Current proton imaging systems

A small number of collaborations have been formed since the early 2000s to develop proton CT systems and work on the associated challenges such as fast data acqui- sition and non-linear image reconstruction. The most significant of these groups include the collaboration between Loma Linda University and the University of California Santa Cruz (LLU/UCSC) with California State University, Baylor Uni- versity, Northern Illinois University and the University of Wollongong each making contributions and the Italian PRIMA Collaboration.

The first proton imaging system in the modern era, used to reduce stop- ping power uncertainties was a proton radiography system capable of taking planar images at the Paul Scherrer Institue (PSI) in Switzerland. The system contains 2 tracking detectors either side of the image subject comprising scintillating fibres and a plastic scintillator range telescope. The active area of the imaging device was 22.0 x 3.2 cm2. The system was used to optimise the HU to RSP conversion curve by comparing the calculated WET through an animal patient with measured WET from the radiography system for each proton path. By optimising the conversion curve, the mean WET deviation was reduced from 3.6 mm to 0.4 mm, suggesting that the modality could reduce a significant systematic error in the delivered proton range if used at the treatment planning stage [39, 40].

The LLU/UCSC collaboration began work in 2005 with an initial study into the electron density, and consequently stopping power, resolution of a potential pro- ton CT system [33]. These studies showed that an accuracy of greater than 1% should be possible for proton CT. The collaboration went on to build two proton CT prototypes, namely the Phase II and Phase II instruments [29]. These proton CT systems were constructed based on silicon strip detector (SSD) technology for

individual proton tracking with scintillators providing the measurement of the resid- ual energy of the beam. The tracking system of the Phase I system used two planes of SSDs in an x-y orientation. Each SSD was 400µm thick with a pitch of 228µm. The detectors were tiled to increase the field of view, however the x and y planes were offset from each other to reduce the effect of the “dead zone” around the edge of the sensor [41]. The energy measurement was performed by a calorimeter con- taining CsI:Tl crystals readout by photodiodes. The system was slow, with a scan time of several hours however did implement reconstructions taking advantage of the proton most-likely paths. Artefacts arose in the produced images due to the gaps between the SSDs, and the authors saw room for development in the calorimeter in the Phase II system [42].

The Phase II system operates between 10-100 times faster than the Phase I system and would be capable of imaging a human lead in less than 10 minutes [42]. The field of view of the system extends to approximately 8.8 cm x 35.0 cm where an object can be scanned through the system. The SSDs were improved by sawing the edges to minimise the “dead space” surrounding the detector such that they could be tiled more effectively. The calorimeter was redesigned to become a novel segmented energy-range detector, where 5 individual stages of a polystyrene-based scintillator, each measuring 5.1 cm in depth, were individually read out by photomultiplier tubes. Segmenting the scintillator reduced the dynamic range required on the readout electronics to determine the residual energy of each proton with the polystyrene material operating significantly faster than the CsI:Tl crystals used in the Phase I scanner. The system was used to image a number of test phantoms as well as a paediatric head phantom [43, 44], with reported results claiming to achieve WEPL resolution close to the theoretical limit of 2.8 mm and RSP accuracy of better than 3% for most materials with many better than 1%.

The PRIMA collaboration demonstrated a prototype system with a field of view of 5.0 x 5.0 cm2, comprising SSDs to track protons and a YAG:Ce scintillator- based calorimeter to perform measurements of the residual proton energy. The YAG:Ce scintillator features a significantly shorter decay time than the CSI:Tl scin- tillator of the LLU/UCSC Phase I scanner and the SSD thickness was much lower than the LLU/UCSC design (200µm), potentially reducing scattering caused by the detectors therefore improving the spatial resolution of the system [45]. Images of a 2 cm diameter plastic phantom have been published with an electron density resolution of 2.4% and a proposed PRIMA II scanner will increase the field-of-view of the system and feature a redesigned DAQ with an event rate exceeding 1 MHz.

animal in 2013 [46]. The system used gas electron multiplier chambers as position sensitive detectors, with a stack of 48 polyvinyl-toluene plastic scintillators each read out by silicon photomultipliers providing the measurement of the residual range. The system only tracks protons as they exit the image subject, therefore the system can only assume linear proton paths through the image subject however this is sufficient for radiography. The active area of the detector is quite large at 30 x 30 cm2. The rate capability of the system is also competitive at 1 MHz.

Other notable efforts include the collaboration between Northern Illinois Uni- versity and Fermilab (NIU/FNAL), who constructed a system comprising scintillat- ing fibres as trackers, and a scintillating range counter to determine the residual proton range [47]. Despite intending to use the system with a 200 MeV clinical beam, issues with the instrumentation and funding expiration unfortunately meant that the system wasn’t tested [29].

Most recently, a group comprising the University of Bergen, Bergen Univer- sity College, Haukeland University Hospital and Utrecht University are planning to develop a new proton CT system using CMOS monolithic active pixel sensors (MAPS) [48]. The group have published results testing a calorimeter that is based on the use of MAPS detectors, providing the opportunity to track multiple protons in each frame whilst measuring their residual range. A new prototype system is planned.