탄소 빔 비정 검증을 위한 속중성자 산란 영상장비 개발 = Development of Fast Neutron Scatter Imager for Carbon Beam Range Verification

다국어 초록 (Multilingual Abstract)

Carbon-ion radiotherapy (CIRT) uses the Bragg peak to deliver precise dose distributions. However, the generation of secondary neutrons and uncertainty in the beam range can affect treatment efficacy and result in damage to organs at risk. To ensure patient safety, it is crucial to characterize secondary neutrons and implement real-time dose verification during treatment. Compared with proton beams, carbon-ion beams produce significantly higher yields of secondary neutrons. In particular, carbon beams generate high-energy neutrons of up to several hundred MeV, making them well-suited for detection through neutron scatter imaging techniques. Therefore, the use of neutron-based monitoring in CIRT affords improved detection efficiency and contributes to enhanced accuracy for range verification.
In this thesis, the design of a fast neutron scatter imager for CIRT beam range verification was optimized using Geant4-based Monte Carlo simulations. The design of the neutron scatter imager was optimized for three key parameters: detector thickness, inter-detector distance, and pixel size. Furthermore, the effectiveness of lead (Pb) shielding in reducing the gamma-ray counts was evaluated. The prototype fast neutron scatter imager developed in this study consists of two position-sensitive detectors utilizing pixelated organic scintillators (EJ-276 and stilbene, 4×4 pixel array, 3×3×10 mm3 per pixel) coupled to 4×4 SiPM arrays (S13361-3050AE-04, Hamamatsu). Signals were read out through dedicated signal-processing electronics provided by AiT, and data acquisition was performed using a VME-based multi-channel digitizer (V1730SB, CAEN).
The performance of the proof-of-principle detector system was evaluated in terms of various parameters (i.e., position, energy, and timing resolution, detection efficiency, and pulse shape discrimination (PSD)). The stilbene and EJ-276 detectors showed average energy resolutions of 17.5% and 22.5% (at 662 keV peak), and constant-fraction discriminator (CFD)-based timing resolutions of 0.328 ns and 0.669 ns, respectively. The average neutron detection efficiencies were measured as 2.08×10⁻⁶±2.96×10⁻⁷ for stilbene and 1.98×10⁻⁶±2.33×10⁻⁷ for EJ-276. The PSD performance demonstrated clear neutron–gamma separation, with a figure-of-merit (FOM) of 1.27 for stilbene and 0.78 for EJ-276. Using a 252Cf source, neutron images were successfully reconstructed through TOF- and PSD-based coincidence event selection. Image resolution was achieved at approximately 4.22 cm FWHM using simple back-projection (SBP) and at 1.97 cm FWHM using maximum-likelihood expectation-maximisation (MLEM).
Finally, the feasibility of a fast neutron scatter imager to carbon-beam range verification was evaluated using Geant4 simulations for carbon-ion beam energies of 150–430 MeV/u. Secondary neutron distributions were reconstructed using the fast neutron scatter imaging technique, and analysis of their correlation with the true beam range confirmed the practical potential of the system for carbon-beam range verification.

목차 (Table of Contents)

• TABLE OF CONTENTS ⅰ
• LIST OF FIGURES ⅲ
• LIST OF TABLES ⅺ
• ABSTRACT ⅻ
• I. INTRODUCTION 1
• II. FUNDAMENTALS OF NEUTRON SCATTER IMAGING 4
• 2.1 Fast Neutron Detection 4
• 2.2 Principle of Neutron Scatter Imaging 10
• 2.3 Effects of Various Parameters on Image Resolution 13
• Ⅲ. DESIGN OPTIMIZATION OF NEUTRON DETECTION SYSTEM 28
• 3.1 Characteristics of Secondary Particles 28
• 3.2 Optimization of Detector Parameters 36
• 3.2.1 Detector Thickness 36
• 3.2.2 Inter-Detector Distance 39
• 3.2.3 Pixel Size 43
• 3.2.4 Pb Shield Thickness 45
• Ⅳ. DEVELOPMENT OF FAST NEUTRON SCATTER IMAGER 47
• 4.1 System Components 47
• 4.1.1 Pixelated Scintillator Array Detector 47
• 4.1.2 Silicon Photomultiplier (SiPM) 51
• 4.1.3 Readout Electronics and Data Acquisition (DAQ) 53
• 4.1.4 Mechanical Structure 60
• 4.2 Pulse Shape Discrimination (PSD) 63
• 4.2.1 Charge Comparison Method (CCM) 63
• 4.2.2 Convolutional Neural Network (CNN)-based PSD 68
• 4.3 Performance Evaluation 77
• 4.3.1 Position Resolution 77
• 4.3.2 Energy Resolution 81
• 4.3.3 Timing Resolution 89
• 4.3.4 Detection Efficiency 95
• 4.3.5 PSD Performance 98
• Ⅴ. IMAGING EXPERIMENTS 101
• 5.1. Experimental Setup and Calibration 101
• 5.2. Data Processing 105
• 5.3. Image Reconstruction 107
• Ⅵ. APPLICATION OF NEUTRON SCATTER IMAGING IN CARBON-ION RADIOTHERAPY (CIRT) 112
• Ⅶ. CONCLUSIONS AND FUTURE WORK 125
• 7.1 Summary and Conclusions 125
• 7.2 Suggestions for Future Work 128
• REFERENCES 129
• ABSTRACT (In Korean) 137
• ACKNOWLEDGEMENTS (In Korean) 139