- Email: [email protected]
New opportunities for the use of digital video in on-line radiologic curricula1
Radiologic Education
New Opportunities for the Use of Digital Video in On-Line Radiologic Curricula 1 Mark S. Frank, MD, Richard B. Gunderman, MD, PhD
Rationale and Objectives. To explore techniques to conveniently and self-sufficiently create high-quality, web-ready instructional digital video snippets suitable for routine use in on-line radiologic curricula. Materials. A commercially available digital camera with an 8-megapixel image receptor and the capability to record webready digital video and audio at a resolution of up to 640 ⫻ 480 pixels and a frame rate of up to 30 per second was used to obtain video snippets intended for inclusion in on-line curricula. Hand-held and tripod techniques were compared. Evaluation focused on the types of snippets deemed most likely to be used within on-line educational content, ranging in length from 10 seconds to 2 minutes. Additionally, basic postprocessing functions to experiment with combinations of video-file size, format, transmission efficiency, and image quality were used. Results. The overall video quality was considered by participating radiologists to be good to excellent for its intended purposes. For most situations tested, a matrix size of 320 ⫻ 240 pixels provided a good balance of visual quality versus file size and transmission overhead. The 640 ⫻ 480 format was occasionally optimal, but was usually larger than necessary and resulted in substantially larger files, especially at a rate of 30 frames per second. A rate of 15 frames per second was considered adequate for most situations, regardless of matrix size, although it did add some barely discernible choppiness to the display. For all but very short snippets, a tripod or some other form of stabilization was necessary to eliminate distracting “camera shake” associated with free-hand acquisition. With a tripod, appropriate ambient lighting, and a favorable acoustical environment, videos of excellent quality when using the highest settings available on the camera (640 ⫻ 480, 30 frames/sec) could be self-sufficiently produced. However, such high-quality video with its associated large file size was rarely the optimal fit for the needs of this curricula. Conclusion. These results indicate that newer-generation digital cameras are useful for quickly and inexpensively producing high-quality, web-ready digital video suitable for use in on-line education. Key Words. Digital video; radiology education; online learning; teaching file. ©
AUR, 2004
The potential for using multimedia in radiologic education is well known (1,2), and digital video is widely used today to distribute radiologic curricula on the World Wide Web. Video is often transmitted in a stream that encom-
Acad Radiol 2004; 11:1144 –1148 1 From the Department of Radiology, Indiana University School of Medicine, 714 N Senate Ave, Suite 100, Indianapolis, IN 46202. Received March 22, 2004; revision requested May 24; revision received July 13; accepted July 15. Address correspondence to M.S.K. e-mail: [email protected]
© AUR, 2004 doi:10.1016/j.acra.2004.07.004
1144
passes an entire lecture or large components thereof, frequently displaying both a presenter’s materials along with a small inset video of the presenter as he or she speaks. Some people believe that this is an effective approach to on-line education, while others believe that this format is less effective than simply providing parcels of text and images to convey the same information. Regardless, we believe there is yet another way to effectively and routinely use digital video for on-line education. Rather than using video as the dominant form of media within on-line content, we have experimented with using relatively short segments of digital video interspersed and complementary
Academic Radiology, Vol 11, No 10, October 2004
with other components of content such as stylized text, still images, and interactive questions with immediate feedback. We have investigated ways to self-sufficiently and inexpensively create high-quality digital video snippets and then integrate them into teaching-file cases and other on-line curricula in our department (Department of Radiology, Indiana University School of Medicine, Indianapolis, IN). We report on our results and the surprising ease with which this can be accomplished when a suitable content-management infrastructure is in place.
MATERIALS AND METHODS We chose the Sony DSC-F828 digital camera (Sony Inc, New York, NY) with a market price of approximately $850. It contains an 8-megapixel image receptor and the capability to record motion picture experts group (MPEG) format web-ready digital video. In video mode, the camera will commence recording when the exposure button is pushed and will continue recording until its memory card is full or until the exposure button is again pushed. Compact-flash, micro-drive, and memory-stick storage are supported. The camera can record at a resolution of up to 640 ⫻ 480 pixels and a rate of up to 30 frames per second (FPS). The camera’s auto-focus system and zoom lens remain operational during video recording. The auto-focus system provides settings to adjust for different recording circumstances, and a manual-focus mode is also supported. White-balance settings facilitate rapid color adjustment for different sources of ambient lighting (eg, sunlight, incandescent, fluorescent). The camera has a built-in microphone but no external microphone jack. We created video snippets in three categories that we deemed most likely to be adjunctive to other forms of online digital content such as still images and stylized text. These were: (1) a brief video of a person introducing or emphasizing something, typically no more than 30 seconds in length; (2) a video snippet ranging from 10 seconds to 2 minutes that demonstrates a technique, such as how to perform a portion of a procedure; and (3) a video snippet that demonstrates an environment or a situation, such as a panoramic view of an angiography suite. Because the camera will record at either 640 ⫻ 480 or 160 ⫻ 120 pixels, but not at 320 ⫻ 240, all recording was performed using a resolution of 640 ⫻ 480 at either 15 or thirty FPS. Some basic postprocessing was performed. When needed for a particular facet of our experiment, we would resize a snippet from 640 ⫻ 480 to 320 ⫻ 240 and when necessary remove from
NEW OPPORTUNITIES FOR DIGITAL VIDEO
a snippet any excess leader or trailer. Video snippets were produced in four combinations: 640 ⫻ 480 at 30 FPS; 320 ⫻ 240 at 30 FPS; 640 ⫻ 480 at 15 FPS; and 320 ⫻ 240 at 15 FPS. In some circumstances we would also anneal some very short postprocessed snippets together to form one longer, continuous snippet. We generated video files in three different formats (Audio-Visual Interleave, MPEG, and Windows Media Video [WMV]) to determine which format seemed most effective for use on the World Wide Web. We experimented with both Camtasia Studio (Techsmith Inc, Okemos, MI) and Microsoft Movie Maker (Microsoft Corp, Redmond, WA) for performing these postprocessing steps. Our department’s digital-education software framework supports the storage and on-line use of digital video (with or without an audio track), digital audio, and static images. The framework contains an authoring system that facilitates construction of interactive case-based exercises as well as thematic curricula, and provides straightforward ways to intersperse video snippets within. Such course work is typically composed of various digital components (eg, stylized text, interactive questions that provide immediate feedback, media objects such as automatically zooming sequences of thumbnail images, and image stacks that simulate a PACS workstation). The software facilitates the construction and arrangement of these components to meet the author’s specific teaching objectives, the results often spanning several web pages. The system supports both direct embedding of a video within a web page and the capability to hyperlink content (eg, fragment of text, thumbnail image) to the default video player invoked as a stand-alone application on the learner’s computer. We placed video snippets within various forms of educational content and asked four participants (two radiologists, one production assistant, one pathologist) to subjectively assess both the quality and potential effectiveness of the videos.
RESULTS In general, our four participants were very pleased with the quality of digital video produced. When obtained in an environment with adequate ambient lighting, the consensus was that image quality at 640 ⫻ 480 pixels and 30 FPS was better than necessary for most on-line teaching applications. A technique of 320 ⫻ 240 pixels at 15 FPS was considered ideal for all head-and-shoulder videos of a speaker and for most “how-to” demonstrations. For demonstrations containing detailed structures
1145
FRANK AND GUNDERMAN
that also encompassed a wide field of view, a matrix size of 640 ⫻ 480 at 15 FPS was considered ideal. For demonstrations encompassing a narrow field of view, the camera lens could be dynamically zoomed to areas of interest, and a matrix size of 320 ⫻ 240 was then satisfactory. Similar perceptions were expressed for the same reasons when acquiring panoramic scenes such as a reading room or angiography suite. Although “full motion” 30 FPS recordings, when obtained in favorable lighting, provided near-professional quality video, consensus was that this did not add to the educational value of the content, primarily because the 15 FPS setting worked very well. In general, we found that the WMV file format provided the optimal balance of compression and image quality when compared with Audio-Visual Interleave and MPEG formats. We found that that Microsoft Movie Maker (a free accessory included with the Windows XP operating system) was more simple to use than Camtasia, but provided fewer options regarding the format and size of an output file. Both software applications provided a user-friendly interface for trimming the ends of a snippet and for annealing several snippets together. Movie Maker was particularly convenient for producing the most frequently preferred output in our experiment (WMV format, 320 ⫻ 240, 15 FPS). However, for video that demanded higher resolution and more control over image quality, we preferred the additional features in Camtasia’s rendering module. We also found both Movie Maker and Camtasia to be superior to the video-editing software bundled with the camera. With either application, a person familiar with the software could transfer an MPEG file from camera to computer, perform the basic editing described above, and produce a web-ready snippet, all in 2 or 3 minutes. Although a well-done MPEG video as obtained from the camera could be used on-line as is, we chose to transform each snippet into a substantially smaller file (usually WMV format) with no perceptible loss in video quality. When audio was recorded by an off-screen narrator (typically the person behind the camera), sound quality was judged to be good. However, when a speaker was the centerpiece of the video, it was necessary for that person to project his or her voice as if giving an unamplified presentation to an audience, and in such circumstances sound quality was influenced to a much greater degree by surrounding acoustics and ambient noise. For all but snippets of only a few seconds duration, partici-
1146
Academic Radiology, Vol 11, No 10, October 2004
pants felt that using some technique to steady the camera was critical to eliminate camera shake, which in addition to being annoying was also considered to negatively influence the overall credibility of the recording. Although a tripod provides the greatest flexibility for alleviating this problem, in some circumstances we found that simply bracing the camera against an available immobile structure was adequate.
DISCUSSION We are not the first to investigate the possible uses of multimedia in on-line radiologic education (1– 4). Our intent was to look more deeply into the feasibility of creating short segments of high-quality digital video selfsufficiently, inexpensively, and essentially on-demand, yet fully suitable for insertion into web-centric radiologic curricula. We are not making recommendations regarding digital cameras, computing platforms, or specific software products for on-line education. To those who primarily use Macintosh or Linux computers, we acknowledge that our experiment involved only computers running Microsoft Windows because it is the pervasive computing platform in our department. Regardless, we believe that the promising results we convey are independent of brand-specific factors. One advantage of using long segments of streaming video as the primary media for on-line curricula is that once the video is produced, few additional content-authoring or content-management capabilities are required to provide it as on-line course work. That is, the video itself conveys all of the educational content, and a simple web page is adequate for housing the video or hyperlinking to it. Some commercially available software products facilitate the integration of a soundtrack with sequences of static images (eg, Powerpoint slides), with an optional accompanying inset video of a speaker as well. Although it might be a labor-intensive endeavor, once produced, such an entity can virtually serve as a stand-alone component of course work. This approach can effectively emulate a lecture, but we believe it represents a relatively instructor-centric model of teaching, and when it becomes necessary, updating a long video can be nearly as labor intensive as originally producing it. Another approach to on-line education is to create curricula with the intention to have it interact with, challenge, and track the learner as the learner works through the course work. We regard this approach to be more Socratic and learner-centric, and
Academic Radiology, Vol 11, No 10, October 2004
streaming video seems to be less effective for this. We use the authoring capabilities of our digital education framework to assemble different forms of content such as stylized text, dynamic image sequences and stacks, and interactive questions with immediate feedback. These components are sequenced, interlinked, and incrementally revealed on-line to meet the specific objectives and desired teaching style of the educator. Such content is easily updated by the educator or a secretarial designee. Here is where we believe that short snippets of digital video can play a particularly enriching role, sometimes demonstrating in seconds what might be nearly impossible to put into words and often surpassing what could be demonstrated with several static images. For example, this approach could be useful when teaching fluoroscopic techniques or interventional procedures. The expectations set for our investigation were: 1. Snippets of video should be easy for an amateur photographer to record. 2. The format of the video produced by the camera should be nonproprietary and essentially web-ready. 3. The quality of video produced should be reasonably high, assuming that ambient lighting and acoustics are appropriate. 4. The snippets of video should be short enough to record with a single “take” (or two) and would therefore require little, if any, editing. 5. Any postproduction steps would be simple, relatively fast, and convenient to perform. 6. The end result could serve to complement and enhance the effectiveness of our on-line curricula. Meeting these expectations implies a form of pedagogical empowerment (ie, that somebody facile with the camera and some simple postproduction steps on a personal computer could single-handedly produce high-quality digital video snippets useful in on-line education). In general, our experience indicates that this implication is justified; however, we believe that there are some helpful caveats. The postprocessing steps we applied were simple to perform yet provided considerable latitude for balancing size versus quality of the final product. For example, a 30-second 320 ⫻ 240, 15 FPS WMV file of a speaker rendered with high compression requires approximately 500 KB of storage, while a less compressed version of the same file format rendered from the same 11-megabyte MPEG file (from the camera) requires approximately 3 megabytes. While the smaller version was functional,
NEW OPPORTUNITIES FOR DIGITAL VIDEO
both image quality and audio quality of the less-compressed version were superior and clearly gave a more professional appearance to the video. Within our departmental Intranet, lag secondary to transmission time was negligible, even with the larger file size. Two participants with broadband Internet access (cable modem) from home tested lag times and expressed a preference for the higher-quality version; they believed that the 5- to 10-second delay before commencement of playback was acceptable. We also found that resourceful techniques with placement and activation of the video snippet could virtually eliminate perceived delays. For example, if the author embeds the video within a web page and places the video below other components of content, such as text or static images, the video will then transmit while the learner is reviewing those components and will usually be fully loaded and ready to play by the time the learner reaches it. However, if the author instead hyperlinks a component of content to a video snippet (which typically will then play independently in a media player on the learner’s computer), the learner must wait for the media player to download some of the video after clicking the hyperlink. We found some additional factors that contributed to obtaining high-quality video with the camera we used. A stabilized camera was essential, but “stabilized” does not imply motionless. On a tripod, the camera can be smoothly panned, and with only minimal practice, the field of view can be effectively zoomed in and out during recording. Occasionally (depending on the subject matter) the camera’s auto-focus mechanism would perform an adjustment when not necessary and very briefly blur the image. Because in many circumstances the distance between subject matter and camera was constant, we could use the auto-focus system to initially focus the image and then switch to manual focus, effectively freezing the focus at its current setting. Although editing or replacing the sound track during postprocessing is always an option, one of our objectives was to acquire audio of high enough quality to avoid the need for this. The camera’s microphone seemed particularly susceptible to an acoustically “hard” environment with background noise, such as noisy air-conditioning or computer equipment. However, in a relatively quiet room with the subject facing the camera and projecting his or her voice, the audio was very satisfactory. Off-screen narration performed by the operator of the camera worked well. We found that using exclusively one technique or the other within the same snippet was preferred because the operator’s voice would record substantially louder than would a subject speaking
1147
FRANK AND GUNDERMAN
several feet away. Our occasional mistake of forgetting to set the camera’s white-balance to match the ambient lighting provided some vivid illustrations of the role that white balance plays. Fortunately, the nature of the videosnippet approach provides a forgiving environment. Few web pages will contain more than one embedded video, and most pages will contain none. We discovered that our “white-balance memory” is relatively short when snippets are encountered infrequently, and small discrepancies in white balance from one snippet to the next were not felt to detract from the educational value of the content. CONCLUSION Newer-generation digital cameras are useful for quickly and inexpensively producing high-quality, web-
1148
Academic Radiology, Vol 11, No 10, October 2004
ready digital video suitable for on-line education. While specific techniques and methods of implementation will no doubt vary among and within departments, the use of video snippets will likely grow in popularity as increasing numbers of departments pursue initiatives in on-line education. REFERENCES 1. Arenson RL. Teaching with computers. Radiol Clin North Am 1986; 24: 97–103. 2. Hennessey JG, Fishman EK, Ney DR. Digital video applications in radiologic education: theory, technique, and applications. J Digit Imaging 1994; 7:85–90. 3. Chou MT, McGinnis P, Tello R. A web based video tool for MR arthrography. Comput Biol Med 2003; 33:113–117. 4. D’Alessandro MP, Galvin JR, Erkonen WE, et al. An approach to the creation of multimedia textbooks for radiology instruction. AJR Am J Roentgenol 1993; 161:187–191.
New Opportunities for the Use of Digital Video in On-Line Radiologic Curricula 1 Mark S. Frank, MD, Richard B. Gunderman, MD, PhD
Rationale and Objectives. To explore techniques to conveniently and self-sufficiently create high-quality, web-ready instructional digital video snippets suitable for routine use in on-line radiologic curricula. Materials. A commercially available digital camera with an 8-megapixel image receptor and the capability to record webready digital video and audio at a resolution of up to 640 ⫻ 480 pixels and a frame rate of up to 30 per second was used to obtain video snippets intended for inclusion in on-line curricula. Hand-held and tripod techniques were compared. Evaluation focused on the types of snippets deemed most likely to be used within on-line educational content, ranging in length from 10 seconds to 2 minutes. Additionally, basic postprocessing functions to experiment with combinations of video-file size, format, transmission efficiency, and image quality were used. Results. The overall video quality was considered by participating radiologists to be good to excellent for its intended purposes. For most situations tested, a matrix size of 320 ⫻ 240 pixels provided a good balance of visual quality versus file size and transmission overhead. The 640 ⫻ 480 format was occasionally optimal, but was usually larger than necessary and resulted in substantially larger files, especially at a rate of 30 frames per second. A rate of 15 frames per second was considered adequate for most situations, regardless of matrix size, although it did add some barely discernible choppiness to the display. For all but very short snippets, a tripod or some other form of stabilization was necessary to eliminate distracting “camera shake” associated with free-hand acquisition. With a tripod, appropriate ambient lighting, and a favorable acoustical environment, videos of excellent quality when using the highest settings available on the camera (640 ⫻ 480, 30 frames/sec) could be self-sufficiently produced. However, such high-quality video with its associated large file size was rarely the optimal fit for the needs of this curricula. Conclusion. These results indicate that newer-generation digital cameras are useful for quickly and inexpensively producing high-quality, web-ready digital video suitable for use in on-line education. Key Words. Digital video; radiology education; online learning; teaching file. ©
AUR, 2004
The potential for using multimedia in radiologic education is well known (1,2), and digital video is widely used today to distribute radiologic curricula on the World Wide Web. Video is often transmitted in a stream that encom-
Acad Radiol 2004; 11:1144 –1148 1 From the Department of Radiology, Indiana University School of Medicine, 714 N Senate Ave, Suite 100, Indianapolis, IN 46202. Received March 22, 2004; revision requested May 24; revision received July 13; accepted July 15. Address correspondence to M.S.K. e-mail: [email protected]
© AUR, 2004 doi:10.1016/j.acra.2004.07.004
1144
passes an entire lecture or large components thereof, frequently displaying both a presenter’s materials along with a small inset video of the presenter as he or she speaks. Some people believe that this is an effective approach to on-line education, while others believe that this format is less effective than simply providing parcels of text and images to convey the same information. Regardless, we believe there is yet another way to effectively and routinely use digital video for on-line education. Rather than using video as the dominant form of media within on-line content, we have experimented with using relatively short segments of digital video interspersed and complementary
Academic Radiology, Vol 11, No 10, October 2004
with other components of content such as stylized text, still images, and interactive questions with immediate feedback. We have investigated ways to self-sufficiently and inexpensively create high-quality digital video snippets and then integrate them into teaching-file cases and other on-line curricula in our department (Department of Radiology, Indiana University School of Medicine, Indianapolis, IN). We report on our results and the surprising ease with which this can be accomplished when a suitable content-management infrastructure is in place.
MATERIALS AND METHODS We chose the Sony DSC-F828 digital camera (Sony Inc, New York, NY) with a market price of approximately $850. It contains an 8-megapixel image receptor and the capability to record motion picture experts group (MPEG) format web-ready digital video. In video mode, the camera will commence recording when the exposure button is pushed and will continue recording until its memory card is full or until the exposure button is again pushed. Compact-flash, micro-drive, and memory-stick storage are supported. The camera can record at a resolution of up to 640 ⫻ 480 pixels and a rate of up to 30 frames per second (FPS). The camera’s auto-focus system and zoom lens remain operational during video recording. The auto-focus system provides settings to adjust for different recording circumstances, and a manual-focus mode is also supported. White-balance settings facilitate rapid color adjustment for different sources of ambient lighting (eg, sunlight, incandescent, fluorescent). The camera has a built-in microphone but no external microphone jack. We created video snippets in three categories that we deemed most likely to be adjunctive to other forms of online digital content such as still images and stylized text. These were: (1) a brief video of a person introducing or emphasizing something, typically no more than 30 seconds in length; (2) a video snippet ranging from 10 seconds to 2 minutes that demonstrates a technique, such as how to perform a portion of a procedure; and (3) a video snippet that demonstrates an environment or a situation, such as a panoramic view of an angiography suite. Because the camera will record at either 640 ⫻ 480 or 160 ⫻ 120 pixels, but not at 320 ⫻ 240, all recording was performed using a resolution of 640 ⫻ 480 at either 15 or thirty FPS. Some basic postprocessing was performed. When needed for a particular facet of our experiment, we would resize a snippet from 640 ⫻ 480 to 320 ⫻ 240 and when necessary remove from
NEW OPPORTUNITIES FOR DIGITAL VIDEO
a snippet any excess leader or trailer. Video snippets were produced in four combinations: 640 ⫻ 480 at 30 FPS; 320 ⫻ 240 at 30 FPS; 640 ⫻ 480 at 15 FPS; and 320 ⫻ 240 at 15 FPS. In some circumstances we would also anneal some very short postprocessed snippets together to form one longer, continuous snippet. We generated video files in three different formats (Audio-Visual Interleave, MPEG, and Windows Media Video [WMV]) to determine which format seemed most effective for use on the World Wide Web. We experimented with both Camtasia Studio (Techsmith Inc, Okemos, MI) and Microsoft Movie Maker (Microsoft Corp, Redmond, WA) for performing these postprocessing steps. Our department’s digital-education software framework supports the storage and on-line use of digital video (with or without an audio track), digital audio, and static images. The framework contains an authoring system that facilitates construction of interactive case-based exercises as well as thematic curricula, and provides straightforward ways to intersperse video snippets within. Such course work is typically composed of various digital components (eg, stylized text, interactive questions that provide immediate feedback, media objects such as automatically zooming sequences of thumbnail images, and image stacks that simulate a PACS workstation). The software facilitates the construction and arrangement of these components to meet the author’s specific teaching objectives, the results often spanning several web pages. The system supports both direct embedding of a video within a web page and the capability to hyperlink content (eg, fragment of text, thumbnail image) to the default video player invoked as a stand-alone application on the learner’s computer. We placed video snippets within various forms of educational content and asked four participants (two radiologists, one production assistant, one pathologist) to subjectively assess both the quality and potential effectiveness of the videos.
RESULTS In general, our four participants were very pleased with the quality of digital video produced. When obtained in an environment with adequate ambient lighting, the consensus was that image quality at 640 ⫻ 480 pixels and 30 FPS was better than necessary for most on-line teaching applications. A technique of 320 ⫻ 240 pixels at 15 FPS was considered ideal for all head-and-shoulder videos of a speaker and for most “how-to” demonstrations. For demonstrations containing detailed structures
1145
FRANK AND GUNDERMAN
that also encompassed a wide field of view, a matrix size of 640 ⫻ 480 at 15 FPS was considered ideal. For demonstrations encompassing a narrow field of view, the camera lens could be dynamically zoomed to areas of interest, and a matrix size of 320 ⫻ 240 was then satisfactory. Similar perceptions were expressed for the same reasons when acquiring panoramic scenes such as a reading room or angiography suite. Although “full motion” 30 FPS recordings, when obtained in favorable lighting, provided near-professional quality video, consensus was that this did not add to the educational value of the content, primarily because the 15 FPS setting worked very well. In general, we found that the WMV file format provided the optimal balance of compression and image quality when compared with Audio-Visual Interleave and MPEG formats. We found that that Microsoft Movie Maker (a free accessory included with the Windows XP operating system) was more simple to use than Camtasia, but provided fewer options regarding the format and size of an output file. Both software applications provided a user-friendly interface for trimming the ends of a snippet and for annealing several snippets together. Movie Maker was particularly convenient for producing the most frequently preferred output in our experiment (WMV format, 320 ⫻ 240, 15 FPS). However, for video that demanded higher resolution and more control over image quality, we preferred the additional features in Camtasia’s rendering module. We also found both Movie Maker and Camtasia to be superior to the video-editing software bundled with the camera. With either application, a person familiar with the software could transfer an MPEG file from camera to computer, perform the basic editing described above, and produce a web-ready snippet, all in 2 or 3 minutes. Although a well-done MPEG video as obtained from the camera could be used on-line as is, we chose to transform each snippet into a substantially smaller file (usually WMV format) with no perceptible loss in video quality. When audio was recorded by an off-screen narrator (typically the person behind the camera), sound quality was judged to be good. However, when a speaker was the centerpiece of the video, it was necessary for that person to project his or her voice as if giving an unamplified presentation to an audience, and in such circumstances sound quality was influenced to a much greater degree by surrounding acoustics and ambient noise. For all but snippets of only a few seconds duration, partici-
1146
Academic Radiology, Vol 11, No 10, October 2004
pants felt that using some technique to steady the camera was critical to eliminate camera shake, which in addition to being annoying was also considered to negatively influence the overall credibility of the recording. Although a tripod provides the greatest flexibility for alleviating this problem, in some circumstances we found that simply bracing the camera against an available immobile structure was adequate.
DISCUSSION We are not the first to investigate the possible uses of multimedia in on-line radiologic education (1– 4). Our intent was to look more deeply into the feasibility of creating short segments of high-quality digital video selfsufficiently, inexpensively, and essentially on-demand, yet fully suitable for insertion into web-centric radiologic curricula. We are not making recommendations regarding digital cameras, computing platforms, or specific software products for on-line education. To those who primarily use Macintosh or Linux computers, we acknowledge that our experiment involved only computers running Microsoft Windows because it is the pervasive computing platform in our department. Regardless, we believe that the promising results we convey are independent of brand-specific factors. One advantage of using long segments of streaming video as the primary media for on-line curricula is that once the video is produced, few additional content-authoring or content-management capabilities are required to provide it as on-line course work. That is, the video itself conveys all of the educational content, and a simple web page is adequate for housing the video or hyperlinking to it. Some commercially available software products facilitate the integration of a soundtrack with sequences of static images (eg, Powerpoint slides), with an optional accompanying inset video of a speaker as well. Although it might be a labor-intensive endeavor, once produced, such an entity can virtually serve as a stand-alone component of course work. This approach can effectively emulate a lecture, but we believe it represents a relatively instructor-centric model of teaching, and when it becomes necessary, updating a long video can be nearly as labor intensive as originally producing it. Another approach to on-line education is to create curricula with the intention to have it interact with, challenge, and track the learner as the learner works through the course work. We regard this approach to be more Socratic and learner-centric, and
Academic Radiology, Vol 11, No 10, October 2004
streaming video seems to be less effective for this. We use the authoring capabilities of our digital education framework to assemble different forms of content such as stylized text, dynamic image sequences and stacks, and interactive questions with immediate feedback. These components are sequenced, interlinked, and incrementally revealed on-line to meet the specific objectives and desired teaching style of the educator. Such content is easily updated by the educator or a secretarial designee. Here is where we believe that short snippets of digital video can play a particularly enriching role, sometimes demonstrating in seconds what might be nearly impossible to put into words and often surpassing what could be demonstrated with several static images. For example, this approach could be useful when teaching fluoroscopic techniques or interventional procedures. The expectations set for our investigation were: 1. Snippets of video should be easy for an amateur photographer to record. 2. The format of the video produced by the camera should be nonproprietary and essentially web-ready. 3. The quality of video produced should be reasonably high, assuming that ambient lighting and acoustics are appropriate. 4. The snippets of video should be short enough to record with a single “take” (or two) and would therefore require little, if any, editing. 5. Any postproduction steps would be simple, relatively fast, and convenient to perform. 6. The end result could serve to complement and enhance the effectiveness of our on-line curricula. Meeting these expectations implies a form of pedagogical empowerment (ie, that somebody facile with the camera and some simple postproduction steps on a personal computer could single-handedly produce high-quality digital video snippets useful in on-line education). In general, our experience indicates that this implication is justified; however, we believe that there are some helpful caveats. The postprocessing steps we applied were simple to perform yet provided considerable latitude for balancing size versus quality of the final product. For example, a 30-second 320 ⫻ 240, 15 FPS WMV file of a speaker rendered with high compression requires approximately 500 KB of storage, while a less compressed version of the same file format rendered from the same 11-megabyte MPEG file (from the camera) requires approximately 3 megabytes. While the smaller version was functional,
NEW OPPORTUNITIES FOR DIGITAL VIDEO
both image quality and audio quality of the less-compressed version were superior and clearly gave a more professional appearance to the video. Within our departmental Intranet, lag secondary to transmission time was negligible, even with the larger file size. Two participants with broadband Internet access (cable modem) from home tested lag times and expressed a preference for the higher-quality version; they believed that the 5- to 10-second delay before commencement of playback was acceptable. We also found that resourceful techniques with placement and activation of the video snippet could virtually eliminate perceived delays. For example, if the author embeds the video within a web page and places the video below other components of content, such as text or static images, the video will then transmit while the learner is reviewing those components and will usually be fully loaded and ready to play by the time the learner reaches it. However, if the author instead hyperlinks a component of content to a video snippet (which typically will then play independently in a media player on the learner’s computer), the learner must wait for the media player to download some of the video after clicking the hyperlink. We found some additional factors that contributed to obtaining high-quality video with the camera we used. A stabilized camera was essential, but “stabilized” does not imply motionless. On a tripod, the camera can be smoothly panned, and with only minimal practice, the field of view can be effectively zoomed in and out during recording. Occasionally (depending on the subject matter) the camera’s auto-focus mechanism would perform an adjustment when not necessary and very briefly blur the image. Because in many circumstances the distance between subject matter and camera was constant, we could use the auto-focus system to initially focus the image and then switch to manual focus, effectively freezing the focus at its current setting. Although editing or replacing the sound track during postprocessing is always an option, one of our objectives was to acquire audio of high enough quality to avoid the need for this. The camera’s microphone seemed particularly susceptible to an acoustically “hard” environment with background noise, such as noisy air-conditioning or computer equipment. However, in a relatively quiet room with the subject facing the camera and projecting his or her voice, the audio was very satisfactory. Off-screen narration performed by the operator of the camera worked well. We found that using exclusively one technique or the other within the same snippet was preferred because the operator’s voice would record substantially louder than would a subject speaking
1147
FRANK AND GUNDERMAN
several feet away. Our occasional mistake of forgetting to set the camera’s white-balance to match the ambient lighting provided some vivid illustrations of the role that white balance plays. Fortunately, the nature of the videosnippet approach provides a forgiving environment. Few web pages will contain more than one embedded video, and most pages will contain none. We discovered that our “white-balance memory” is relatively short when snippets are encountered infrequently, and small discrepancies in white balance from one snippet to the next were not felt to detract from the educational value of the content. CONCLUSION Newer-generation digital cameras are useful for quickly and inexpensively producing high-quality, web-
1148
Academic Radiology, Vol 11, No 10, October 2004
ready digital video suitable for on-line education. While specific techniques and methods of implementation will no doubt vary among and within departments, the use of video snippets will likely grow in popularity as increasing numbers of departments pursue initiatives in on-line education. REFERENCES 1. Arenson RL. Teaching with computers. Radiol Clin North Am 1986; 24: 97–103. 2. Hennessey JG, Fishman EK, Ney DR. Digital video applications in radiologic education: theory, technique, and applications. J Digit Imaging 1994; 7:85–90. 3. Chou MT, McGinnis P, Tello R. A web based video tool for MR arthrography. Comput Biol Med 2003; 33:113–117. 4. D’Alessandro MP, Galvin JR, Erkonen WE, et al. An approach to the creation of multimedia textbooks for radiology instruction. AJR Am J Roentgenol 1993; 161:187–191.