The Pathology Education Instructional Resource (PEIR) has repurposed the NLM videodiscs by converting the contents of each of the videodiscs into an online digital format. What follows is an e-poster from the APIII 2002 meeting detailing the conversion process.
NLM's Basic Medical Pathology Videodisc Series Repurposed for Web Delivery
Kristopher N. Jones, BS, BA; Nagaraju Sarabu, MBBS; Kristina T. C. Panizzi, MAE; and Peter G. Anderson, DVM, PhD
Background
Between 1983 and 1993, Dr. James W. Woods of the National Library of Medicine oversaw the creation of the Basic Medical Pathology series of 11 videodisc-based learning modules--an interactive multimedia learning project designed to introduce medical students to the basic concepts of microscopic pathology. Each of the learning modules consisted of a 25-30 minute audio-video mini-lecture and an associated quiz component of approximately 100 questions that referenced images on a given module’s videodisc. The computer system on which the student took the quiz was connected to a videodisc player and video monitor via an RS-232C interface that allowed the computer to cue up images on the second monitor. Most quiz questions referenced multiple magnifications of pathologic and normal slides and feedback for a given answer choice also often included additional images. In the days before computers could natively display high-quality digital images, the ability of analog videodisc technology to combine audio, video, and still image resources on a single platter which could be controlled by an external computer via an RS232 interface was a high tech marvel. As a result, these videodiscs became very popular among medical students over the years and came to be in use at over 100 medical schools in the United States and throughout the world.
Obsolete Medium, Relevant Material
What made the Basic Medical Pathology series popular and unique also defined its limitations—namely, use of the system required specialized equipment not available outside of a given medical school’s physical plant. Further, even before the completion of the series, videodisc technology was already become obsolete; and, by the late 1990s, CD-ROM technology could accomplish the work of the videodisc for less cost (both for production and delivery). The final blow to use of the system was the explosion of WWW-based delivery of medical education materials that allowed students to access video, images, and testing materials from any Internet-capable device. Following an upgrade in 1999 of the student computing facilities at our institution (University of Alabama School of Medicine), the videodiscs were still available for students to borrow from the Learning Resource Center but there were no longer any computer-controlled videodisc players on which they could be used; the mini-lectures could be viewed on standalone videodisc players, but the quiz components were no longer available. Students who stumbled across the videodiscs in recent years highly-rated the mini-lectures and—when told of the quiz component—frequently requested the full resource be made available to them for self-study (even if not required in the curriculum).
Purpose of this Project
Recognizing these trends, in the late 1990s, Dr. Woods converted the series’ two Neoplasia mini-lectures to a WWW-based format (http://tlc2.nlm.nih.gov/neoplasia)—including RealAudio, text, and screen captures of selected video images—that are frequently referenced on medical school pathology course Web sites. However, the other 9 videodiscs were never similarly converted and none of the quiz components were ever converted for Web-based delivery.
The purpose of this project was to convert the videodisc-based Basic Medical Pathology series to a WWW-delivered system using a process for automated digital conversion of videodisc resources previously described (http://apiii.upmc.edu/apiii2000/epwin.htm) at APIII 2000 (http://apiii.upmc.edu/apiii2000/).
Design/Methods
First Step: Reproducing Program Functionality
Before proceeding with the project, a delivery system for the quiz component of the Basic Medical Pathology series had to be identified. Hot Potatoes Version 5.4, a freeware (for educational purposes) program for WWW-based delivery of multiple-choice exercises developed at the University of Victoria (http://web.uvic.ca/hrd/hotpot), was selected due to its cost (free!), the source code of its output (DHTML/JavaScript) was open and modifiable, and its functionality closely mirrored that of the program developed by and under Dr. Woods’ direction. Custom HTML templates would be needed to display the multiple image magnifications required to emulate the previous videodisc-based quiz component.
Second Step: Extracting ASCII Data from Legacy Materials
The functionality of the Basic Medical Pathology system was provided by a combination of CLIPPER programs, C programs, and third party object code library members, while all of the data for the quiz components were stored in 11 individual dBASE-formatted files. The functionality of the program was to be provided by a self-contained testing program, thus it would not be necessary to view or manipulate the code of the original programs; however, the quiz data were needed. Both Microsoft’s Access database application and Microsoft’s Excel spreadsheet application can read dBASE-formatted files, and it was decided to use Microsoft Access to manipulate these data due to its ability to execute custom queries against and produce reports from the data tables. These features would be needed to reformat the data into JavaScript arrays needed for the new testing program.
Third Step: Digital Capture of Analog Videodisc Resources
Videodiscs contain data in analog—not digital—format. Therefore, the audio-video mini-lectures and the images on each of the Basic Medical Pathology videodiscs had to be converted to a digital format through a “capture” process. A still and motion video capture card from ATI (ALL-IN-WONDER 128) was chosen to perform this conversion from a videodisc player with high-quality digital comb filters (http://www.cs.tut.fi/~leopold/Ld/FAQ/Hardware.html#TechComb), S-video output, and an RS-232C serial communications port (Pioneer CLD-V2600). Capture of the 11 video mini-lectures utilizes the ATI MPEG-2 video capture software installed as part of the ATI Multimedia Center. Once captured, the digital videos can be edited and converted to the RealVideo8 Web-compatible streaming media format using RealProducer Plus software from Real Systems (http://www.realnetworks.com/products/producer).
Capture of the individual images referenced by the quiz questions was performed using an automated digital capture process previously described (http://apiii.upmc.edu/apiii2000/epwin.htm) at APIII 2000 (http://apiii.upmc.edu/apiii2000/). Briefly, a batch file is made that repetitively launches a keystroke automation program (Launchkey, http://www.splinterware.com/products/launchkey.htm) which first executes an RS-232C communications program that advances the videodisc to the appropriate frame to be captured and then executes a screen capture program that digitizes the desired image and then saves it to a local file. In this way, 5 images per minute may be captured from the videodisc. The images can then be automatically batch edited (color correction, cropping, etc.) using Atalasoft’s EyeBatch batch image processing software (http://www.atalasoft.com).
Results
First Step: Reproducing Program Functionality
Though the Hot Potatoes program has a user-friendly graphical user interface (GUI), due to needed modifications in the templates and the ability to have Access create the necessary data arrays, it was decided not to enter the question data field-by-field through the interface. Rather, a single question was entered through the GUI and all settings were chosen and then a single-question HTML output file was generated. This file was modified with custom JavaScript code and HTML formatting and then duplicated 10 times. JavaScript arrays generated in the second step were then to be pasted into these shell exam files.
Second Step: Extracting ASCII Data from Legacy Materials
Each of the 11 dBASE-formatted data files was converted into a Microsoft Access database containing 85 fields:
-
DB {DB identifier—two letter code for each module}
-
QID {Question number}
-
TYPE
-
PREFERRED
-
QUESTION {Question Stem}
-
ANSWER {Number of correct answer choice—0 to 4}
-
FOIL1 ... FOIL5 {Answer choices}
-
PATHFRAME1 ... PATHFRAME7 {Frame numbers of question images}
-
NORMFRAME1 ... NORMFRAME7 {Frame numbers of normal reference images}
-
STARTFRAME {Which PATHFRAMEn field (magnification) should be displayed first}
-
HELPSTART {Frame number of first video frame for student to review mini-lecture}
-
HELPEND
-
QUESTSTART {Frame number of first video frame for student to review mini-lecture}
-
QUESTEND
-
FDBK1 ... FDBK5 {Feedback text corresponding to FOIL of same number}
-
FRAME1 ... FRAME5 {Feedback image corresponding to FDBK and FOIL of same number}
-
FRAME1STRT ... FRAME5STRT {Frame number of first video frame for student to review mini-lecture}
-
FRAME1END ... FRAME5END
-
FRAME1MAG1 ... FRAME1MAG7 {Additional magnifications of feedback images}
-
FRAME2MAG1 ... FRAME2MAG7
-
FRAME3MAG1 ... FRAME3MAG7
-
FRAME4MAG1 ... FRAME4MAG7
-
FRAME5MAG1 ... FRAME5MAG7
The fields in eight of the 11 data files were in the above order, with the remaining three files in a variant order. The fields in these three files were reordered in Excel before importation to Access. The TYPE, PREFERRED, HELPSTART, HELPEND, QUESTSTART, QUESTEND, FRAMEnSTRT, and FRAMEnEND fields were unneeded or unusable for our purposes and, thus, deleted. Due to design limitations in the Hot Potatoes quiz templates, it was decided not to use multiple magnifications of feedback images in this first attempt at converting the quiz components and, therefore, the 35 FRAMEnMAGn fields were deleted. It was also decided to limit magnifications on the question and normal images to five images, since almost all records were limited to five, rather than seven; thus, the PATHFRAME6, PATHFRAME7, NORMFRAME6, and NORMFRAME7 fields were deleted. The remaining fields were exported into ASCII flat files and spell-checked and reformatted to remove previous formatting characters. The text files were then edited to remove language specific to the previous interface—e.g., references to function keys, TV screens (videodisc monitor), etc. Finally, the text files were reimported to Access and a custom report was created to reformat the remaining fields into the JavaScript array format used by Hot Potatoes:
-
I[3] = new Array();
-
I[3][0]='Before release into the blood stream, hepatic lipid must be bound to<script> parent.primePics("00063","05970","06778", "07462","07976","02563","0","0","0","0","1"); </script>';
-
I[3][1] = new Array();
-
I[3][1][0] = new Array();
-
I[3][1][0][0]='glycogen.';
-
I[3][1][0][1]='Incorrect. Glycogen can alter hepatic metabolism and lead to increased fatty acid synthesis and fat accumulation, but lipid does not bind to it.<Form><input type="button" value="View Feedback Slide" class="typebutt" onClick="javascript: parent.feedPic(\'05970\');"></form>';
-
I[3][1][0][2]= 0;
-
I[3][1][1] = new Array();
-
I[3][1][1][0]='glycerol.';
-
I[3][1][1][1]='Incorrect. Increased glycerol phosphate can lead to increased esterification and fatty liver, but glycerol is not a secretory lipid acceptor.<Form><input type="button" value="View Feedback Slide" class="typebutt" onClick="javascript: parent.feedPic(\'07462\');"></form>';
-
I[3][1][1][2]= 0;
-
I[3][1][2] = new Array();
-
I[3][1][2][0]='apoprotein.';
-
I[3][1][2][1]='Correct. Hepatic lipid must be bound (to apoprotein) prior to release into the circulation. Thus, a low titre of apoprotein can lead to a fatty liver.<Form><input type="button" value="View Feedback Slide" class="typebutt" onClick="javascript:parent.feedPic(\'03433\');"></form>';
-
I[3][1][2][2]= 1;
-
I[3][1][3] = new Array();
-
I[3][1][3][0]='glucose.';
-
I[3][1][3][1]='Incorrect. Hepatic lipid binds to protein and becomes lipoprotein.<Form><input type="button" value="View Feedback Slide" class="typebutt" onClick="javascript:parent.feedPic(\'03433\');"></form>';
-
I[3][1][3][2]= 0;
-
I[3][1][4] = new Array();
-
I[3][1][4][0]='glycoprotein.';
-
I[3][1][4][1]='Incorrect. Hepatic lipid binds to protein and becomes lipoprotein.<Form><input type="button" value="View Feedback Slide" class="typebutt" onClick="javascript:parent.feedPic(\'03433\');"></form>';
-
I[3][1][4][2]= 0;
-
I[3][2] = 0;
Each of the 11 arrays was then modified with a Perl regular expression query to strip out all feedback other than “Correct” and “Incorrect” for use in the Pre-Test component of each learning module, where this is the only feedback the student should see. Thus, 22 arrays were created and pasted into the shell exam files created above in the first step, thus creating 11 Pre-Test quiz files with limited feedback and 11 Post-Test quiz files with full feedback. The settings in the shell exam called for only 10 random questions to be presented, whereas the study module should present all the questions. Thus, a third set of quiz files was created from the Post-Test quiz files, in which an internal setting was changed to allow all questions—not just 10—to be presented.
In all, there were a total of 1058 questions (940 multiple-choice, 118 true/false) in the 11 data files:
Cell Injury: 91 MC 9 T/F
Necrosis: 75 MC
Cellular Alterations: 97 MC
Cellular Accumulations: 100 MC
Exudates and Phagocytosis: 100 MC
Chemical Mediators: 98 MC
Chronic Inflammation: 100 MC
Edema, Congestion, and Shock: 87 MC
Thrombosis and Embolism: 100 MC
Neoplasia: Nomenclature: 43 MC 58 T/F
Metastasis and Differentiation: 49 MC 51 T/F
Third Step: Digital Capture of Analog Videodisc Resources
The 11 mini-lectures were captured and converted to RealVideo8 format using SureStream settings to insure the files could be delivered in a variety of bandwidths for different access speeds—from modem to LAN. These files were uploaded to two different servers to provide duplication of video access in case of server downtime or slow access.
All images specified in the dBASE data files were captured—even those not to be used in this initial version (e.g., the feedback magnification slides). In the end, 4,618 images are referenced in the eleven quiz files. These files were batch processed as listed above and uploaded to the server for access by the quiz files.
