateral chest examinations. In automated display, all these studies are automatically placed for rapid access in a virtual, chronologically appropriate stack on each screen. In essence, the PACS performs a directed search through the patient's folders to automatically retrieve pertinent studies. The system then places the studies in the most appropriate viewing order and location.
During this study, some minor variability in image display time using both automated display and work list methods was introduced by incoming image data for other cases. When this incoming data was being written to the workstation disk, retrieval of selected images from the disk for display was deferred to this higher priority function. This situation occurred occasionally because the workstation used for this study routinely received body MR images in its interpretation queue.
For a typical set of 16 examinations, the total session time averaged 16.8 min for film interpretation, 15.5 min for work list interpretation, and 10.4 min. for automated display interpretation. This represents a 38% timesavings for PACS with automated film hanging as compared with film interpretation, the standard in radiology practice. Although our study was limited to ICU chest film examinations, our experience is that the efficiency of automated display interpretation is at least as great in multiple-image cases as in single-image cases because the PACS automatically assembles the images according to coherent screen display rules for both new and comparison images.
Although this study investigated only single-film examinations, we believe that even more time may be saved in multiple-image examinations. Multiple-image examinations typically require the radiologist to search through several folders to obtain the appropriate comparison films. At our institution, the timesaving has been evident since the September 1997 transition from film-based interpretation to soft-copy interpretation. Staff radiologists come to the department in the morning and late afternoon when they are on call. A typical morning interpretation session previously took an average of 3 or more hours. After conversion to PACS with automated display, the average session now requires less than 2 hr for the same quantity of work. Further, in daily practice when both automated display and work list interpretation methods are available, the staff routinely choose to interpret computed radiography with automated display.
In our experience, the time saved is less evident when using automated display over work list display to interpret other techniques such as CT, MR imaging, and sonography. In any interpretation session, the proportion of time devoted to studying these examinations relative to the inter-interpretation time is far greater, and the total number of inter-interpretation intervals tends to be less. This is reflected in practice when radiologists use both automated display and work list for these techniques depending on circumstances such as the presence of a resident on service and consultation with clinicians. Unlike software from many vendors for whom the development focus has been on cross-sectional imaging and image manipulation such as maximum-intensity-projection or three-dimensional reconstructions, the current version of our PACS software has been optimized for computed radiography, overwhelmingly the largest volume technique in most radiology departments.
Integration of PACS with the hospital information systems and radiology information systems facilitates more informed and presumably more accurate interpretations [10]. When interpreting film, access to previous reports required logging in to the hospital information system on a separate computer and navigating through the menu structure to view the report. Typically, the radiologists chose not to review prior reports during the film interpretation portion of the study, probably because this was a relatively time-consuming process and disrupted work flow. When using either automated display or work list display, previous reports were available for every image on display by clicking a single icon. Though the time accessing reports was not specifically tracked, review of old reports was occasionally part of the process on soft-copy interpretation during the study. This time was included within the interpretation-time measurement. Had it been tracked and excluded, the PACS interpretation time results, both automated display and work list display, would likely have been shorter than that with film interpretation.
In conclusion, attention to the details of interface design can have a significant impact on the cost-effectiveness of a technology such as PACS in which capital outlay for start-up and maintenance costs can be the most important determinant of viability. Although film interpretation using PACS with a conventional work list algorithm is comparable in efficiency and accuracy with film interpretation, the automated image-display method increased efficiency and resulted in presumably more accurate interpretations.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  C.-S. Yam, A. Sitek, V. Raptopoulos, and M. Larson A Simple Method for Extracting DICOM Images from a Magnetooptic Disk Am. J. Roentgenol., August 1, 2004; 183(2): 529 - 533. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. I. Reiner, E. L. Siegel, F. J. Hooper, S. Pomerantz, A. Dahlke, and D. Rallis Radiologists' Productivity in the Interpretation of CT Scans: A Comparison of PACS with Conventional Film Am. J. Roentgenol., April 1, 2001; 176(4): 861 - 864. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
