Speaker
Description
Elena Fogazzi1,2, Mara Bruzzi3,4, Elvira D’Amato1, Paolo Farace2,5, Francesco Fracchiolla2,5, Stefano Lorentini2,5, Roberto Righetto2,5, Monica Scaringella4, Marina Scarpa1,2, Francesco Tommasino1,2, Carlo Civinini4*
1 Physics department, University of Trento, via Sommarive 14, Povo (TN), Italy 2 Trento Institute for Fundamental Physics and Applications (TIFPA), Italian National Institute of Nuclear Physics (INFN), via Sommarive, 14, Povo (TN), Italy
3 Physics and Astronomy department, University of Florence, via G. Sansone 1, Sesto Fiorentino (FI), Italy
4 Italian National Institute of Nuclear Physics (INFN), Florence section, Via G. Sansone 1, Sesto Fiorentino (FI), Italy
5 Medical Physics Unit, Hospital of Trento, Azienda Provinciale per i Servizi Sanitari (APSS), Via Paolo Orsi 1, Trento, Italy
*Responsabile Nazionale del progetto XpCalib
Introduction: The dose computation in the proton treatment planning system (TPS) is based on the proton relative stopping power normalized to liquid water (RSP) distribution in the target volume. Presently, the RSP maps are extracted from x-ray computed tomographies (xCT) of the patient. Namely, the photon attenuation coefficients (CT Hounsfield Units – HU), are translated into RSP values using empirical methods based on conversion tables. These methods introduce an uncertainty on the actual position of the Bragg peak inside the patient, which has to be mitigated by means of the use of safety margins around the target and organs at risk. To avoid this two-step process and to reduce the intrinsic errors, we propose a different approach based on the direct use of 3D RSP maps obtained with a proton computed tomography (pCT) system. Herein, we present the main results of the experiment XpCalib, funded by CSN5 INFN (2020-2023) and based on a pCT system previously developed [1].
Methods: The pCT system, tested at the Trento Proton Therapy Center, is made of four planes of silicon micro-strip trackers and a YAG:Ce scintillating calorimeter [1] (Fig1). A filtered backprojection algorithm, taking into account the protons’ most likely path, allowed reconstructing the phantoms’ RSP 3D maps. The imaging performances (i.e. spatial resolution, noise power spectrum, RSP accuracy) were assessed on a custom-made phantom, made of plastic materials with different densities (0.66-2.18 g/cm3), in two background conditions (liquid water or air) 2 . Then, moving towards more clinical scenarios, we designed the first clinical application for the INFN proton computed tomography (pCT) system through the realization of biological phantoms [3]. Namely, the bio-phantom is made of biologic inserts of a bovine/porcine specimen, stabilized with a formalin solution and embedded in agar-agar gel in a plastic housing (Fig2). Both pCT and xCT images were acquired on these phantoms. The direct, voxel-by-voxel comparison of HU and RSP maps of the biological phantom provides a cross-calibrated xCT calibration curve, i.e. a RSP-HUs look-up table, improving the description provided by the existing calibration methods [4].
Results: Overall, the system results to have imaging performances comparable to the x-ray CT with standard imaging protocol for proton therapy. Moreover, the system is highly accurate, with a mean absolute percentage error on the measured RSP values well below 1% [2,5]. Comparing the bio-phantom data acquired with the pCT system with the one calculated with conventional xCT calibration curve, we obtained that the vast majority of pixel data, falling within regions of fat and muscles, shows differences within 2.46% on average. In the bone region, the conventional calibrations overestimate the pCT-measured SPR of the phantom, with a maximum discrepancy of 4% on average, corroborating previous results in literature. As a result, a new cross-calibration curve is directly extracted from the pCT data in the HU range ([-109, 1536]). Additionally, the associated uncertainty is below 3%, that is less than the standard error of conventional calibration curves adopted in clinics.
Conclusion: The obtained performances showed that the INFN pCT system provides a very accurate RSP estimation, and it can be used as a reference RSP measuring method for the verification of the xCT calibration in proton treatment planning, and, eventually, for the implementation of a new cross-calibration curve. This could allow reducing range uncertainty and margin size in proton therapy treatments.
[1] Civinini C, Scaringella M, Brianzi M, Intravaia M, Randazzo N, Sipala V, Rovituso M, Tommasino F, Schwarz M, Bruzzi M. Relative stopping power measurements and prosthesis artifacts reduction in proton CT. Phys. Med. Biol. 2020; 65(22), 225012
[2] Fogazzi E, Trevisan D, Farace P, Righetto R, Rit S, Scaringella M, Bruzzi M, Tommasino T, Civinini. Characterisation of the INFN proton CT scanner for cross-calibration of x-ray CT. Phys. Med. Biol. 2023; 68 124001
[3] Fogazzi E et al, in preparation
[4] Farace P, Tommasino F, Righetto R, Fracchiolla F, Scaringella M, Bruzzi M, Civinini C. Technical Note: CT calibration for proton treatment planning by cross-calibration with proton CT data. Med Phys. 2021 Mar;48(3):1349-1355.
[5] Scaringella M, Bruzzi M, Farace P, Fogazzi E, Righetto R, Rit R, Tommasino T, Verroi E, Civinini C.The INFN proton computed tomography system for relative stopping power measurements: calibration and verification. Phys. Med. Biol. 2023; 68 154001