Cooperative Tele-Surgery

Welcome to my part of the tele-surgery project at MIT. The Under Contruction sign is likely to be a permanent feature here, as I'll be updating the page as my research progresses. Enjoy!

-- Mark Ottensmeyer --

Overview

In most tele-surgical procedures, some form of assistance will be available from someone else at the scene. This person might be an army medic (battlefield), ambulance attendant, or a companion of the patient. Given that this person experiences none of the disadvantages of the tele-surgery system (time delay, imperfect visual/audio/haptic feedback, etc.), but may not have surgical skills, what is the best way to allocate tasks and coordinate the motions of the surgeon and assistant to produce the best possible surgical outcomes?

Further, is there a general framework that can be developed which will use tele-operator characteristics, type of procedure, competance of assistant and other factors as inputs and define the interaction as it's output?

Focus

The experiments that will be performed will revolve around laparoscopy, a form of minimally invasive abdominal surgery. Laparoscopic techniques are already used for gall bladder removal, appendectomies, gynecological procedures and other operations, and are being evaluated as a method of minimally invasive inspection to avoid unnecessary laparotomies. (references will be added at a later date) In the field, laparoscopic techniques could be used to inspect the inside of the abdomen for damage (e.g. did a bullet wound enter the peritoneum or not?) and perform preliminary repairs (e.g. stop bleeding with clip applier or approximate tissue with stapler).

In terms of cooperation, laparoscopic procedures typically involve the simultaneous use of three or more instruments (e.g. laparoscope, probe or gripper and shears or other cutting tool). Because of this, at least one tool must be operated by the assistant, as the tele-operator system will likely have only two surgeon operated arms. Thus, laparoscopy provides a valid model to study cooperation between a tele-surgeon and a local assistant.

Current Tasks

Experimental

Experiments with surgically naive assistants (likely MIT seniors and grad students) will begin in the fall. The experiments will focus on a suite of practice tasks that are used by surgeons new to laparoscopy, to master the skills needed for the task (Steele et. al., 1994). Some of these are precision manipulation with grippers, application of surgical clips, and as a final task, we'll look at performing dissection of some sort of animal tissue.

To measure the competance of the assistants, they will perform manual dexterity tests before they participate in the surgical tasks.

Hardware

The teleoperator system that we are using in our project is a pair of Phantom haptic interface arms, developed at MIT's AI lab and built by SensAble Devices. The arms are kinematically identical, making basic control of the system simple. The control strategies that will be used in the experiments are being developed by Hu (link soon), also in the HMSL.

(full size: 17k and 26k jpeg)
Master and Slave Phantom Arms--Note tool handle and tool

I'm in the final stages of designing, building and interfacing a tool handle and actuator pair that will work in conjunction with our teleoperator system. I've planned the actuator so that it will be possible to quickly and easily switch from one type of tool to another--e.g. dissecting forceps to scissors. When complete, we'll be able to use any of the tools with the master-slave system to evaluate which ones should be operated by the surgeon and which by the assistant.

(full size: 44k and 31k jpeg)
Master tool handle and slave tool--full size images include extra detail

As a stage for our experiments, I built a laparoscopic surgical trainer, which includes an opaque surface to obstruct direct viewing, and a number of ports for cannulae and surgical instruments. The trainer also houses the emitter unit for a Polhemus 6 DOF position sensor. Using the Polhemus, it is possible to record the position and orientation of an instrument within the box.

(full size: 50k jpeg)
Laparoscopy Experiment Box

Software

I've developed a few graphic overlays which will eventually be used in predictor displays. The real analogs to the overlay are the Phantom master, and an interior view of the surgical trainer, from the viewpoint of the laparoscope.

The overlays are being implemented on an SGI Indigo2 with a Galileo video card which allows us to combine video images and computer generated graphics. I've networked the SGI with a Pentium housing the Phantom control card and software so the motions of the Phantom drive the overlay. The Polhemus is run from a '486, also networked with the SGI, and sends the position and orientation data of the CCD camera to the overlay program.

The overlays will permit us to examine the usefulness of a predictive display in dealing with the problem of time delay.

Completed and "back-burner" tasks

A preliminary experiment involving a "tele-expert" and a "novice" cooperating on a geometric puzzle task was performed. The results showed that fewer errors were made when the expert used a pointer (substitute for tele-operator) and vocal directions than when only vocal directions were used (more detail will be included later).

(full size: 42k jpeg)
Point Placement and Line Tracking Apparatus. Polhemus field emitter mounted under board, sensor at end of tool (both are blue). Probe is a modified laparoscopic clip applier, shown with 10mm cannula.

I've built and tested equipment to test probe placement accuracy. Using a Polhemus 6 degree of freedom position/orientation sensor it will be possible to measure point placement or line tracking accuracy.

A literature search is continuing, to combine knowledge from a number of different related fields, because very little research examining cooperation on manual tasks has been done. (If you know of some, please let me know by e-mail! Thanks)

That's it for now. From here you can return to the main Human Machine Systems Lab Page. Comments or questions? We have a MAIL FORM for your responses.

-- Genesis: Jan. 7, 1995 -- Last Update: Sept. 5, 1995 --
-- Created by: Mark Ottensmeyer -- e-mail: markott@mit.edu --
-- Disclaimer --
If you hold a copyright that I don't know about on material that I've used here, give me a shout and I'll try to fix things.