X-ray CT is commonly used for proton therapy planning, but uncertainties caused by both the conversion of Hounsfield units to proton relative stopping power and the beam hardening effect of the CT scanner necessitate the addition of a 2.5–3.5% margin to both edges of the spread-out Bragg peak. This margin detracts from one of proton therapy's key advantages – the sparing of healthy tissue by ensuring that the proton beam's Bragg peak is contained within the target volume.

An alternative approach for planning lies in directly measuring the anatomic distribution of the proton relative stopping power, using proton imaging techniques such as 2D proton radiography or 3D proton CT. While medical proton delivery systems have reached an advanced level of development, corresponding imaging technologies have been relatively under-investigated – and there is a need to develop practical and cost-effective proton imaging detectors.

Systems currently under development can be categorized as either tracking or integrating type detectors. Tracking imagers use two detector arrays and a calorimeter to measure individual proton trajectories and residual energy – but are incompatible with passive scattered proton beams and have a slow image acquisition speed that makes them impractical for clinical use.

In contrast, integrating systems are compatible with both pencil-beam scanning and passive scattered proton beams – making the image acquisition time short enough for clinical use. While a number of such detectors have been proposed, all have had significant limitations – for example, plastic nuclear track detectors require an unsuitable wet chemistry developing process, while CCD camera and scintillator detectors come with a relatively low energy resolution, as do stacked CMOS sensors.

In their new study, Peng Wang – of the Texas Center for Proton Therapy – and colleagues report a prototype detector design with a one-dimensional silicon diode detector array that was originally developed for measuring planar photon dose distributions. The detector is composed of four monolithic silicon sensor modules – each of which is 64 mm in length and comprises 64 pixels. When placed along the central axis of the proton beam, the position of the beam's Bragg peak along the diode array correlates with the water equivalent thickness (WET) of the object being imaged.

"The design has the potential to have a very high spatial resolution – in the x–y direction – and energy resolution – in the z direction – since the size of diodes can be made quite small," said Wang. The researchers tested the detector with a block phantom made of polymethyl methacrylate slabs, of known WET, which were designed to calibrate the prototype and assess its spatial and contrast resolution.

By moving the phantom relative to the proton beam axis, the researchers produced a proton radiography image of the phantom; the resolution and maximum error in WET measurements were found to be 2.0 and 1.5 mm, respectively. They also reconstructed a proton CT slice using filtered backprojection, with the resultant image showing good agreement with simulation – clearly resolving slits in the phantom down to the 2 mm scale.

"There is increasing interest in proton radiography and proton CT as 2D tools for range and dosimetry verification in adaptive proton therapy," comments Reinhard Schulte, a medical physicist from Loma Linda University who was not involved in this study. He adds that the new detector design is a low-cost solution that "could find its way into clinical application much sooner than the more technologically demanding solutions based on individual proton tracking and analysis".

Having demonstrated the feasibility of their approach, Wang and colleagues are now looking to build and calibrate a complete prototype detector as the next step towards commercialization and clinical application.

Related stories

• Proton CT could ease treatment planning
• Proton radiographs improve range prediction
• Proton CT improves stopping power accuracy
• The potential of proton CT