Is it really true? Single Gantry impt units are still going for ~$20 million are they not? Sure that's better than the $100-150 million of the original 3 story non impt units, but hundreds of percent higher than a $3 million sbrt/imrt capable linac.
Sorry but the imrt vs 3d argument being used as an analogy to justify examining protons and carbon ions is way off base and smacks of a red herring argument.
Considering how much the US consumer and patient bears the cost from big pharma for drug development compared to rest of the developed world, I'm fine with Japan and Europe paying up and figuring out carbon ions for us first
Regarding costs coming down, there is an interesting comparison at hand in Spain, where Madrid is getting 2 proton centers in the next 2 years.
Two private hospitals are due to open centers in Madrid, where children and patients with rare tumors will especially benefit from a treatment with fewer side effects than X-rays
elpais.com
The first one, due in 2019, with an IBA compact cyclotron and half-gantry, costs roughly $40 Million Euro, machine and building included.
The second center, due only 1 year later, has a Hitachi synchrotron and full gantry costing $40 million Euro, machine and building included,
with an option to expand.
A full gantry is a big added value to clinical through-put compared to a half-gantry, probably saves 5 minutes on every complex treatment, but the option to expand is an even bigger deal, because as program growth occurs, the cost of the accelerator is further amortized. It only costs the gantry and vault, roughly $5-10 million, to double capacity, and another $5-10 million to triple capacity.
Each added proton room is at nearly the cost of building a new MRI linac and vault, is it not? Non-gantry rooms for a chair or laterals only, are even smaller and cheaper. Chicago's proton center has such a chair room, as do many carbon facilities for lateral-only beam rooms.
Someone mentioned range uncertainty of particles is an issue. True, but there is a difference between an unknown and the unknowable.
There are some technology advances that are already improving accuracy of delivery. One is daily 3D imaging at isocenter, which is near-ubiquitous with IMRT and common with proton v2.0 (pencil beam scanning era), but many proton centers built in the proton v1.0 era (passive scatter, 2D imaging) are now being retro-fitted with CT and PBS, or augmented by new proton centers (eg MDA).
One way to improve stopping power calculation is correction of the Hounsfield Lookup Table by using Dual Energy CT calibration curves, which Wolhlfahrt, Mohler, Troost, Geilich and Richter in Dresden and Heidelberg have shown can improve stopping power calculations by about 1.2% from the usual +/- 3.5%.
Doing patient-specific stopping power calculations using Dual Energy CT can further reduce stopping power uncertainty by another 1% or so, to about +/- 1.5% for mixed tissues, using a technology that is in everyone's dept (a CT scanner run at 80 KVp, then again at 150 KVp = dual energy CT).
Here is their latest manuscript, not yet published in the Red Journal:
For those looking to the future, proton v3.0 will feature further advances like proton CT, in which we can capture and image an exiting proton beam for direct verification of beam path trajectory and stopping power. Physicists like Mark Pankuch, PhD in Chicago are already looking at this for brain, H&N and lung using existing hardware. Arc therapy will probably be part of proton v3.0 as well, which may help improve robustness by splitting one or two large directional uncertainties into multiple small opposing uncertainties. LET-based or RBE-based planning will also help move hotspots into GTVs, and away from structures like optic nerves or brainstem, further improving plan quality and safety. Just my 2 cents.