• in-beam PET
• range monitoring
• charged particle therapy

Charged particle therapy is a tumour treatment whose aim is to eradicate cancer cells by using mainly protons and carbon ions. In the past few decades, it gained more and more importance: with respect to conventional radiotherapy (i.e., with photons), it offers many theoretical advantages in both physical and biological aspects. In fact, conventional radiotherapy beams show an exponential decrease in dose deposition, while charged particles release most of the energy at the end of their range in tissue. This would allow to deliver higher doses to the tumour and to spare surrounding healthy tissues. Nevertheless, particle therapy is very sensitive to range uncertainties, which could cause severe under-dosage to the tumour or overdosage to the healthy tissues and organs at risk. Thus, large safety margins are introduced in clinical practice, preventing to fully exploit the intrinsic potential of particle therapy. Hence, precise in vivo range monitoring is required. Over the years, several strategies for treatment verification have been proposed. Positron Emission Tomography (PET) imaging is the most consolidated technique. PET exploits the beam-induced activity inside the patient to detect possible differences between the prescribed and delivered dose. In particular, in-beam PET is one of the monitoring options. The acquisitions are performed during the treatment fraction by using dedicated PET detectors, compatible to the beam delivery system and treatment room geometry.
The work presented in this thesis was carried out in the framework of the INnovative Solutions for In-beam Dosimetry in hadronthErapy (INSIDE) project, whose aim was to develop and validate a bi-modal range verification system composed of an in-beam dual head PET scanner and a charged particle tracker. Since 2016 it has been installed and tested at Centro Nazionale di Adroterapia Oncologica (CNAO), Pavia.
Based on the results obtained in the pre-clinical phase, a clinical trial was approved by the ethics committee of the San Matteo polyclinic in Pavia to investigate the performances of the INSIDE bi-modal system. The trial consists of two phases, phase I and phase II. The first one started in July 2019 and ended in March 2020, while the second one will start in the next months. Phase I ended with the recruitment and monitoring of 20 patients (10 treated with protons and 10 with carbon ions), while during phase II will be recruited and monitored other 20 patients to increase the statistics collected during the first phase of measurement. In particular, patients were monitored during the therapy period, which consisted of a variable number of daily treatment fractions. Patients suffering from diseases that do not present an immediate response to the treatment and patients affected by early response pathologies (i.e. commonly showing morphological changes during the treatment period) were selected.
First objective of the present work was the quantification of the INSIDE PET sensitivity in detecting activity range variations. Second objective was testing the system capability to spot morphological changes occurred during the treatment period.
Two different analysis methods, Beam Eye View (BEV) and Most-Likely Shift (MLS), were implemented and tested on the acquired PET images. In particular, for each patient the activity distribution of the first monitored fraction was compared with the subsequent ones. Both BEV and MLS allow to perform a profile-by-profile evaluation along the beam direction and to obtain a bi-dimensional matrix of range deviations (called activity range
difference map) on the axial plane. The sensitivity of the INSIDE PET range verification system was statistically quantified as the dispersion of the obtained activity range difference distributions (i.e. the unidimensional representation of the range difference maps).
BEV and MLS analysis methods were tested on 9 cases of patients treated with protons. A sensitivity in activity range difference detection of 6.4 mm and 5.0 mm Inter Quartile Distance (IQR) was found for BEV and MLS, respectively. The analysis allowed to detect morphological changes occurred in 3 patients. For 2 patients, the result was proved by the control Computed Tomography (CT) acquired during the therapy period. The comparison between the two range verification methods highlighted their differences. In particular, both MLS and BEV were sensitive in detecting morphological changes. However, BEV needed to be adapted for each studied case: an optimization of the threshold value used for the iso-activity surface extraction was required to detect the occurred changes and to maximize the range differences. On the contrary, MLS was more sensitive in detecting changes that affect a considerable part of the distal fall-off region.
Based on these results, it was proposed an optimization study for the BEV method and a further evaluation of the MLS approach robustness and stability. To this purpose, the data collected during the upcoming second phase of the clinical trial will be essential to validate the studied analysis methods and their impact in the clinical practice.