ESTRO 2024 - Abstract Book

S3291

Physics - Detectors, dose measurement and phantoms

ESTRO 2024

Purpose/Objective:

Positron emission tomography (PET) is often considered as an useful method for in-vivo protontherapy range verification. It is based on the detection of the activity arising from β+ isotopes generated by protons through nuclear reactions [1] in irradiated tissues. The β+ activity can be correlated to the dose deposited in the patient. Contrary to other methods to obtain dose maps from PET activation maps, in-beam PET is more attractive in terms of obtaining fast feedback from the treatment, allowing for the detection of intra-fractional errors and for the possibility to reduce the biological washout of the PET activity. However, it is challenging for the detectors and electronics due to the high event rates during the beam-on period. For its clinical implementation, the dose reconstruction algorithm must be integrated on-the-fly and should be fast enough to ensure real on-line feedback of the dose in the patient. This has not been achieved to date.

In this work, we report about on-the-fly dose reconstruction from clinical in-beam PET data, using a novel In-beam Dose Estimation tool (IDE-PET), capable of obtaining on-line dose and of detecting range deviations.

Material/Methods:

We have designed, developed and constructed an in-beam PET scanner (miniPET) to measure on-line PET activity. It consists of one ring of 6 phoswich detector blocks, each consisting of 338 pixels of 1.55 mm crystal pitch with a dual layer (phoswich) made of LYSO (7mm)+GSO(8mm) crystals attached to PMTs. The scanner is coupled to a fast data acquisition system able to sustain rates of up to 10 Msingles/s. All the experiments were performed at the Quirónsalud proton therapy center in Madrid, Spain. Data were acquired in singles mode and coincidences were sorted out in real-time, with 10 ns and 450-650 keV coincidence and energy windows. Here we report on the results with a monoenergetic proton beam of 70 MeV oriented along the longitudinal axis of the PET scanner. Several cylindrical (50-mm diameter and 50-mm height) homogeneous PMMA phantoms were irradiated. Phantoms were fixed in a holder placed in the center of the Field Of View (FOV) of the miniPET scanner and aligned with the gantry isocenter on the treatment bed. Next to the homogeneous configuration, 5 PMMA range shifters foils of varying thickness (from 1 to 5 mm) were also placed at the proximal surface to investigate range shift prediction accuracy. We have developed an IDE-PET tool that can be used for real-time dose estimation. It combines a GPU-based 3D reconstruction algorithm [2], which is used to reconstruct the 3D images from the list file, in less than a second. For dose estimation, we have implemented a dictionary-based software capable of estimating deposited doses from the 3D PET activity images [3]. By combining these two tools, IDE-PET enables online display of deposited dose and proton range deviations.

Results:

3D maps of the PET activity were reconstructed on-the-fly every 0.5 seconds. The dose estimation algorithm requires from 0.5 to 2 seconds to calculate and display the deposited dose (see Figure 1). The dose is computed with good accuracy even during the first seconds after the beginning of the irradiation, during the beam-on period. For a 2 Gy dose fraction, the method was able to spot range variations as small as 1 mm, demonstrating the fall-off region of the Bragg Peak (BP) is well reproduced. Mean differences in range shift, between reconstructed dose profiles and