Visualising protons during the radiotherapy treatment

Proton radiotherapy already allows the irradiation of tumours with an accuracy of up to a few millimetres, thereby sparing the surrounding healthy tissues. Thanks to a recent innovation in PET-detectors from TU Delft, a radiation oncologist with a predisposition towards image-guidance and a coffee machine at HollandPTC, this radiation treatment can become even more accurate in a few years’ time.

HollandPTC, the proton therapy institute located on the TU Delft campus, houses a mix of radiation oncologists and engineers. It was at the HollandPTC coffee machine that Coen Rasch, professor and head of the radiotherapy department at the LUMC, expressed his wish to be able to verify the quality of the treatment while the patient is being irradiated. ‘As a clinician, I want to know the accuracy of the treatment I prescribe to the patient. For years now, such so-called in-vivo dosimetry has been standard in the more traditional high-energy x-ray therapy. Proton therapy is a high-quality treatment, but it is lagging behind when it comes to treatment verification.’ Also at the coffee machine was Dennis Schaart, associate professor at TU Delft and theme leader Oncotech within the Delft Health Initiative. He described a recent technical innovation in PET-detection (Positron Emission Tomography) developed within his group. Their conversation led to a research proposal# for the implementation of in-vivo dosimetry for proton therapy, which has already been approved for funding by the Dutch Cancer Society (KWF – Koningin Wilhelmina Fonds).

A small amount of radioactivity
Because of small uncertainties in the radiation treatment, it is necessary to irradiate some extra tissue – a safety margin a few millimetres wide around the tumour. Using in-vivo dosimetry, it is possible to determine whether or not this safety margin is sufficiently large, of if it can be decreased (for this patient). The latter is important as treatment related complications are often caused by a too high radiation dose to healthy organs in close proximity to the tumour. Schaart: ‘When treating with x-rays, the radiation beam will pass through the patient, just like when making x-ray images of broken bones. When you put a detector behind the patient, you can measure this exiting beam and back-calculate the radiation dose delivered to the patient. In the case proton therapy, however, all protons will stop in the patient at the location of the tumour. There is no beam to image behind the patient.’ Protons do, however, cause a tiny fraction of atomic nuclei in the patient to become unstable – radioactive – inducing them to emit a positron. When the positron meets its antiparticle, the electron, they will annihilate (cancel each other out). In this process, their combined mass is transformed into two gamma rays flying in opposite directions. Schaart: ‘A state-of-the-art PET-scanner can detect these gamma rays and produce an image of where they originated, providing information as to where the protons stopped inside the patient.

Immediate intervention
Verifying the quality of proton radiation treatments using PET-detection is not a new idea, but it never really gained traction. ‘It required the patient to walk from the treatment room to a diagnostic PET-scanner located elsewhere, a very cumbersome workflow,’ Schaart says. ‘By that time most atoms would already have decayed, leaving very little signal to measure. Some institutes therefore installed their own PET-detectors in the radiation treatment room, but the resulting image quality was too low to be really useful.’ More importantly, out-of-room treatment verification can only take place after delivery of the daily treatment dose in its entirety. ‘It means that any deviations from the intended treatment can only be taken into account during next day’s dose delivery,’ Rasch says. ‘That is too late. From a clinical perspective it is important to allow immediate intervention, when only a fraction of the daily dose has been delivered to the patient. I expect the TU Delft PET-detectors make this possible.’

Large crystals, small detectors
A traditional PET-scanner consists of a ring of small crystals that can convert a gamma ray into a flash of visible light – with each crystal having a single light sensor. The TU Delft innovation consists of using much larger and newer crystals (3 cm by 3 cm) with an array of sensors attached to each of these crystals. Schaart: ‘By applying machine-learning techniques to the signal measured by this array, we can exactly reconstruct at what location, depth and moment the gamma ray interacted with the crystal. Because of this added intelligence, our PET-detector has a much higher resolution and sensitivity when compared to traditional detectors. It allows us to image the tiny signals of (only a fraction of) the daily proton therapy treatment dose with high accuracy.’ The increased sensitivity of the detectors also means that it takes less than a minute to complete this measurement. That is important as a shorter treatment allows more patient to receive proton therapy each day. Rasch: ‘The shorter, the better. But proton therapy is a high-end treatment, taking as long as needed.’

The ideal location
‘HollandPTC is of paramount importance to this project,’ Schaart says. ‘The experts who reviewed our project proposal on behalf of KWF, stress that it is the only location where a project like ours can succeed, as it combines the expertise of the participating medical centres with the technical expertise of TU Delft.’ The PhD student of the project will also need to do a lot of measurements using realistic, human-like phantoms. This is where the HollandPTC R&D bunker comes into play. It features a proton beam of clinical quality that researchers can use without interfering with the clinical programme. ‘We also need access to anonymised clinical data,’ Schaart says. ‘Think of realistic proton therapy treatment plans, CT-scans, you name it. HollandPTC has been setup with this in mind. Even a TU Delft researcher can access such information without risking privacy violations. This could otherwise have been a considerable hurdle to take for this kind of research.’

A long-term project
The end goal of the project, after four years of research, is to show that it is technically feasible to verify the dose delivered to a patient to within a few percent. After this proof-of-concept it will take a follow-up project – lasting another four years – to build a prototype that can be tested within the clinical treatment program at HollandPTC. ‘We want take the idea up to the point where commercial parties believe it to be a viable product,’ Schaart says. ‘It requires us, at TU Delft, to not only do research and development, but to also be continually aware of the regulations and standards of medical products.’

Plenty more coffee
For Rasch, that is much too long a timeline. For his entire career he has been a front-runner regarding the introduction and use of image-guidance for radiation treatments. ‘The clinical consequences of this development will become apparent much faster, such as the minimum dose needed to achieve a certain accuracy in where the protons stop in the patient’s body. Once you know these consequences, you can already start to think about clinical implementation of this technology. We could, for example, choose to deliver a very low dose with a somewhat larger safety margin. Based on the in-vivo dosimetry measurement, we would be confident in delivering the remaining dose using a much smaller margin.’ Medical-technical innovation is often the culmination of an iterative design process in which technical outcomes and clinician’s wishes are continually harmonized. Schaart: ‘We will continue to meet regularly for the duration of the project. It will take plenty more coffee for our idea to come to full fruition.’

#Jaap Zindler of ErasmusMC and Dennis Vriens of LUMC co-applicants.