Impact of dose uncertainty of tracking irradiation in three-dimensional measurements using a polymer-gel dosimetry with deep learning

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K08097
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 52040:Radiological sciences-related
• Research Institution: National Cancer Center Japan
• Principal Investigator: Hidenobu Tachibana 国立研究開発法人国立がん研究センター, 東病院, 室長 (20450215)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: ゲル線量計 / 深層学習 / 放射線治療 / MR

Outline of Final Research Achievements

The principles and implementation of deep learning for high resolution and noise reduction (DL-HRNR) have been completed, but it has been discovered that obtaining high-quality data is difficult and challenging. In the future, it is believed that obtaining training data for DL-HRNR can be achieved through high-speed sequences and low-noise sequences in MR imaging. Additionally, conducting MR imaging of gel dosimeters with various dose distribution patterns will be necessary to obtain a larger amount of training data.

Free Research Field

放射線治療医学物理学

Academic Significance and Societal Importance of the Research Achievements

深層学習が全てを解決できるわけではなく、よい教師データが必要であり、本研究におけるゲル線量計の教師データ取得が難しいことがわかった。
MRによるゲル線量計の画像化は歴史が古く、実績が多いため、デファクトスタンタードであり、信頼性も高いが、利便性の観点からすると、別のモダリティへのシフトが必要であると言える。