• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Realization of next-generation adaptive radiation therapy using virtual CT image constructed with convolution neural network

Research Project

Project/Area Number 18K15563
Research Category

Grant-in-Aid for Early-Career Scientists

Allocation TypeMulti-year Fund
Review Section Basic Section 52040:Radiological sciences-related
Research InstitutionJuntendo University

Principal Investigator

USUI KEISUKE  順天堂大学, 保健医療学部, 講師 (20714132)

Project Period (FY) 2018-04-01 – 2021-03-31
Project Status Completed (Fiscal Year 2020)
Budget Amount *help
¥4,160,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥960,000)
Fiscal Year 2019: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2018: ¥3,510,000 (Direct Cost: ¥2,700,000、Indirect Cost: ¥810,000)
Keywords放射線治療学 / 適応的放射線治療 / コーンビームCT / 人工知能 / 深層学習 / 放射線治療技術学 / シミュレーション / がん放射線療法 / 畳み込みニューラルネット / 放射線治療 / 前立腺癌 / 画像誘導放射線治療 / 畳み込みニューラルネットワーク
Outline of Final Research Achievements

The purpose of this task was to develop an adaptive radiotherapy planning method with prediction of toxicity after treatment using cone-beam CT images acquired during radiotherapy. In this study, we examined three studies: ①improvement of image quality of cone beam CT using deep learning, ②image quality evaluation using quantitative index, and ③construction of a training model for predicting toxicity from cone beam CT images. In ①, we constructed a convolutional neural network to improve the image quality deterioration due to scattered photons generated from the subject. In ②, the image quality was evaluated using the image similarity and the maximum signal-to-noise ratio, and an improvement effect of about 10% was achieved compared to the conventional method.
In ③, a model for predicting toxicity was created using cone-beam CT images of 20 cases of prostate cancer and clinical information, and accuracy verification was performed.

Academic Significance and Societal Importance of the Research Achievements

本研究によって、深層学習が画質が劣化したCT画像の改善に効果的に寄与することを実証した。放射線治療時のコーンビームCT画像は世界中で使用されており、その画質は依然として課題である。本研究成果はこの問題に対し簡易的で実用性の高い手法を提案することができた。さらに、このコーンビームCT画像から放射線治療後の障害を予測するという新しい適応放射線治療計画法を考案した。本手法が実装できることで、放射線治療時の障害発生の頻度と程度を最小限に留めることができると考えている。本研究ではその初期検討を行い、実現可能性を示すことができた。

Report

(4 results)
  • 2020 Annual Research Report   Final Research Report ( PDF )
  • 2019 Research-status Report
  • 2018 Research-status Report
  • Research Products

    (2 results)

All 2020

All Presentation (2 results) (of which Int'l Joint Research: 1 results)

  • [Presentation] Quantitative evaluation of deep convolutional neural network based denoising for ultra-low-dose CT2020

    • Author(s)
      臼井桂介
    • Organizer
      20th Asia-Oceania Congress on Medical Physics
    • Related Report
      2020 Annual Research Report
    • Int'l Joint Research
  • [Presentation] 機械学習を用いたコーンビームCTの散乱線除去2020

    • Author(s)
      村田一心
    • Organizer
      第39回日本医用画像工学会大会
    • Related Report
      2020 Annual Research Report

URL: 

Published: 2018-04-23   Modified: 2022-01-27  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi