This doctoral thesis focuses on the design, implementation, and experimental validation of a gamma-ray detector for real-time single photon emission computed tomography (SPECT) imaging during boron neutron capture therapy (BNCT). The work introduces artificial neural networks (ANNs)-based reconstruction algorithms and their deployment on a field-programmable gate array (FPGA) for low-latency online processing, a first SPECT setup, and a comprehensive experimental validation campaign. BNCT is a highly selective form of hadron therapy that exploits the nuclear reaction between thermal neutrons and 10B, which accumulates in tumour cells through suitable carrier compounds. This reaction yields two particles with high linear energy transfer (LET) that deposit their energy over short distances, thereby confining the biological effect to the targeted tumour volume. Due to this intrinsic selectivity, BNCT is ideal for treating infiltrative or diffuse malignancies and lesions in close proximity to critical organs. Despite its potential, the widespread clinical implementation of BNCT has historically been challenged by several factors. These include the limited availability of neutron sources capable of providing clinically relevant flux, the necessity of highly selective and non-toxic boronated agents, and the absence of reliable systems for real-time dose monitoring. However, recent progress in compact accelerator-based neutron generators has renewed interest in BNCT. A growing number of clinical and preclinical facilities are either in operation or under development worldwide. In this context, there is an urgent need for advanced BNCT-related technologies. In-beam dosimetric imaging systems, such as the one investigated in this study, play a pivotal role in this regard. The main objective of this thesis is to quantify and localise the dose delivered to patients undergoing this therapy. To achieve this, we developed a prototype of a single-module BNCT-SPECT system by integrating the BeNEdiCTE detector, developed within a parallel PhD project, with a dedicated imaging chain comprising a collimator and a reconstruction algorithm based on ANNs. Moreover, this study investigates the feasibility of implementing the ANN-based algorithm on the system’s FPGA for real-time information processing. Throughout this work, three successive versions of the reconstruction algorithm are introduced. Initially, the algorithm relies solely on the reconstructed x and y coordinates of interaction. Subsequently, the algorithm is extended to incorporate depth of interaction (DOI) information. A comprehensive experimental campaign was carried out at two neutron facilities: the TRIGA Mark II research reactor at the University of Pavia in Italy and the NUANS accelerator at Nagoya University in Japan. At the former facility, the proposed system enabled the first tomographic reconstruction of two boron-containing vials placed 1 cm apart under neutron irradiation. At the latter facility, vial activity profiles at different boron concentrations were successfully reconstructed under clinically representative neutron field conditions. Taken together, these results provide strong evidence that the BNCT-SPECT system proposed and developed in this thesis has substantial potential for clinical application in the future.
Questa tesi di dottorato è dedicata alla progettazione, implementazione e validazione sperimentale di un rivelatore di raggi gamma per l’imaging tomografico a emissione di singolo fotone (SPECT) in tempo reale durante la terapia di cattura neutronica del boro (BNCT). Il lavoro introduce algoritmi di ricostruzione basati su reti neurali artificiali (ANN) e la loro implementazione su una field-programmable gate array (FPGA) per l’elaborazione online a bassa latenza, oltre alla realizzazione di un primo sistema SPECT e a una campagna completa di validazione sperimentale. La BNCT è una forma altamente selettiva di adroterapia che sfrutta la reazione nucleare tra neutroni termici e ^10B, il quale si accumula nelle cellule tumorali attraverso opportuni composti vettori. Questa reazione produce due particelle ad alto trasferimento lineare di energia (LET), che depositano la loro energia su distanze molto brevi, confinando così l’effetto biologico al volume tumorale bersaglio. Grazie a questa selettività intrinseca, la BNCT risulta particolarmente adatta al trattamento di tumori infiltranti o diffusi e di lesioni situate in prossimità di organi critici. Nonostante il suo potenziale, l’ampia implementazione clinica della BNCT è stata storicamente limitata da diversi fattori. Tra questi vi sono la limitata disponibilità di sorgenti neutroniche in grado di fornire flussi clinicamente rilevanti, la necessità di agenti boronati altamente selettivi e non tossici e l’assenza di sistemi affidabili per il monitoraggio della dose in tempo reale. Tuttavia, i recenti progressi nello sviluppo di generatori di neutroni compatti basati su acceleratore hanno riacceso l’interesse per la BNCT. Un numero crescente di strutture cliniche e precliniche è oggi operativo o in fase di sviluppo a livello mondiale. In questo contesto emerge un’esigenza urgente di tecnologie avanzate dedicate alla BNCT. I sistemi di imaging dosimetrico in-beam, come quello investigato in questo studio, rivestono un ruolo fondamentale in tal senso. L’obiettivo principale di questa tesi è quantificare e localizzare la dose somministrata ai pazienti sottoposti a questa terapia. A tal fine è stato sviluppato un prototipo di sistema BNCT-SPECT a singolo modulo, ottenuto integrando il rivelatore BeNEdiCTE, sviluppato nell’ambito di un progetto di dottorato parallelo, con una catena di imaging dedicata composta da un collimatore e da un algoritmo di ricostruzione basato su ANN. Inoltre, questo studio analizza la fattibilità dell’implementazione dell’algoritmo basato su ANN sulla FPGA del sistema per l’elaborazione delle informazioni in tempo reale. Nel corso di questo lavoro sono state sviluppate tre versioni successive dell’algoritmo di ricostruzione. Inizialmente l’algoritmo utilizza esclusivamente le coordinate ricostruite di interazione x e y. Successivamente viene esteso per includere anche l’informazione di profondità di interazione (DOI). È stata inoltre condotta un’ampia campagna sperimentale presso due strutture neutroniche: il reattore di ricerca TRIGA Mark II dell’Università di Pavia, in Italia, e l’acceleratore NUANS della Nagoya University, in Giappone. Presso la prima struttura, il sistema proposto ha consentito la prima ricostruzione tomografica di due provette contenenti boro poste a 1 cm di distanza sotto irraggiamento neutronico. Presso la seconda struttura, sono stati ricostruiti con successo i profili di attività di provette con diverse concentrazioni di boro in condizioni di campo neutronico rappresentative di un contesto clinico. Nel complesso, questi risultati forniscono solide evidenze che il sistema BNCT-SPECT proposto e sviluppato in questa tesi possiede un notevole potenziale per future applicazioni cliniche.
Design and validation of a SPECT imaging detector towards real-time dose monitoring in BNCT
Ferri, Tommaso
2026
Abstract
This doctoral thesis focuses on the design, implementation, and experimental validation of a gamma-ray detector for real-time single photon emission computed tomography (SPECT) imaging during boron neutron capture therapy (BNCT). The work introduces artificial neural networks (ANNs)-based reconstruction algorithms and their deployment on a field-programmable gate array (FPGA) for low-latency online processing, a first SPECT setup, and a comprehensive experimental validation campaign. BNCT is a highly selective form of hadron therapy that exploits the nuclear reaction between thermal neutrons and 10B, which accumulates in tumour cells through suitable carrier compounds. This reaction yields two particles with high linear energy transfer (LET) that deposit their energy over short distances, thereby confining the biological effect to the targeted tumour volume. Due to this intrinsic selectivity, BNCT is ideal for treating infiltrative or diffuse malignancies and lesions in close proximity to critical organs. Despite its potential, the widespread clinical implementation of BNCT has historically been challenged by several factors. These include the limited availability of neutron sources capable of providing clinically relevant flux, the necessity of highly selective and non-toxic boronated agents, and the absence of reliable systems for real-time dose monitoring. However, recent progress in compact accelerator-based neutron generators has renewed interest in BNCT. A growing number of clinical and preclinical facilities are either in operation or under development worldwide. In this context, there is an urgent need for advanced BNCT-related technologies. In-beam dosimetric imaging systems, such as the one investigated in this study, play a pivotal role in this regard. The main objective of this thesis is to quantify and localise the dose delivered to patients undergoing this therapy. To achieve this, we developed a prototype of a single-module BNCT-SPECT system by integrating the BeNEdiCTE detector, developed within a parallel PhD project, with a dedicated imaging chain comprising a collimator and a reconstruction algorithm based on ANNs. Moreover, this study investigates the feasibility of implementing the ANN-based algorithm on the system’s FPGA for real-time information processing. Throughout this work, three successive versions of the reconstruction algorithm are introduced. Initially, the algorithm relies solely on the reconstructed x and y coordinates of interaction. Subsequently, the algorithm is extended to incorporate depth of interaction (DOI) information. A comprehensive experimental campaign was carried out at two neutron facilities: the TRIGA Mark II research reactor at the University of Pavia in Italy and the NUANS accelerator at Nagoya University in Japan. At the former facility, the proposed system enabled the first tomographic reconstruction of two boron-containing vials placed 1 cm apart under neutron irradiation. At the latter facility, vial activity profiles at different boron concentrations were successfully reconstructed under clinically representative neutron field conditions. Taken together, these results provide strong evidence that the BNCT-SPECT system proposed and developed in this thesis has substantial potential for clinical application in the future.| File | Dimensione | Formato | |
|---|---|---|---|
|
PhD_Thesis_Tommaso_Ferri_FINALE.pdf
non disponibili
Licenza:
Tutti i diritti riservati
Dimensione
18.34 MB
Formato
Adobe PDF
|
18.34 MB | Adobe PDF |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/376642
URN:NBN:IT:POLIMI-376642