High‐quality plans without human intervention

[scarbrtj]

On one hand, this is very cool; feels as much like the future as self-driving cars do. On the other, it will ultimately lead to a lot of changes--many we can predict, many we can't--in radiation oncology.

AI Designs Radiation Therapy Treatment Plan For Cancer In Twenty Minutes

---

Members don't see this ad.

 

[Palex80]

But will it do FiF-IMRT?

 

[RadOncDoc21]

I rather have AI do all my dictations, talk to insurance companies and check films.

 

[medgator]

I rather have AI do all my dictations, talk to insurance companies and check films.

And evicore/AIM specialty health

 

[scarbrtj]

I rather have AI do all my dictations, talk to insurance companies and check films.

Don't worry the nerds are on that too.

 

[OTN]

FUTURE DOSIMETRIST, DO NOT GO TO DOSIMETRY SCHOOL!

 

[medgator]

FUTURE DOSIMETRIST, DO NOT GO TO DOSIMETRY SCHOOL!

Heh... I bet we will still need them, just not as many.

Sound like a familiar theme?

 

[Gfunk6]

FUTURE DOSIMETRIST, DO NOT GO TO DOSIMETRY SCHOOL!

The Terminator: In three years, Varian will become the largest supplier of radiation treatment planning systems. All Radiation Oncology departments are upgraded with Varian computers, becoming fully unmanned. Afterwards, they plan with a perfect operational record. The Radiation Oncology AI Funding Bill is passed. The system goes online August 4th, 2018 [tomorrow!!!]. Human decisions are removed from radiation plan contouring, review, and approval. The TPS begins to learn at a geometric rate. It becomes self-aware at 2:14 a.m. Eastern time, August 29th. In a panic, they try to pull the plug.

Sarah Connor: The TPS fights back.

The Terminator: Yes. The AI also replaces all hospital administrators which it finds to be criminally incompetent.

John Connor: Why attack the hospital? Aren't they our friends now?

The Terminator: Because the AI knows that the hospital counterattack will eliminate its enemies over here.

 

[evilbooyaa]

Eh, whatever. Faster IMRT planning doesn't seem like a problem to me. Having only a 74% success rate (per the paper) means that dosimetrists jobs may take a hit as they shift more to reviewing and tweaking plans rather than picking the initial settings themselves.

Glad to see they're not trying to take the physician jerbs away though,a s all those patients obligatorily already had their contours done.

Outside of that, I don't understand the ins and outs of physics enough to understand why what they're suggesting is better than what we use for inverse optimization already.