| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Computers in Radiology |
1 Department of Radiology, Harborview Medical Center, Box 359728, 325 Ninth
Ave., Seattle, WA 98104.
2 Harborview Injury Prevention and Research Center, Harborview Medical Center,
Seattle, WA 98104.
3 Department of Radiology, University of Vienna, Waehringer Guertel 18-20,
A-1090, Vienna, Austria.
4 Department of Surgery, Harborview Medical Center, University of Washington,
Seattle, WA 98104.
Received August 22, 2002;
accepted after revision September 19, 2002.
Selected to be presented at the annual meeting of the American Roentgen Ray
Society, San Diego, May 2003.
Abstract
 Top Abstract IntroductionThe Research ProjectDiscussionReferences |
|---|
CONCLUSION. The Web-based system that we present uses a commercially available Internet browser and Web platform and allows automated data entry that can be easily uploaded into standard data analysis programs. The system simplifies the complex logistics of using multiple sites and reviewers for radiology research and can preserve human subject confidentiality. We tested the system using a large-scale multicenter cohort study of pelvic fracture–related hemorrhage (the "Evaluating Pelvic Hemorrhage" study). Program testing revealed seamless remote image interpretation and data acquisition.
|
TopAbstractIntroductionThe Research ProjectDiscussionReferences |
|---|
The advent of PACS (picture archiving and communication system) has brought tremendous speed and flexibility to image display. With PACS, electronic review of images from geographically distant sites is possible. However, special interfaces may be required, and the inclusion of patient-specific data on PACS images may violate human subjects' confidentiality requirements. Furthermore, PACS contains no capability for integrated data collection. The use of a Web-based interface for simultaneous image review and data collection has the potential to avoid the limitations of these other systems.
In this article, we describe a user-friendly Web-based system for multisite image interpretation and automated data collection developed for an ongoing large-scale multicenter cohort study. This system is based on a commercially available Internet browser and Web platform and allows automated entry of data that can then be easily uploaded into standard data analysis programs. The system simplifies the complex logistics of using multiple sites and reviewers for radiology research and can satisfy the need to preserve patient confidentiality in research studies. This system has broad potential applications in radiology research.
|
TopAbstractIntroductionThe Research ProjectDiscussionReferences |
|---|
Image interpretation.—The pelvic radiographs for each study subject were analyzed by one or more of five investigators located at multiple sites in the United States and Europe. For each subject, a series of 14 determinations regarding the location and extent of bony injury were made from the radiographs by the investigators and entered into a database for analysis. Included were identification of any fractures of the iliac wing, the sacrum, the obturator ring, the acetabulum, and the sciatic notch. If a fracture was identified, the extent of displacement was characterized as 5 mm or smaller or as larger than 5 mm. Reviewers noted if the sacroiliac joints were normal, or if diastatic, whether they were less than 10 mm, 10–30 mm, or greater than 30 mm. From the radiographic findings, the radiologist reviewers also determined the suspected injury pattern (anteroposterior compression; lateral compression; vertical shear; combination; anterior pelvic ring injury only; unknown mechanism). The pelvic radiographs were interpreted, the questions answered, and the data automatically collected through the Evaluating Pelvic Hemorrhage study Web site.
Web site.—The Evaluating Pelvic Hemorrhage study Web site is based on a commercially available Web content management system (Manila; Userland Software, Burlingame, CA) and operates using an on-site server system (ProLiant 2500; Compaq Computer, Cupertino, CA). Access to the Web site is password-protected and limited to the investigators only. The home page contains an introduction explaining the purpose of the research study as well as a summary about how the Web site works.
Each of the radiologist reviewers has a home page on the study Web site with links to detailed instructions about the research questions. The pelvic radiographs are organized into modules of 10 cases each. Reviewers navigate through the modules by way of links with access to detailed instructions available on each page. Once a module has been completed, the research Webmaster removes the link to the module from the reviewer's work list. The investigators' Web sites also provide lists of modules completed, with links to new modules that have not yet been interpreted (Fig. 1). Each reviewer has a different set of modules, but to assess inter- and intrareviewer variability, we assigned some cases and modules to more than one reviewer. This joint assignment was performed simply by adding links from the image files to the investigator Web pages. Web site management and image loading are performed by a part-time undergraduate in the work–study program.
|
The basic layout of a individual case overview is a quarter-screen image of the pelvic radiograph with adjacent questions to be answered (Fig. 2). Reviewers click on the image to open a full-screen image of the pelvic radiograph in a different window. The size of the full-screen image approximates the size of the digital images produced in hard-copy format for clinical use. The New window option allows reviewers to go back and forth between the questions and the actual-size image. Questions are answered by clinking on one of several radio buttons, each of which represents a different response. At the end of the question list, a series of indicators prompt the reviewers if any questions remain unanswered.
|
Data entry.—After each case has been completed, the reviewer clicks the "Submit" button, which sends the data to a CGI (Common Gateway Interface) program written in Perl (Practical Extraction and Report Language). This program performs several tasks, including validating the data, sending feedback to the reviewer who submitted the data, storing the data in a database, and sending the data to the study coordinators via e-mail.
If a reviewer attempts to submit data with unanswered questions, the Web server prompts the user to go back and enter any missing data. Once all the questions have been answered, clicking on the Submit button presents a screen to the reviewer echoing all the responses to the study questions. This feedback screen shows the reviewer that the information was received safely and was entered correctly into the database (Fig. 3). The reviewers are then prompted to start the next case. The assessment data received from reviewers are time- and date-stamped and appended automatically to the study database. As a backup, the program immediately sends the data to an e-mail account managed by study coordinators. These e-mails are filed and organized according to their case number. The data accumulating in the database are periodically harvested and uploaded to a standard database (Excel 2000, Microsoft, Redmond, WA; or Stata 6.0, Stata, College Station, TX) for management and analysis. The system is flexible and can be set to allow reviewers to return and change responses at a later date. The automated e-mail backup function corrects for multiple or erroneous data submission by the reviewers.
|
Protection of subject identities.—The posting of medical information on the World Wide Web presents potential risk of invasion of patient privacy due to public access to the Internet. We have made our Web site password-protected. To gain access to the Web site, reviewers must obtain an access code from the Web site study coordinator. More important, the images and case pages are all coded by study number only. No patient-identifying information is available anywhere on the Web site. These protections have met with the approval of our human subjects review committee and allow blinded review of study images.
Limitations
Setup of the CGI program requires advanced knowledge in Web server
administration and Perl. The Perl script was created by one of the
investigators and is not a part of the Web management content software.
Composing the image cases requires some knowledge of HTML (Hypertext Markup
Language) coding. However, in our system, coding is easily handled by a
premedical undergraduate student who completed a 2-hr HTML training course. In
addition, a reliable Web server must be available and maintained by technical
staff. The server used in this study was also used for managing the
departmental home page as well as Web-based electronic teaching files. A
pitfall that we encountered was that the system could not be used during
periods of scheduled and unscheduled server maintenance. We also encountered
some sporadic variations in Web site layout between Macintosh and PC operating
systems.
JPEG File Interchange Format was used for broad compatibility across various computer platforms and commercially available Web browsers. Loss of image information using the JPEG compression algorithm potentially limits diagnostic assessment of soft-copy radiographs. However, degradation of radiographic image quality due to JPEG compression is negligible at compression ratios up to 1:20 [1, 2]. Alternatively, image compression without data loss could be achieved using file formats such as JPEG-LS, PNG (Portable Network Graphics), or TIFF (tagged image file format) at the cost of larger image files.
|
TopAbstractIntroductionThe Research ProjectDiscussionReferences |
|---|
Systems, such as the one described in this article, have the potential to enable large-scale multicenter trials of image interpretation and diagnostic accuracy. Studies need not be limited to a few investigators at a single institution, but investigators can take advantage of the widespread expertise and case experience of individuals at a virtually limitless number of institutions. This Web-based format also protects patient confidentiality and easily allows the study investigators to ensure the interpreting radiologist is unaware of patient information. Additional applications for this technology are also possible in continuing medical education and competency testing [3].
Acknowledgments
We thank Steve Breathour, Timothy Quagliaroli, and James Lyon-Hall for
outstanding technical computer systems support.
|
TopAbstractIntroductionThe Research ProjectDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  A. G. Ryan, L. J. Louis, and W. C. Yee Informatics in Radiology (infoRAD): HTML and Web Site Design for the Radiologist: A Primer RadioGraphics, July 1, 2005; 25(4): 1101 - 1118. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
