I could see two ways of designing this
Option one would be to design a new cautery device (hook, for instance) that has a light sensor on it. The amount of light that it senses would be used to determine whether or not its in the camera's field (I'm assuming that laparoscopic cameras are calibrated to have their light align with their visual field) The light sensor would regulate whether cautery can be activated. If its in the dark, it's probably not on the operator's screen and it shouldn't be activated. If its in the light, it's probably is the surgeon's field and is safe. This system might require calibration to be accurate with sensing the limits of the camera, but it has the advantage of only requiring one new piece of technology (a new hook)
Option two would be to have an in-line video processing unit that the camera plugs into, which analyzes the video in real time and identifies whether or not the cautery instrument is present on the screen. This unit would communicate with a second unit (on the cautery) to give a GO or NO GO on whether its safe to fire. This option requires more computing (ie a computer has to recognize what a hook cautery looks like) and requires two pieces (camera piece and cautery piece) that can communicate. It has the added benefit of increase precision in identifying if the cautery is on the screen.
Anyone else have thoughts? Do you think hospitals would invest in a technology like this for patient safety? Are there any other safety mechanisms that we could "hard wire" in to make surgery safer?