This just so happens to be my area of academic interest and I have some experience with both of these machines. I do my best to stay neutral when it comes to one machine versus the other. I don't know if I'll have many on this forum agree with many of my views on MR-RT. That's okay with me.
The core issue is that radiation oncology depends on devices instead of pharmaceuticals. Unlike pharmaceuticals, these devices get approved through the FDA substantial equivalence pathway, which means that the ViewRay or Elekta products only have to be "substantially equivalent" (i.e. deliver radiation therapy at least as well as its existing predicate device) to be approved. These devices are built by these companies and then sold without a clear and known benefit. It's up to us to decide what it might be good for and hopefully prove it. If the devices had to undergo full testing for safety and efficacy for specific indications, like new medications, these new machines would likely not exist, be delayed many years, and be far more expensive to recoup the costs of pre-clinical studies. It would even be hard to make the newer treatment machines at all--like a Truebeam or a Versa HD, as they would all have to demonstrate their efficacy and safety in large patient cohorts. You can argue that this should not be the case--that these new radiation devices should have a proven indication for use before release. While it's a separate argument, I think it's a good and interesting argument to have.
Whenever something is novel like this, it's easy to be a nay-sayer. The technology is brand new. Cone beam CT was widely panned when it was first released as a product back in 2003. Interestingly, I'm not aware of any randomized data showing that CBCT improves outcomes. There are some retrospective series that suggest it. Regardless, I think we all use it for selected indications. Some trial protocols mandate its use. The evolution in image guidance allowed for newer therapies to occur. It's hard to know whether CBCT ever improved outcomes directly, but it's a technology that allows improved image guidance and that has allowed for improved outcomes. Certainly CBCT today is far better than CBCT was in early 2000. Varian, who actually has an MRI product manager at conferences (for what product, exactly?), claims that MR-RT is a stupid idea because they're going to make CBCT look even better than it does today. That's fine by me--4D CBCT is cool technology, though it really only works in lungs so far to my knowledge. Maybe there are other tricks up their sleeve.
In my mind, CBCT is a mature technology with wide acceptance, but it has taken over 10 years. I think MR-RT will be the same. I put the MR-RT start date at 2014 since ViewRay was cleared in 2012, but it wasn't treating then to my knowledge. It's my belief that in another 10 years, we'll all accept it for selected indications. It may not make its way into every private practice the way CBCT has since it's easy to just bolt on a kV source and an EPID on an existing linac. But, I suspect it'll be fairly widespread within academics. It's the challenge for people like myself to find good things to do with it. I have a lot of such ideas. The easiest most compelling ideas are moving structure targeting and easy online adaptive radiotherapy. The adaptive portion is just now beginning to happen with the latest ViewRay software, but was previously quite labor intensive to make it to prime time. It still is--you've got VU in Amsterdam basically taking an hour per patient and adapting everyone's therapy every day--but this will improve with better software and workflows.
Currently there is no billing code for MR guidance, which is unfortunate given its cost. Fortunately, it is much cheaper than protons--the machines retail at somewhere around $5-$10 million + vault costs--which isn't crazy compared to something like a Truebeam Edge which is somewhere around $3 million. I'm not clear that single gantry proton solutions can be had for less than $20 million, though there's certainly a lot of downward pressure on costs and that's a very moving target with a lot of startup companies promising questionable cheaper products. A lot of people are shocked to hear that IBA and ProBeam proton systems are upwards of $120 million. To me, MR-linacs and protons aren't even in the same league when it comes to cost. The only issue is that protons have their own reimbursement codes and MR-linacs don't. This does make the financial justification/business model for MR-linac difficult in the current environment. When CBCT had its own reimbursement codes, that really drove proliferation especially in private practice. This may make MR-RT an academic device or one for hospital systems to brag about/try to draw patients with for the near future.
I agree with mcrue. Cobalt is not a great way to deliver RT. The MLCs are 1 cm width on the ViewRay device. MR-guidance is allowing me to do some cool stuff with image guidance, but I'm not sure it's letting me do anything I couldn't do with fiducials and a respiratory tracking system. When I want to be tight with MR guidance, I want to be very tight, and I need a plan that will also be very tight. Still, I emphasize that this is just the beginning. ViewRay is planning to release a linac upgrade to their systems with narrower MLC widths and make themselves a true MR-linac solution. Will the Elekta system still make them obsolete? It depends. The Elekta system has a real 1.5T Philips MR system under the hood, which might allow for better imaging. However, the ViewRay system has been treating patients for well over two years now and is the more mature technology from a software standpoint. Who knows what the actual real life performance of the two systems will look like.
As it stands, both machines have serious vault considerations. The cobalt system needs shielding as a cobalt room--for us that was a significant cost to upgrade from a linac vault. The Elekta system is big and has its own vault considerations. I can't be more specific because I don't know how much of what I know is still considered proprietary.
As for skin toxicity, secondary electron effects on dose distributions due to magnetic fields are real on both devices. There is plenty of data to demonstrate that this exists at both 0.35T and 1.5T field strengths. While it's easy to demonstrate the potential overdosing of skin with one or two beams, the reality is that with many beam angles or arcs and IMRT based planning, the effect is largely mitigated. If anyone wants to review that further, I can pull papers, as these effects have been widely modeled by numerous MR-RT groups.