Always trying to come up with clever headlines for these blogs, today I settled upon Random Access Memory. We in the video business are always talking about RAM as it relates to using multiple programs at once, having enough juice to run your software. But in this case I am talking about the availability of information in a randomly accesible fashion.
Take a database, or a simple Excel speadsheet - randomly accessing desired information is as simple as a search or a query. However as we all know, the specificity of the search is limited by the details of the data stored.
For example, if I have a database of oh, I don't know, surgical videos. We store the name of the procedure, the author/s, the year produced, the running time, the available formats and a brief description of the video, either something written by the author or the first few lines of narration.
Thus, i can search for: Cholecystectomy AND Jones to find videos on Cholecystectomy by anyone named Jones.
But what if I want to find a video in which the gallbladder is removed using the top-down approach, which is less commonly performed. If the term top-down is not included in the title or the description, we are out of luck.
Thus, we arrive at the very popular subject of Metadata. For any new record in the database, ideally, one would want to add more keywords, additional description of the video not submitted by the author, and anything else that might be useful for future reference. There now exist numerous software solutions for tagging video with metadata, and presumably the XML can be inserted into a database, although the softwares probably include searching abilities.
So, a more ideal record for a video would be:
Laparoscopic Cholecystectomy
Author, George Harrison, MD Date, 1998 RT, 6:18 Format, BetaSP
Description, A patient with recurrent cholecystitis presents with diffuse abdominal pain, guarding and elevated WBC. We demonstrate our standard technique for laparoscopic cholecystectomy using 4 ports, an articulating clip applier and endoscopic retrieval bag.
Keywords, Veress needle, snake retractor, top-down approach, cholangiogram performed, xyz inc. trocars, wxyz-o-matic retrieval bag
Additional description, The authors take the gallbladder down using hook cautery, then switch to ultrasonic dissection (device name) for the cystic duct and artery. Etc.
So how does this relate to the price of bread?
Presently, the only metadata exists in the brains of our production team. Since one or more of us have been on hundreds of shoots over the years, we tend to remember selected important or unimportant metadata about each experience(type of abdominal access performed, temperature of the OR, did the case start on time, where did I park my car?). The amazing human brain has instant access to some of this data, and delayed access to other data, depending upon coffee intake, sleep, age frequency of performing crossword puzzles, and other factors. Needless to say a computer is not prone to fluctuations in data availability, unless you have some flaky software or hardware, a failed disk, inadequate backups or poor processes. Ok, so maybe the brain and the computer are closer in reliability than we thought, only for different reasons. However there is of course a distinction between the brain's ability and system for data management and that of a computer. A computer needs to be told how to manage data - the brain works from the factory (out of the package?). Batteries not included, your parents put it together.
Here is a hypothetical example:
1 - I need some video showing the extracorporeal ileorectal anastomosis using a circular stapler during a total colectomy.
A - Search the database - we may or may not get a result because the detail we needmore than likely is not in the title (the type of stapler and anastomosis may in fact be incidental to the overall procedure) and not in the description (the descriptions vary between "The video presents the case of a 45 year old man with primary adenocarcinoma..." and "This video depicts our technique for total colectomy, as performed on 340 patients..") See what I mean
B - Search the human database - "Hey, does anyone remember a video showing a circular stapled external anastomosis in the colon?" "Oh yeah, didn't Dr. Jetson do that last year during the live teleconference we produced?" "I think you might be right. I worked with Dr. Flintstone last year and he may have done that too." So I have two videos to check - in either case you would still have to eyeball the footage to know if it is what you need.
A solution for the above dilemma is to add metadata to all 2000 existing records - this is not so easy, because someone with some surgical expertise would be required to view each video and come up with the data to enter.
Moving forward, new records would be easier to populate, due to the availability of the original author and/or producer of the project.
Institutional memory, however remains a viable option as a complement to an electronic system.
Thanks for reading.
Mike Cohen
Digg it
When Walter Murch starts to
When Walter Murch starts to put the hundreds (and sometimes thousands) of shots together to make a movie he relies on a storyboard with a frame or sometimes up to 10 frames of pictures taken from that particular shot that captures the essence of the contents; something that sets it apart from the rest. This way when he scans his storyboard for all takes of the same shot in an instant the pictures will jog his memory into remembering what set that particular take apart from all the others and help him decide if that's the one he wants. He says that pictures helps him recall the take much more readily than metadata because of how the mind works and perceives things.
Non linear editing has become too good and too exact, because when you call up "Scene 12 Take 24" the computer will give you exactly what you ask for and in the process you might perhaps miss something from one of the other takes that would be better suited for what you had in mind.
Walter goes on to add that there's no use scrolling through all the takes at say, 10 times speed because the NLE will drop frames to keep up. More often than not the frames the NLE dropped are crucial to the essence of the shot.
Perhaps adapting Walter's storyboard idea to become part of what you're trying to achieve would go a long way to help find the needle in your medical haystack.
- Don