2014: During year 2014 more than 140 treatment rooms were serving 14,500 patients

2015: In year 2015 only 0.5% of radiation-needy patients were treated with proton therapy.

2019: By year 2019 330 patient treatment rooms are expected to be available, but even then only 1% of radiation-needy patients will be able to receive particle therapy.

2030: By year 2030 it is expected that 1200 to 1800 treatment rooms will be open to patients worldwide. But even 1800 rooms will allow only 5% of radiation-needy patients to receive particle therapy.

2000: In year 2000 the proton therapy market was valued at US$ 1 million

2015: During year 2015 the market reached US$ 800 million (1, 2)

2019: In year 2019 the market is expected to reach more than US$ 1 billion

During 2000-2014 the average annual growth rate was almost 15% p.a.

During 2010-2014 the growth rate underwent acceleration to 22% p.a.

2030: By year 2030 the proton therapy world market is expected to be worth US$ 3.5 to 6.6 billion in response to an annual growth rate of 11% to 16%.

In 2015 there was reported to be 15 manufacturers or developers of ion therapy equipment, and it was predicted that by 2019 the market would be dominated by 5 of those companies. Important manufacturers include: Ion Beam Applications (IBA), Varian, Sumitomo, Hitachi, Toshiba, Siemens, and Mitsubishi, the latter having installed at least 11 proton- or carbon-ion centers in Japan. Of the more than 60 proton therapy facilities worldwide 25 are located in the US. After a slow start the proton therapy market began to accelerate in response to: (i) a growing recognition of the effectiveness of particle therapy, (ii) the recent introduction of lower cost compact systems and single-treatment-room centers. CERN, with its experience in proton accelerator technology (e.g. the LHC) continues to make important contributions to ion therapy. Proton/carbon-ion centers in Japan, Europe-and-Elsewhere, and the U.S. are listed in Tables 1 – 3 , respectively.

An increasing interest is also being shown in carbon/heavy-ion therapy, institutions offering which include: The National Institute for Radiological Sciences (Chiba), Hyogo Ion Beam Medical Center (Hyogo), Gunma University Heavy Ion Medical Center (Gunma), Saga Heavy Ion Medical Accelerator in Tosu (Saga), and Shanghai Proton and Heavy Ion Center (SPHIC, Siemens). In 2019 the program “Next Ion Medical Machine Study” (NIMMS) was established to support R&D based on CERN accelerator technology relevant to heavy ion therapy. It is also interesting to note that in a partnership between IBA and Toshiba the latter will become the Japanese distributor of IBA’s Proteus ONE compact single-room proton therapy facility, and IBA will become Toshiba’s carbon-therapy agent outside Japan. Carbon-ion centers worldwide are listed in Table 4 .

Review articles on proton therapy (6, 7) begin by citing the work of Robert Wilson of the Lawrence Berkeley Laboratory who in a paper published in 1946 (8) was the first to point out the importance of the proton Bragg peak for targeted radiation therapy. As protons pass through tissue at velocity v they continuously lose kinetic energy by inelastic Coulomb scattering against atomic electrons. The rate of energy loss, being proportional to 1/v² , rises sharply as the protons slow down, the end of their range, and form the Bragg peak. Protons also undergo repulsive non-elastic interactions with atomic nuclei and will be deflected from their original path. The product of such interactions may be secondary protons, heavier ions, neutrons, and gamma rays. These non-elastic proton-nucleus interactions, although less frequent than proton-electron ones have a stronger effect (9).

For a detailed description of the history of proton therapy systems reference (7) is recommended. That same article also gives the locations and provides technical details (as of 2010) of 6 commercial proton beam therapy systems (PTS), viz: The IBA Proteus^® 235 PTS, Sumitomo PTS, Varian PTS, Still River Systems Monarch 250 (Mevion Medical Systems) PTS, Hitachi PROBEAT PTS, and Mitsubishi PTS. The Hitachi and Mitsubishi systems are based on synchrotron acceleration and the others on cyclotrons. Mitsubishi is particularly active in Japan; a list of their installed proton- and carbon-ion systems is given in Table 5 . A complete list of Japanese installations is given in Table 2 and lists of European and US particle therapy installations are given in Tables 1 , 3 .

Excellent companions to the present document are two recently published reviews. In a paper entitled “Superconducting Magnets for Medical Accelerators” (11) S. Prestemon offers an introduction to hadron therapy and considers the challenges presented by superconducting technology. Cyclotrons and synchrotrons are reviewed, also gantries in general including TULIP (TUrning Linac for Proton Therapy) and ULICE (for Union of Light Ion Centres in Europe) which involves 20 European installations coordinated by CNAO (Pavia, Italy, see Section 4.1). L. Rossi assembled a document entitled “HITRI+ and I-FAST: Next Eu programs for SC heavy ion therapy machine” (12) which outlined the status of ion therapy in Europe and Asia. Considered were HIT (the Heidelberg Ion Beam Therapy Centre, see Section 5.5), and HIMAC (the Heavy Ion Medical Accelerator in Chiba, see Section 5.7.1). Also described was the carbon ion superconducting gantry collaboration involving CNAO (Pavia, Italy, see Section 4.1), MedAustron (Weiner, Austria, see Section 4.1), CERN, and INFN.

In convention radiation (photon) therapy the absorption of x-rays or gamma rays is intensive at the surface and decreases with distance into the subject. Thus in traditional therapy a high intensity of photons must be administered using the isocentric convergence technique to allow multiple beams with the diminished intensity focus at the tumor site. As a result, much of the photon’s energy causes damage to healthy tissue.

In contrast to photons, charged particles are “silver bullets” whose interactions with matter are characterized by the Bragg curve. The energy loss by charged particle through matter is described by this curve which rises to a maximum (the Bragg peak) just before the end of the particle’s track. The peak occurs because the cross section for particle-matter interaction increases just before the particle comes to rest. In particle therapy the beam energy is adjusted either electrically (synchrotron accelerators) or by filters (cyclotron accelerators) to ensure that the Bragg peak occurs at the tumor site. Figures 1 , 2 illustrate relative dose versus depth from the body surface for photon- and charged-particle radiation. This indicates that while photon therapy is characterized by relatively high entrance and exit doses, proton therapy has not only a lower entrance dose but a negligible exit dose. It also shows that the position of the Bragg peak can be adjusted to ensure that the tumor receives the intended radiation. This allows patients to receive high radiation doses with low risk of collateral tissue damage.

The radiobiological effectiveness (RBE) of photon (traditional, x-ray) therapy is arguably equivalent to that of proton therapy, however as illustrated in Figures 1 and 2 , a relatively large proportion of the photon’s energy is deposited in the entrance and exit healthy tissue regions. A distinct advantage of proton beams is that by modulating the energy to create a spread-out Bragg peak (SOBP) a large fraction of the beam energy can be deposited in the tumor site. This advantage is shared by heavy (in particular carbon) ions but in addition, being heavier than protons, they provide a higher RBE, that also increases with depth and rises to a maximum at the end of their range in the tumor region (15, 16).

Photon and proton radiation tend to produce only single-strand DNA breakage. Since cells can repair such breaks damage to both strands is required for successful treatment. Carbon ions are able to produce double-strand breaks that cannot be repaired. In other words the RBE of carbon ions against DNA is up to three times greater than that of x-rays while protons are only slightly more effective. Whereas a full treatment with protons may require 30 sessions just four days may be needed for carbon.

For hadron therapy in general, ions of He, Li, B, N, O, Ne (7), Ar, and Si (17) have also been considered, Section 4.2.

Beams of ionized particles are accelerated and bent by the magnetic fields of cyclotrons, synchrotrons, and dipoles. If the particle velocity, v, is greater than one-third c, the velocity of light, a relativistic correction γ = 1/√[1-(v/c)²] is applied to some of the equations. Thus the velocity (m/s) of an E MeV particle of rest mass m₀ can be deduced from E = m₀c ²(γ-1) in SI units (1 MeV = 1.602x10^-13 J). The bend radius ρ of a beam of particles of momentum p = m₀γv and charge q deflected by a magnetic field B is given in SI units by:

The quantity p/q = (m₀γv/q) which has the units Tm is known as the “magnetic rigidity”, e.g (5). It represents the reluctance of the ion beam to being deflected by the field B. The energy-dependent rigidities of some proton and carbon beams are listed in Table 6 , see also Appendix A . The table demonstrates that heavy ions beams are more difficult to deflect than proton beams and hence require stronger magnetic fields

Hadrons are accelerated in circular paths by cyclotrons or synchrotrons. In a modification of Equation (1) the path radius is proportional to (M/Q)(v/B) in which M represents ionic mass (number of protons and neutrons in the nucleus) and Q represents the ionic charge (number of stripped electrons). Thus as suggested in (7) a hadron synchrotron can accelerate a long list of 0.5-(Q/M) ions, in particular:

Protons are the commonly used ions, accelerated to 220-230 MeV/u they can travel 30-33 cm into the body. Fewer systems use carbon ions which have about the same range at 400 MeV/u.

For injection into a synchrotron that accelerates them to final beam energy (21) particles are typically pre-accelerated by a radiofrequency quadrupole (RFQ) cavity accelerator (7, 22) followed by a drift-tube linear accelerator (DTL) (7, 23). The numerous processes that exist for injection into cyclotrons have been discussed in detail by Mandrillon (24). Some commercial accelerators are listed in Table 7 .

CERN, with its experience in proton accelerator technology (e.g. the LHC) continues to make important contributions to ion therapy. Some 20 years ago CERN initiated the program “Proton Ion Medical Machine Study” (PIMMS) whose purpose was to produce a synchrotron tailored to treat tumors with protons and carbon ions. The design evolved into a proton and carbon ion machine built for CNAO (Italy’s National Center for Oncological Hadron Therapy, Pavia Italy, Tables 1 and 4 ). Subsequently MedAustron (Weiner, Austria, Tables 1 and 4 ) with technical support from CERN based its clinic on the CNAO design. In 2019 the program “Next Ion Medical Machine Study” (NIMMS, led by M. Vretenar) was established to support R&D based on CERN accelerator technology relevant to heavy ion therapy.

Therapy synchrotrons yield beams of relatively low intensity and of variable energy. Protons are injected into a ring of dipoles at 2-7 MeV and accelerated up to 70-250 MeV as the dipole field is ramped. Synchrotrons can be up to 20 m in diameter. The first hospital-installed proton therapy system was based on a Fermilab synchrotron to be further developed by Optivus Technology. Other vendors such as Hitachi, Mitsubishi, Toshiba, and Siemens followed suit, Table 7 . But as Jongen has pointed out (7) about 75% of today’s proton therapy systems are based on cyclotron technology

The cyclotron consist of an RF system situated between the poles of a normal-wound or superconducting electromagnet. Considered below are the classical “traditional” cyclotron, the isochronous cyclotron, and the synchrocyclotron.

A Sumitomo-ProNova Solutions collaboration produced a normal-conducting 230 MeV cyclotron 4.4 m in diameter weighing 220 tons. The smaller ProNova SC360 system used a superconducting 230 MeV isochronous cyclotron 2.8 m in diameter weighing 50 tons. The Massachusetts Institute of Technology (MIT) in collaboration with ProNova has been working on the design and construction of an iron-free variable energy (70-230 MeV) synchrocyclotron also 2.8 m in diameter but weighing only 5 tons (28). Several advantages accrue from the use of this accelerator: (i) its variable-energy capability removes the need for graphite energy degradation, (ii) its low weight would make it very attractive for gantry mounting, (iii) gantry mounting eliminates the need for beam-directing magnets. Taken together these advantages lead to an attractive proton beam therapy system.

Before it enters the treatment area or gantry the ion beam requires energy adjustment. Therapy cyclotrons generate a fixed-energy beam of typically 230 MeV, the needed energy variation between 60 and 230 MeV being achieved by passing the beam through an “energy degrader, an absorber of variable thickness such as two opposite-facing wedges of graphite. These are followed by a magnetic analyzer consisting of a combination of dipole magnets and collimators (20). The “Energy Selection System” of Ion Beam Applications S.A. (IBA) allows tuning from 60-230 MeV in less than 1 second. Although the beam loss by degradation can be as much as a factor of one hundred or more typical cyclotrons deliver sufficient beam intensity to make up for this. With regard to synchrotron sources, rapid energy variation can be achieved by extracting the ion beam at different times in the acceleration cycle (20). The synchrotron can deliver proton beams in more than 90 energy steps between 73 MeV and 222 MeV corresponding to penetration depths in water of 4 to 31 cm (29).

In the present context a gantry is a massive structure that rigidly holds in place the guidance magnets of ion beam therapy. A typical magnet arrangement is shown in Figure 3 . The figure also indicates the so-called “isocenter” or axis about which the gantry rotates thereby enabling the ion beam to enter the tumor in all directions. The term “gantry” may refer to the mechanical structure, the magnet string (5), or the entire system as illustrated in Figure 4 .

Figure 3 illustrates a Pavlovic-type (30) arrangement of magnets and devices in an ion therapy gantry. The beam is bent by three dipole magnets BM 1, BM 2, and BM 3, and kept in focus by six quadrupole magnets Q1 to Q6. The beam enters the gantry through a monitor PRM that monitors and records the center of the beam and its size. Three other PRMs are associated with steering magnets for beam trajectory correction and two more, PRM1 and PRM2, are located in the nozzle. Several nozzle arrangements have been used to passively or actively spread the mm-size beam over a treatment area that may be as large as 30 cm² (5, 34). Described in detail by (31) are: (i) uniform scanning nozzles, (ii) pencil scanning nozzles, (iii) single scattering nozzles, and (iv) double scattering nozzles. The latter version is represented in Figure 3 .

During a typical treatment session a proton beam energy may need to be varied from 70 to 250 MeV. This produces a change in rigidity of from 1.231 to 2.431 Tm requiring the field of a 1-m-radius dipole to track the beam energy and hence increase from 1.23 to 2.43 T. In the absence of field-change a typical normal dipole will accept a beam momentum change, dp/p, of less than 1% ( (35), p.2). Tumors are typically scanned in layers 5 mm deep enabled by a momentum sweep of 1% ( (35), p.4), the layer-scan taking of order 100 ms. A dp/p of 1% corresponds to a beam energy change dE/E = 2dp/p = 2%. Thus to control a 100 MeV beam (proton magnetic rigidity 1.483 Tm) a 1-m-radius dipole would require a field change of 15 mT. On this basis the average sweep rate of the scan is 150 mT/s (20 times faster than LHC). Taking another approach, a dE/E of 2% implies 50 scanning steps covering the energy range from 70 to 250 MeV and a 1-m-radius dipole field range of 1.23 to 2.43 T. At 100 ms/step this leads to an average sweep rate of 240 mT/s. From another standpoint, at 5 mm per layer a proton-beam-depth range of 10-35 cm would require 50 scanning steps, leading to the same result.

Sweep rates of hundreds of mT/s cause stability and energy-loss problems for superconducting magnets. So when considering a gantry upgrade from normal-conducting to superconducting magnets these high ramp rates created a demand for achromatic bending dipoles with wide momentum acceptances, such dp/p = 5-10%, 25%, and 40-50% [(35), p.3]. The use of such magnets, which enable the entire energy range to be covered in just a few steps, eases the ramp-rate requirement.

As outlined in (20) the use of normal-conducting iron-core dipole magnets with their maximum bore field, B, of about 1.8 T governs the size of most commercial gantries. At a typical 250 MeV proton beam rigidity, R = 2.43 Tm the 1.8 T field calls for a bending radius ρ = R/B = 1.35 m. The addition of 3.5 m for distance from the nozzle entrance to the isocenter leads to a gantry rotation radius of about 5 m ( Figure 3 ). Thus proton gantries are about 10-12 m in diameter, 7-10 m long ( Figure 3 also (18)) and weight about 100-200 tons (20). Further details are provided in Appendix B .

Much larger than the above proton gantries is the famous carbon-ion gantry of the Heidelberg Ion Beam Therapy Centre (HIT). The maximum treatment beam energy of 425 MeV/u yields a beam rigidity of 6.58 Tm ( Table 6 ) which calls for a 1.8 T dipole radius of ρ = R/B = 3.65 m. The supporting gantry is ≈ 12 m in diameter, ≈ 21 m long, and the overall system (including 135 tonnes of magnets) weighs ≈ 650 tonnes (36). These statistics signaled the beginning and end of normal-conducting carbon-ion gantries and ushered in the need for superconducting magnets (36). In conclusion we note that HIT should not be confused with HITRI+ which stands for “Heavy Ion Therapy Research Initiative” a design study to assess the relative merits of CT and CCT magnets (Sections 5.7.2 and 5.7.4) for synchrotrons (Section 4.2) and CT/CCT and toroids (Section 5.7.6) for gantries.

The designers of ion-beam therapy systems are continually striving for improved beam optics, smaller size, and lower cost. Regarding optics, pencil-beam scanning was early introduced by the vendors. Introduction of the alternating gradient CCT magnet with its large momentum acceptance (~ 25%) enabled large energy ranges to be transmitted at fixed field, thus reducing the need for low-AC-loss magnet windings. Therapy systems consisting of large accelerators, beam lines and gantries are large and expensive. Superconductors went a long way towards reducing their size. Another important step was taken by MEVION which, by mounting a compact 17 ton synchrocyclotron on the gantry frame, eliminated the need for a string of quadrupoles and bending dipoles. Although the MEVION could be fitted with the desired pencil-beam scanning system, the fixed 250 MeV of the synchrocyclotron called for graphite energy degraders to vary the bean energy. A potential improvement would be to introduce the MIT-ProNova-designed 5 ton ironless variable-energy (70-230 MeV) synchrocyclotron.

Sections 5.5 through 5.7 have outlined the evolution of magnet systems and gantries from proton to carbon-ion, from normal-conducting to superconducting, from standard cosθ and cos(2θ) windings, to tilted double-helix (CCT) windings. We see proton magnet system weights decreasing from an estimated 20~40 tons to less than 5 tons. Gantry layouts have been simplified and magnets have been improved, for example: (i) Bontoiu’s proton gantry (37) consisted of a string of 36 combined-function superconducting magnets, (ii) Toshiba’s carbon-ion gantry consisted of 10 combined-function superconducting magnets arranged in 3 groups for bends of 70° (3 magnets), 70° (3 magnets), and 90° (4 magnets) (18, 33, 46) (47), (iii) Another Toshiba carbon-ion gantry consisted of just 3 superconducting combined function 90° bend magnets (48), (iv) substitution of the standard cosθ and cos(2θ) windings by tilted double-helix (CCT) windings, for which several advantages have been claimed viz: (a) an order of magnitude reduction in weight, (b) possible gantry size reduction, e.g. diameter 5 m and length 8.3 m, and (c) rapid scanning (35).

To perform an efficient treatment accurate image guidance techniques that can distinguish a tumor from healthy tissue are required. Imaging is associated with target accuracy, of targeting, dosimetry, treatment outcome assessment, and prognosis. Images are used in radiation therapy during the following processes: (1) offline treatment planning – tumors and other anatomical structures are identified and the treatment is simulated on a treatment planning computer. During this process CT, MRI and even PET-CT or other type of images are commonly used; (2) imaging is applied to set up the patient in the treatment room and to convert the virtual treatment (a treatment plan) into the real treatment; (3) images are used to assess the treatment outcome: effectiveness, toxicities, and prognosis, etc. While Processes 1 and 3 are usually carried out offline, Process 2 is often performed with the imaging device attached to the treatment machine inside the treatment room to align the patient in the treatment position or during the treatment to monitor the targeting of tumors and organs-at-risk. Process 2 is complex and needs to conform to the gantry and the beam delivery system. The currently available image-guided particle therapy systems are mainly based on 2D orthogonal X-ray imaging, in-room 3D computed tomography (CT) or on-board cone-beam CT (CBCT) imaging. However, 3D or 4D CT and MRI images are commonly used off-line for treatment planning (57). Imaging technologies such as the proton Computed Tomography (or pCT) (58, 59) or MR-guided proton therapy have been proposed but are not yet available for clinical use (60). While X-ray based imaging has difficulty resolving soft-tissue the converse is true for MRI-guided particle therapy. MRI guidance in photon (radiation) therapy MRgRT, which entered clinical practice not long ago, gradually became popular during the past decade (61). Currently there are several commercial available MRgRT manufacturers in the market: ViewRay (62), Elekta (63), and Varian Medical Systems/IMRIS (64).

In order to take advantage of the peak effect (discussed in Section 2.1) the exact depth of the tumor (the target) must figure into the treatment. An error in the target depth of a few mm that may result in only a few percent change in the photon dose may lead to a 100% change in the proton dose (67). This emphasizes the need for direct visualization of the tumor position. Such “image guidance” using MRI has been proposed and modeled by numerous researchers (see section 7.3). The principle of MRI guided proton therapy is illustrated in Figure 6 .

Photon irradiation requires a well-defined beam aimed in the right direction. In proton therapy it is also necessary to predict ensure that the beam will terminate at the tumor side. The additional margin needed for range uncertainty detracts from the benefit of proton therapy. Image guidance is introduced to overcome this uncertainty (67). In a combined MRI-proton system it is necessary to consider mutual electromagnetic interaction between the proton beam and the MRI field. For example Monte Carlo computer simulations and measurements have revealed that a 190 MeV proton beam would be deflected by about 1 cm upon entering a 1 T field. Such effects would influence treatment planning and dose delivery. External magnetic fields that could influence MR image quality come from sources such as: (i) the proton generating cyclotron, (ii) the gantry beam line and steering magnets; the fringe field of such magnets, which can be up to 100 µT, could detrimentally affect the MR image quality. To further investigate such effects Hoffmann and his team (65, 66), in association with OncoRay, attached an open 0.22 T MRI scanner to a fixed horizontally mounted proton research beam line, Figure 7 .

In a useful experimental study Inaniwa et al. (68) investigated the effects of magnetic fields (0.3 and 0.6 T) applied transversely (B_T ) and longitudinally (B_L ) a proton beam adjusted to deposit energies of 1.1 and 3.3 keV/µm into normal and cancer cells. Effectiveness was gauged by the index R₁₀ ≡ D_WO/D_W which is the ratio of the dose that would result in a survival fraction of 10% in the absence (D_WO ) and presence (D_W ) of the magnetic field.

For cancer cells exposed to 1.1 keV (3.3 keV) proton beams R₁₀ increased to 1.10 (1.11) and 1.11 (1.12) in longitudinal magnetic fields (B_L ) of 0.3 T (0.6 T).

For normal cells R₁₀ increased to 1.13 (1.17) and 1.17 (1.30) in these longitudinal magnetic fields.

For both normal and cancer cells R₁₀ showed no significant change in the transversely applied fields

Inaniwa et al. pointed out that the longitudinal field enhancement effect should be taken into account in the design of an MRI-proton system (68).

No commercial scale versions of the research system described in Figure 7 can be made available for experimentation. However numerous modeling studies have been undertaken on the general topic of beam-line/magnetic-field interactions (see Table 10 , Bibliography). Oborn et al. (69) have modeled the path of a proton beam through the 3D field of a 1 T split bore MRI magnet. Used for the modeling was the software GEANT4 (GEometry ANd Tracking), a platform for simulating the passage of charged particles through matter using a Monte Carlo method. Significant rotation of the beam was observed in the longitudinal orientation while a more complex path was seen in the transverse field. It was concluded that pencil beam scanning was favored for either longitudinal or transverse field orientations.

MRI guidance should be introduced in order take advantage of proton over photon therapy. In so doing the influence of external magnetic fields emanating from the imaging and beam guidance systems on image quality needs to be investigated. In so doing it would be useful to produce an engineering design for a full-scale proton-beam/MRI system (e.g. Figure 6 ) using GEANT-Monte-Carlo method modeling. In addition (or alternatively) a small-scale experimental prototype MRI-integrated proton therapy system (e.g. Figure 7 ) could be constructed and used for investigating beam-line/magnetic-field interactions. Given that longitudinal field dose enhancement and other such effects exist, experiments using such a system could be designed to investigate the underlying mechanisms.

To build an effective image-guided or MR-guided proton or particle therapy system, a comprehensive and sophisticated design of accelerator, beam-line, gantry, magnets, and the imaging components needs to be carefully calculated so that all sub-systems could be integrated into a unit that operates efficiently and meets the clinical requirements while reducing cost to a minimum.