Oral presentationExperimental validation of the filtering approach for dose monitoring in proton therapy at low energy
Introduction
Most efforts in external beam radiotherapy are aimed to achieve the deposition of higher radiation doses to the tumor regions sparing as much as possible the surrounding healthy tissues.
In this context, the irradiation with protons offers physical advantages over the common irradiation modalities, thanks to a reduced lateral spreading when penetrating in matter and to a localized dose deposition in depth.
However, the full clinical exploitation of proton selectivity could be hampered, even during the pre-treatment phase, by the difficulty in superimposing the dose distal fall-off on the tumor boundaries. Currently the treatment planning (TP) calculations can be affected by few percent range uncertainties, which are intrinsic to the usage of calibration curves between CT photon attenuation coefficients and proton stopping power [1]. Besides, since the total therapeutic dose is often delivered in more than one fraction, further different sources of errors may arise from daily setup variations, internal organ motions and anatomical or physiological changes. As a result, the availability of a dosimetry tool turns out to be a prerequisite for quality assurance in proton therapy.
A non-invasive method to verify the accuracy of treatment delivery is provided by Positron Emission Tomography. It is based [2], [3], [4], [5] on the detection of the β+-activity induced by nuclear reactions between incident particles and target nuclei. Unfortunately, due to the energy thresholds of the isotope activation reactions (typically 15–20 MeV), the distal 50% activity level is located few millimeters before the dose peak position. This fact, along with the different nature of the interactions leading to the isotope activations and the energy delivery, makes the dose reconstruction a non-trivial task.
The dose verification is usually limited to an indirect test: the measured PET image is compared with the simulated activity by means of a Monte Carlo (MC) code. However, a simpler and more direct way to reconstruct the expected PET signal from the planned radiation dose distribution has been recently tried out in [6], where a filtering approach is proposed to replace of the full-blown MC simulation use.
This work is focused on the possibility to apply this method to an unexplored energy range ([40, 70] MeV) which could be more critical with respect to higher energy ranges, because of the greater relative incidence of the uncertainties about the nuclear cross sections in the lower part of the energy spectrum. Moreover, due to a smaller relevance of the straggling at the energies which we are dealing with, any imprecision in the cross section peak positions survives to the straggling smoothening and therefore will be reflected in a more severe disagreement between theory and experimental observation.
The promising results obtained, that are shown in this paper, provide support to the hypothesis of the existence of a universal filter function to convert dose into activity profile whatever is the beam energy. This encourages the efforts to invert the filter itself and to obtain a direct information about dose localization.
The paper is organized as follows. In the Materials and methods we describe our experimental setup and the process of data collection performed at CATANA (Centro di AdroTerapia e Applicazioni Nucleari Avanzate) at INFN–LNS (Istituto Nazionale di Fisica Nucleare – Laboratori Nazionali del Sud) in Catania (Italy), where 62 MeV proton beams are employed for the treatment of ocular melanomas [7]. The achievable accuracy in range monitoring is presented and discussed in the Data analysis. In the Unfolding we briefly describe the filtering approach and the model we used to calculate the dose curve and the associated activity profile. The application of theoretical filter to measured activity images is reported in the Results and discussion, along with a new separation method of the activity contributions of several isotope species. The conclusions follow in the last section.
Section snippets
The PET prototype
The dedicated PET prototype consists of two planar heads, each one providing an active area of about 5 cm × 5 cm. Each head is made up of a squared multianode photomultiplier (Hamamatsu H8500 [8]) coupled to a matrix of the same size of LYSO scintillating crystals (2 mm × 2 mm × 18 mm pixel dimensions). A multiplexed readout [9] is used to reduce the number of collected signals from each head to four signals, while dedicated, compact electronic boards are used for signal amplification and digitization.
Feasibility and accuracy of PET dose monitoring
It is a plain issue that the first, most meaningful, information one hopes to recover through the PET image concerns the depth–dose profile rather than its transverse distribution.
Fig. 1 shows the lateral activity distributions as reconstructed for three spread-out Bragg peaks (SOBPs) of 10.8 mm plateau width. Range shifters were placed along the beam line so that each irradiation differs from the other ones only in the proton range, with variations less than 2 mm.
The distal half-maximum activity
The filtering approach
The method proposed in Ref. [6] relies on the description of the PET image as a convolution of the dose distribution with a filter function. At its basis there is the introduction of functions, defined as the convolution of a Gaussian with a power law function. The convolution integral has a closed form representation in terms of the parabolic cylinders functions [12], [13] which depend on the parameter ν. In particular it has been proved that they are particularly suitable to
Testing the theoretical filters on experimental data
Fig. 2 shows the experimental β+-activity depth profiles for a monoenergetic proton beam and two spread-out Bragg peaks of 10.8 mm and 17.9 mm plateau width, respectively, obtained by integrating the measured 3D-data over the entire lateral extension of the field. For comparison we superimpose the distributions obtained by applying the filter to the depth dose profiles, smoothed with a Gaussian point spread function of 1.7 mm FWHM to account for the spatial resolution of the PET prototype. In all
Summary and conclusions
Proton dose monitoring with PET techniques and its application to proton radiotherapy has been investigated in the energy range [40, 70] MeV by the activation of PMMA phantoms, with promising results as for the achievable accuracy in range monitoring along beam direction.
The poor spatial correlation between the depth dose profiles and the corresponding depth activity profiles limits, at the moment, the dose verification to an indirect test. At the purpose, a novel approach for one-dimensional
References (14)
- et al.
Positron emission tomography as an aid in “vivo” dosimetry for proton radiotherapy: a Monte Carlo simulation
Nucl Instrum Meth A
(1994) - et al.
Charged hadron tumour therapy monitoring by means of PET
Nucl Instrum Meth
(2004) - et al.
Performance of a four-output front-end electronics for multi-anode PMTS readout of scintillator arrays
Nucl Instrum Meth A
(2007) - et al.
The precision of proton range calculations in proton radiotherapy treatment planning: experimental verification of the relation between CT-HU and proton stopping power
Phys Med Biol
(1998) - et al.
Visualization and transport of positron emission from proton activation in vivo
Science
(1978) - et al.
Proton dose monitoring with PET: quantitative studies in Lucite
Phys Med Biol
(1996) - et al.
A filtering approach based on Gaussian–powerlaw convolutions for local PET verification of proton radiotherapy
Phys Med Biol
(2006)