High-speed neutron imaging by using AI super-resolution technique

2023 Fiscal Year Final Research Report

• Project/Area Number: 21K18624
• Research Category: Grant-in-Aid for Challenging Research (Exploratory)
• Allocation Type: Multi-year Fund
• Review Section: Medium-sized Section 15:Particle-, nuclear-, astro-physics, and related fields
• Research Institution: The University of Tokyo
• Principal Investigator: Inada Toshiaki 東京大学, 素粒子物理国際研究センター, 助教 (20779269)
• Project Period (FY): 2021-07-09 – 2024-03-31
• Keywords: 中性子

Outline of Final Research Achievements

Understanding the basic property of charge tracks generated by two kinds of secondary particles in neutron capture events in a 10B layer (n + 10B -> 7Li + α) is very important in this research. The progress of such study was obtained as planned, by using GEANT4 Monte Carlo simulations. Our deep-learning approach improved the performance of secondary-particle identification and neutron spatial resolution, compared to the conventional method using a cluster size and total charge. We obtained a steady progress in the feasibility study of AI-based super-resolution and established the R&D environment for future hardware integration.

Free Research Field

素粒子・原子核・宇宙線・宇宙物理

Academic Significance and Societal Importance of the Research Achievements

AIを活用して粗視画像の高解像化を図る技術の開発は、一次データのサイズを削減可能であるため、高いフレームレートの動画撮影や大面積のイメージングに有用となる。本研究はそのような高効率撮像技術を実現するための基礎研究として重要であり、中性子に限らず様々な２次元測定に応用可能であると考える。