Theriac revisited

[xela]

So Cedars-Sinai altered one of the programs on their CT scanner and introduced a bug resulting in over 200 patients receiving radiation overdoses, reminding many of us (geeks with a healthy interest in how technology fails, that is), at least superficially, of Theriac-25.

Now, I only know about CT scanners from the laying-there-with-the-godawful-taste-of-barium-solution-in-my-mouth perspective, but how fucking hard can it be, say once a year (and immediately after reprograming or installing patches), to run a CT scanner through each of its programs with a dosimeter inside it instead of a patient? Why isn't that standard procedure?

Comments

[sammason.livejournal.com]

That does sound like a fairly serious case of medical malpractice.

[yakshaver.livejournal.com]

Yeah. Sadly, malpractice doesn't restore anyone's health.

[nakor.livejournal.com]

It is. But (a) these people "knew what they were doing," and so didn't test their work, and (b) most systems have a special test mode. In a case like this, the system might do the right thing in test mode and the wrong thing to live humans. The test mode really tests that the radiation sources do what they're told, not that you're telling them safe things.

[yakshaver.livejournal.com]

If you really know what you're doing, you know your you're* fallible, and need to test your work. And that's the standard I expect when human lives aren't involved.

A test mode that doesn't test realistically is worse than useless.

But I reckon I'm preaching to the choir here.

---

* No, I didn't typo, submit, read the live version, and edit it just to prove my own point about testing. But yeah.