Will this fix my crushed Lumbar Disc in my lower back?
No but it might make it well done, I wonder what the ideal cuisson is on a lumbar disc.
That being said, if we consider this a valid medical device for alleviating things like back aches there are some guidelines. If we keep it simple and look to the giant microwave oven that a MRI scanner is: the MRI guideline is somewhere around 4-5W per kg of tissue, with up to 20 W/kg for some specific cases (e.g. you can do this to someone's foot), if I remember well from my medical imaging days. This is only the average power over something like 5-10 minutes - depending on local regulations. You should be able to find tables of the exposure limits fairly quickly for your region. (e.g. a quick google:
https://www.medtronic.com/content/dam/medtronic-com/mri-surescan/documents/SAR_White_Paper.pdf ) So, yes, you can go way above that instantaneously. The amplifiers on the average MRI scanner can easily reach a few tens of kW of instantaneous output power (very shortly), because you got to be generating fields in the order of mT/m with those gradient coils and the RF transmitter itself is also a piece of work. Really depends on the scan sequence, and they tend to over design some of these things massively in my humble opinion. But surprisingly little of that power goes into the patient. I remember the power dissipation estimation of the scanner I worked with often overestimated it by 50% or more, it easily burned up 30-40% of the power in the coils themselves. And given how many times I got microwaved by one of those, I'm surprisingly ok with this microwave heating concept in a way. So yeah that means that at 250W you could stick a 50 kg human underneath for 5-10 minutes if you ignore the whole list of boundary conditions, and I'd be amazed if this wasn't pulsed in some way which would make it fairly safe in the grand scheme of things.
And before you ask, the actual permissible power level for MRI also tends to depend on the operator qualifications, medical diagnosis, and the patient itself. Transmitter average output power during scan sequences is in the 10 - 100W range usually, partially being the reason why some scan sequences take so long (you don't want to go over SAR limits). The real question is what's considered "acceptable", e.g. if you want to get a good look at something that's about to kill someone, you're not exactly going to have much moral qualms going a bit over those limits if it gives a better image. Also sometimes by dumping in a lot more power you can get the image a lot quicker as well. But you don't want to boil the innards of someone with a bad case of tendinitis. And I haven't touched one of these things in 5 years so my apologies if any part of it is wrong.