Section 4: Scenarios: Examples of the use of Digital Video in Teaching and Research

Department of General Medicine, University of Bristol

Within the department of general medicine, especially within the area of endocrinology, virtually all patients are now seen as out patients. This means that students do not gain experience of the range of cases that is required. A solution to the problem has been to video the patients during referrals for viewing at a later date. Patients visit the consultant as normal and are captured to videotape using a small, unobtrusive Hi-8 camera and microphone. The manner in which the video is captured is low-key and informal, and does not interfere with the referral procedure. This allows patient/consultant discussions to occur naturally without altering the dynamics of the consultation. The use of video in this scenario allows important visual considerations to become more apparent. Much of our everyday communication is non-verbal, and culturally we 'read' such things as facial expressions, hand movements, posture, etc. The video material offers a window onto the non-verbal communication and presents a holistic approach to the consultation.

"....visuals of some sort and variety are the main vehicle of expression and communication. Consider how influential visuals such as facial gestures and other body movements (usually referred to as nonverbal communication ) are in face-to-face conversations and social interactions"

Patients are filmed over a period of time, before, during and after treatment. The video is digitised using a motion JPEG capture card and Apple's Quicktime. A typical film from a single consultation uses 3 GB of disc space. The required sequences are edited and assembled in Adobe Premiere and exported back out to tape. A typical video is 8 minutes in length. No voice-over or narration is added as this may prove influential and detract from the naturalness of the consultation.

This scenario deals with the teaching of disease manifestations that are uncommon, with symptoms gradually becoming more apparent with time. To represent the passage of time, a technique known as "Morphing" is used. Morphing is a process which merges short sequences (or still images, graphics and other visual elements) into one another. Adobe Premiere has a number of special effects of which Morphing is one. Merging these sequences in quick succession has resulted in a powerful educational tool. Text and images alone could not have had such an impact, and this example highlights benefits of the immediacy of video when combined with techniques such as morphing.

Users (trainee solicitors) are presented with the first video sequence from one of the scenarios, eg interviewing a suspect at the police station. As the sequence plays, users interrupt the video where they think an alternative course of action should have been taken by the solicitor. Users are invited to enter the alternative course of action and are then presented with a model answer against which they can score themselves.

The scenarios are filmed and edited using Betacam SP video systems. Two specifications of delivery platform were determined at the outset of the project:

1. A mixture of 486 DX2 66 MHz and Pentium PCs present within the department's computer assisted learning laboratory (CAL Lab)
2. Dual speed CD-ROM

Initially the video was captured at 25 frames per second at 320 x 240 pixels using the VideoBlaster RT 300 capture card. The sequences were compressed in real-time with Indeo 3.2 and with a key frame every 10 frames. This decision was reached by digitising the first two sequences of the first scenario above at different window sizes, frame rates and key frames. Ideally, the video should have been captured raw. However, our PC and card at that time was not capable of raw capture at 25 frames per second. For the next two scenarios (annual general meeting and briefing of expert witnesses), the video will be captured raw using a Pentium 133 MHz with 32 MB RAM. By capturing raw video, the video files can be re-compressed and optimised for delivery from dual-speed CD-ROM at a later date.

In this scenario there is also the sound track to be considered. The sound was captured at 22 kHz, 16 bit, momo as this gave good sound quality at the available bandwidth. The sound track will be compressed when optimising for CD-ROM delivery.

Any soundtrack has to share the available bandwidth with the video and there are often tradeoffs between the quality of the video and sound track. This can lead to problems with lip sync with the audio track often lagging behind the video.

"An experienced TV viewer (i.e. most people) will immediately spot if they are in conflict. An obvious example is lip-sync sound. Deliver lip-sync video at 10 frames a second with audio lagging behind, and the disjointed look detracts from the quality of the communication. The viewer spends more time wondering what is wrong with the picture, than listening to the words".

The article, however, goes on to say that a good interactive designer would have defined this issue before it became a problem by understanding the audience, the demands of the communications required and the constraints of the technology which was being selected. In short - lip sync would not have been used.

Evidence from formative evaluations of the DPLS project, however, showed that although users were initially put off by the sound being slightly out of sync with the video, the user's attention was soon detracted from this because of the quality and completeness of the information being conveyed. What this perhaps demonstrates, is how, with good instructional design, a deviation from the 'norm' may challenge our expectations, or notions. It is also important to realise that a generation brought up with television will have certain expectations when viewing digital video. However, television and computers should not be in competition with each other. Each provides a separate channel for communication and has its own set of issues and criteria. Viewing from television and computer screen are different: for the former video is viewed at a distance, the latter, close up.

1. together
2. in separate bed/cot

The mother and baby stay overnight in a room that is equipped with an infra-red sensitive video camera (and infra-red light). Both mother and baby are connected to equipment that measures and records various physiological data (e.g. respiration; heart rate; temperature; blood gases). Currently the physiological data is recorded onto computer, while the video output, overlayed with the physiological traces, is recorded onto video tape. A typical study will use 4 x 195 minute tapes.

The physiological data (approx. 50-100MB) stored on hard disc can be easily accessed for fast searches, reviews and reporting. Physiological data is backed up to tape. The video data is reviewed and any movements by mother or baby coded. This takes approximately 15 hours, even using accelerated playback, for a typical 10-12 hour study. The video tapes have no backup.

There are a number of problems with the data being stored in this way:

• Raw physiological and video data are stored separately
• Retrieving physiological data from backup tapes is not easy
• Accessing a specific part of a specific tape is time consuming
• Tape storage
• Sharing the data, for teaching and research means duplicating video tapes and finding a suitable method for distributing the physiological data

We considered whether or not digitising the video signal could help with some of the above together with the possibility of storing all data on CD-ROM.

As a first step some questions were asked

Do we need all the data?

Two immediate ways of reducing file size are:

• as the video signal is greyscale, 8 bit capture is sufficient
• at smaller frame sizes the overlayed physiology data is not easy to read and poses additional problems for the codec. Given that the physiological data is being recorded separately, it was decided not to capture with the physiological data overlayed.

Should the videotape record be continued?

What interval and frame size should be chosen for the 14 kB per second limit?

A frame size of 240 x 480 at 1 frame per second was decided on. It may turn out that too great a safety margin has been built in. It may therefore be possible to increase the frame size to 320 x 240 at a later date. This would probably prove more useful than increasing the frame to 2 frames per second.

Having taken these decisions the following benefits became apparent:

• As the same computer would be performing both the video capture and data acquisition, lowering the video data rate to such an extent meant that the PC would not be overloaded by data throughput. Jitter on the physiology data is unacceptable. At the lower frame rates this is no longer noticeable.
• Playing back the video at an acceleration of 10-20fps also accelerates the review of the nights activities by the same amount.
• It becomes possible to present overviews, e.g. displaying 10 frames taken at 60 frame intervals across the screen (at reduced resolution) would give a 10 minute overview. There is also the ability to zoom in or out in time and/or pixel resolution while showing the corresponding view of the physiology data.

Other considerations and requirements:

• The capture card has to be capable of capturing/saving the video straight into an AVI file which can then be played back without the capture card being present. Compressing 35,000 plus frames is a very time consuming task.
• The capture must be able to do interval (frame by fame; time lapse) capture. It should also yield gracefully to other processes, in this instance the physiology data capture. Check if the capture driver supports standard Video for Windows calls (or that the suppliers will release the API information) should you need custom capture software.
• To minimise head thrashing, two external hard discs (SCSI, in a shared case) are used; one for video, the other for the physiology data. These can then be unplugged and taken to a CDROM writing PC, where they will be written to the same CD.

Graphics     Multimedia      Virtual Environments      Visualisation      Contents