Good g..d. What could this do to physics job market? Of course the few catastrophes that this saves on a national level will never be "statistically" meaningful.
Focus will be on efficiency. This is not what I do well.
I have a very (surprise!) contrarian view on IMRT QA. The short answer is: it has never saved a catastrophe. (Or, at least, no one can
prove that it ever saved a catastrophe/prevented a disaster.) As I sip coffee this weekend, the long answer is...
In the early days of IMRT QA, each individual field would be port filmed on X-ray film just like the old light box-able port films of old. The physicist would make a dot-matrix-y grayscale print, scaled to 1:1 port film size, of the fluence. The film would be viewed side by side with this and/or superimposed over the print off on the light box. It was all visual analysis. The IMRT field either passed the eyeball test or it didn't. I can't recall a time that a field ever did not pass the eyeball test.
If you're comparing a film "printed" by the linac versus a field printed by a printer, the linac is analogous (analogous to me, anyways) to a color inkjet printer. (In the old days an IMRT parlor trick was to make a port film of Einstein's face,
or anybody's face, e.g.) The linac "sprays" dose into the patient just like an inkjet printer sprays ink onto a piece of paper. In the treatment planning system, we see a dose distribution; on the linac, we try to recreate that dose distribution as close as possible in real life. If you are working with a color photograph on a computer screen and want to print it out, you want the colors on the printout to match the computer screen as much as possible. However the screen and printer have different "
gamuts." It is very interesting, and seemingly entirely coincidental, that the word gamut comes from a combination of the words "gamma" and "ut" (
from music) and that the IMRT QA process looks at a "
gamma."
At some point all physicists abandoned the aforementioned subjective eyeball test in favor of the objective gamma. You know how physicists are: subjectivity makes them very uncomfortable. Using the printer analogy, a catastrophe would be if you were printing a picture of a zebra and instead the printer gave you a picture of an ostrich. However sometimes the images we create on screen are far outside the printer's gamut and when you print that zebra you notice the stripes are dull grey and just blah looking. This can be a catastrophe, too, if you are super anal retentive about image quality. Or maybe a red bird on-screen looks pink off the printer... total catastrophe if you're hanging the picture in a gallery. When you take a treatment plan and "print" it on the linac, a disaster would be if you got wholly different fluences than the ones you see on screen. But, very rarely, you can make a plan which is outside the linac's "gamut." And it will fail the gamma. To a very anal retentive physicist, this is a catastrophe.
But failing gamma is arbitrary (and ironically subjective). One can choose more permissive, or
more stringent, values in the gamma analysis to make gamma pass versus fail. All of the choices physics makes in gamma analyses have been arrived at WHOLLY detached from any clinical outcomes. From my viewpoint, a gamma fail plan might lead to
increased local control versus decreased local control in a patient. We just don't know. We have never tested gamma in ANY clinically rigorous fashion. EVER.
When you print a picture of a zebra, you never get an ostrich (unless there's been severe user error). When you run gamma analysis of an IMRT plan for a zebra, the analysis never shows ostrich. It occasionally shows the zebra's stripes are not the right color or a little mis-aligned. If there's a clinical reason to do IMRT QA on every IMRT plan, there's a clinical reason to do QA on every 3DCRT plan with electronic wedges (who uses real wedges anymore?). But of course no one is doing QA on every 3DCRT plan. The large academic centers that stopped doing IMRT QA on every IMRT plan did so for the same reason that it would get boring, and be inefficient, to check for ostriches on zebra printouts.
