4

First of all, I'm a biologist, so please let me know if this question is being asked in the wrong place. I regularly perform operations on rats and recently stumbled upon the following paper (http://ar.in.tum.de/pub/padoy2008IAAI/padoy2008IAAI.pdf) where surgical activity is classified primarily based on the use of specific surgical tools. I thought this would be a very interesting thing for me to do with my own work and may help me be more efficient with my time. I have a good CS background so the use of the required computational models is not problematic. What is a problem though is the fact that the authors of this paper use manual tagging to identify whether a given tool is being used or not. This seems inefficient; the authors admit this, saying automatic tagging is possible.

I am wondering if anyone can suggest a good way to tag the tools being used? My instinct is to use a simple RFID system, where each tool is tagged and the tool tray could contain a reader. This would allow me to collect a binary signal vector from each tool across the duration of an operation (successful read indicates tool is not being used, failed read indicates that it is). Is this a good approach to my problem? If so, could someone recommend a specific tagging system to use? If not, could someone recommend another approach?

As an aside, this project is for my own edification and enjoyment: it is not a part of my funded research, I am not going to publish on it, and I am not going to sell anything I put together.

UPDATE:

Assume that the cost of the sensing system is not a constraint when offering suggestions.

user93189
  • 141
  • 2
  • 4
    Are you willing to have a holder so that each instrument fits in a specific place in a specific way, and then always replace each instrument into its designated holder when done with it? That would simplify things a lot as the instruments would not need to be added to or modified at all. – Olin Lathrop Jan 04 '14 at 21:40
  • 3
    Will support staff be autoclaving these instruments? – RedGrittyBrick Jan 04 '14 at 21:42
  • @Olin Lathrop, I have no objection to using a holder, but I do have space constraints. I have anywhere from 20-100 tools to be placed on two tables, each approximately 3 ft x 2 ft. Not all of the tools are unique though, for any given operation, I will have about 10-20 unique tools. – user93189 Jan 04 '14 at 22:18
  • @RedGrittyBrick, I use a combination of disposable tools and tools that have been previously used (and have been autoclaved). Each is held in a plastic wrapping until the operation. Ideally, I would like to come up with a way to monitor use of disposable tools as well as reused ones. – user93189 Jan 04 '14 at 22:23
  • @OlinLathrop the idea of a holder would be something that surgeons would not be willing to use. Its a pretty big eorkflow invasion. It would also be hard to clean and sterilize – Scott Seidman Jan 04 '14 at 22:46
  • @user93189 If you're storing sterilized instruments in one time use package, then you could perhaps get some RFID stickers and use them on the package itself. Once the package is disposed, you'd get rid of the sticker itself, but at least you'll know which instrument has been used. What would be a bit complicated would be synchronizing one time use stickers with some sort of a database which would connect the sticker to a particular instrument. I've heard of RFID systems being used in warehouses, so there are definitely cheap programmable stickers available. – AndrejaKo Jan 04 '14 at 23:51
  • @AndrejaKo, the plastic wrapping usually contains between 10-40 tools (either disposable or reused). Could this solution still work? If so, how could I recognize the removal of a single tool from the plastic wrapping? Also, for reused tools, after removal from the plastic wrapping, they are not placed back in, but usually placed in a sterilizing solution until reuse during the same operation or cleaning after the operation is over. Could this solution be adapted to tag them? – user93189 Jan 05 '14 at 00:06
  • Should the link to the paper say "viewer discretion recommended?". Anyway, RFID seems the way to go, although I think it may not be desirable to have the RF signal in the place. Computer vision could be used, with a camera on top of the table, the problem is that some instruments are near the shape of the other, but maybe that can be considered for the same type. It has the advantages to not need tags on the instruments, they can go to autoclaving, and disposable instruments will be recognized without putting tags each time you open the package. – Diego C Nascimento Jan 05 '14 at 00:30
  • Not to count the possible contaminant problems with the RFID tags. Even that they don't go on an intrusive part of the instrument, that is a factor of risk depending on the contaminant. – Diego C Nascimento Jan 05 '14 at 00:38
  • @user93189 In that case, I don't think that sticking a RFID tag to the whole set would be very interesting, because I don't think that you could easily track removal of a single tool. Do you perhaps have pictures of the package? Maybe some procedure could be worked out. In any case, I don't think that it would be very convenient. Right now, sticking a tag at each instrument seems to be the best way, but the problem is that cheap and accessible tags aren't really constructed to survive horrors of sterilization and aren't really fit for a sterile environment. – AndrejaKo Jan 05 '14 at 01:14
  • @user93189 I've seen some medical RFID tags meant for implantation, but they are usually in a glass capsule of sorts and some of them might be a bit large, at least compared to size of surgical instruments I've seen. Still, such tags would probably be able to survive anything you can toss at them, as long as there's a secure way to attach them to the instruments. If I can dig up any specific model numbers, I'll be sure to report back. – AndrejaKo Jan 05 '14 at 01:20
  • 1
    @DiegoCNascimento, it seems like the consensus is RFID. I've found a manufacturer that makes custom surgical equipment with built-in RFID tags (http://www.haldor-tech.com/ORLocate%C2%AE_OR.aspx): something like the Mayo Tray or the Back Tray seems to be what I want. With a little work, it seems entirely reasonable to put one of these together on my own. Any ideas for doing this? – user93189 Jan 05 '14 at 01:38
  • @user93189 well, that's not my opinion and is sported by facts and at some level by research not supported by facts, like what RFID or long RF exposure can do. Some studies shows that 2.4Ghz at long exposure time can be hazardous, although there's no fact to support that, many transmission towers has warnings to RF exposure. That's totally opinion dependent to certain power levels. Anyway you seen to have found the solution to the major problem that is contaminant in the RFID tag. – Diego C Nascimento Jan 05 '14 at 02:59
  • @user93189 Few more commercial solutions for tracking surgical tools: [RFID based](http://www.corerfid.com/rfid%20shop/xerafy/rfid_revolutionizing_healthcare.pdf), [bar-code based](http://www.censis.net). – Nick Alexeev Jan 05 '14 at 03:00
  • @user93189 the question I think about RFID, is at some level cost, but you said that this is not a problem for you. For widespread use that probably would be a problem. The other is the time you take tagging each disposable instrument. With "computer vision" you solve these three problems, and only need to "add" an object when it shape or color changes. – Diego C Nascimento Jan 05 '14 at 03:04
  • The bar code link provided by @NickAlexeev says something with contamination using barcode. I has thinking this to, to help identify the object, the unique problem are the disposable objects, if they don't come with the barcode you would need to put, so you need to "tag" it like RFID, and you can contaminate it putting the tag. – Diego C Nascimento Jan 05 '14 at 03:09
  • @DiegoCNascimento, I am skeptical about the immediate practicality of computer vision in this project, though I agree in theory, it is the best solution. First the authors of the paper cited above cite poor resolution with the techniques used. They were unable, as you hinted above may be problematic, to identify certain tools from others. This is a problem. I am also not up to date with the computer vision literature and implementing a custom solution would be interesting, but would take some time. I think that as the field advances, or as I get more free time, this is the best solution. – user93189 Jan 05 '14 at 03:36
  • @user93189 you don't need something much new, and with stereo vision, the detection can be more precise. For non disposable objects the bar codes can be read by the software so there's no need to recognize the objects by the shape. – Diego C Nascimento Jan 06 '14 at 04:55

2 Answers2

1

There are two challenges with using RFID tags, one of which has been covered in the comments.

  1. The tag, presumably, must be autoclavable.

  2. As these tools are metal, the presence of a metal plane so close to the tag will cause problems, or completely prevent operation of normal RFID chips.

As a possible solution, you could use the RFID tags manufactured by Xerafy which are specifically designed for this sort of application and are autoclavable. They implement RFID-on-metal (ROM), so they are designed for tagging metal objects such as surgical tools.

A link to their brochure is here.

There may be other suppliers in this market as well, but this should get you started.

As well as tracking the use of tools, tracking surgical instruments could help prevent them from being misplaced (such as inside a patient).

Spehro Pefhany
  • 376,485
  • 21
  • 320
  • 842
  • Can I ask specifically what problem are you trying to solve? How would "tracking your tool use" help you? Are you trying to make sure you have them all back, or trying to assess how long you spend with each one, ...? Tracking your tools is your solution to some problem, and it would help to know what that problem is. – Scott Seidman Jan 17 '14 at 22:56
1

You could consider using image analysis with a camera positioned over the instrument tray as long as the instruments have distinctive enough shapes.

George White
  • 473
  • 3
  • 8