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

Radiation dose reduction in CT by improving the image quality of ultra low dose CT images by means of deep learning

Research Project

Project/Area Number 17H06679
Research Category

Grant-in-Aid for Research Activity Start-up

Allocation TypeSingle-year Grants
Research Field Medical Physics and Radiological Technology
Research InstitutionTokyo Institute of Technology

Principal Investigator

鈴木 賢治  東京工業大学, 科学技術創成研究院, 特任教授 (00295578)

Project Period (FY) 2017-08-25 – 2019-03-31
Project Status Completed (Fiscal Year 2017)
Budget Amount *help
¥2,730,000 (Direct Cost: ¥2,100,000、Indirect Cost: ¥630,000)
Fiscal Year 2017: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Keywords深層学習 / 機械学習 / CT / 被曝低減 / 雑音除去
Outline of Annual Research Achievements

1)画像出力型深層学習をベースとするCTのための被曝線量低減手法の開発
本研究で開発したCTのための被曝線量低減手法は,我々独自の深層学習をベースとし,学習と実行ステップから成る.学習ステップでは,超低線量で撮像されたCT像を入力画像,高線量で撮像されたCT像を教師画像に使い,深層学習モデルを学習する.深層学習は,ニューラルネット回帰モデル(NN)で構成され,入力は超低線量CT像の局所領域の画素値,出力はそれに対応する高線量CT像中の1画素の推定値である.学習は,教師画素と出力画素の誤差が小さくなるよう,NNの重み係数を調整することにより行われる.実行ステップでは,学習後のモデルに未学習の超低線量CT像を入力すれば,あたかも高線量で撮ったようなCT像(仮想高線量CT像)に変換できる.
CTデータの3次元化に伴い,深層学習モデルを3次元に拡張する場合,情報量の増大に伴う学習時間の増大が問題となる.2次元モデルの学習は,通常のPCにおいて約72時間を要した.3次元化すると情報量が10倍程に増える.この問題に対処するため,我々が以前に開発したLaplacian Eigenmapに基づく非線形入力次元削減手法を応用した.
2)3次元胸部ファントムによる被曝線量低減手法の性能検証
深層学習3次元モデルの学習と検証のため,精巧な3次元胸部ファントムを,CT装置の最低線量から最高線量までの複数の線量で撮像した.超低線量と最高線量のCT画像を入力画像と教師画像とし,深層学習モデルを学習した.本手法で低減できる被曝線量を定量的に明らかにするため,最高線量CT像を理想画像とし,出力CT像の画質をSSIM (Structural Similarity)を用いて評価した.仮想高線量CT像の画質と,実際に線量を変えて撮像したCT像の関係を調べることにより,本手法で低減できる線量を定量的に明らかにした.

Research Progress Status

29年度が最終年度であるため、記入しない。

Strategy for Future Research Activity

29年度が最終年度であるため、記入しない。

Report

(1 results)
  • 2017 Annual Research Report
  • Research Products

    (26 results)

All 2018 2017 Other

All Int'l Joint Research (1 results) Journal Article (7 results) (of which Int'l Joint Research: 3 results,  Peer Reviewed: 4 results) Presentation (13 results) (of which Int'l Joint Research: 12 results,  Invited: 2 results) Book (5 results)

  • [Int'l Joint Research] Illinois Institute of Technology/University of Chicago/Harvard University(米国)

    • Related Report
      2017 Annual Research Report
  • [Journal Article] Comparing Two Classes of End-to-End Learning Machines for Lung Nodule Detection and Classification: MTANNs vs CNNs2017

    • Author(s)
      Nima Tajbakhsh and Suzuki K.
    • Journal Title

      Pattern Recognition

      Volume: 63 Pages: 476-486

    • DOI

      10.1016/j.patcog.2016.09.029

    • Related Report
      2017 Annual Research Report
    • Peer Reviewed / Int'l Joint Research
  • [Journal Article] Machine Learning in Medical Imaging Before and After Introduction of Deep Learning2017

    • Author(s)
      Suzuki K.
    • Journal Title

      Medical Imaging and Information Sciences

      Volume: 34 Issue: 2 Pages: 14-24

    • DOI

      10.11318/mii.34.14

    • NAID

      130006846726

    • ISSN
      0910-1543, 1880-4977
    • Related Report
      2017 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Overview of Deep Learning in Medical Imaging2017

    • Author(s)
      Suzuki K.
    • Journal Title

      Radiological Physics and Technology

      Volume: 10 Issue: 3 Pages: 257-273

    • DOI

      10.1007/s12194-017-0406-5

    • Related Report
      2017 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Survey of Deep Learning Applications to Medical Image Analysis2017

    • Author(s)
      Suzuki K.
    • Journal Title

      Medical Imaging Technology

      Volume: 35 Issue: 4 Pages: 212-226

    • DOI

      10.11409/mit.35.212

    • NAID

      130006108080

    • ISSN
      0288-450X, 2185-3193
    • Related Report
      2017 Annual Research Report
    • Peer Reviewed
  • [Journal Article] Special issue on Machine Learning in Medical Imaging2017

    • Author(s)
      Suzuki K., Zhou L., Wang Q.
    • Journal Title

      Pattern Recognition

      Volume: 63 Pages: 465-467

    • DOI

      10.1016/j.patcog.2016.10.020

    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Journal Article] Special issue on Machine Learning Applications in Medical Image Analysis2017

    • Author(s)
      El-Baz A., Gimel'farb G., Suzuki K.
    • Journal Title

      Computational and Mathematical Methods in Medicine

      Volume: 2017 Pages: 1-2

    • DOI

      10.1155/2017/2361061

    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Journal Article] Deep Learning Applications, Research and Development in Medical Imaging: Introduction2017

    • Author(s)
      Suzuki K.
    • Journal Title

      Medical Imaging Technology

      Volume: 35 Issue: 4 Pages: 177-179

    • DOI

      10.11409/mit.35.177

    • NAID

      130006108057

    • ISSN
      0288-450X, 2185-3193
    • Related Report
      2017 Annual Research Report
  • [Presentation] Radiation dose reduction in digital breast tomosynthesis (DBT) by means of deep-learning-based supervised image processing.2018

    • Author(s)
      Liu J., Zarshenas A., Wei Z., Yang L., Fajardo L., and Suzuki K.
    • Organizer
      SPIE Medical Imaging (SPIE MI)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Reduction in training time of a deep learning (DL) model in radiomics analysis of lesions in CT.2018

    • Author(s)
      Makkinejad N., Tajbakhsh N., Zarshenas A., Khokhar A., and Suzuki K.
    • Organizer
      SPIE Medical Imaging (SPIE MI)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Neural Network Convolution (NNC) for Converting Ultra-Low-Dose to "Virtual" High-Dose CT Images.2017

    • Author(s)
      Suzuki K., Liu J., Zarshenas A., Higaki T., Fukumoto W., and Awai K.
    • Organizer
      International Workshop on Machine Learning in Medical Imaging (MLMI)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Recent Advances in Medical Image Understanding and Diagnosis with Artificial Intelligence.2017

    • Author(s)
      Suzuki K.
    • Organizer
      Hiroshima Medical Engineering School (hBMEs)
    • Related Report
      2017 Annual Research Report
    • Invited
  • [Presentation] Deep and Shallow Machine Learning in Medical Image Analysis and Diagnosis.2017

    • Author(s)
      Suzuki K.
    • Organizer
      IEEE 5th Workshop on Data Mining in Biomedical Informatics and Health (DMBIH)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research / Invited
  • [Presentation] Investigating the Depth of Convolutional Neural Networks (CNNs) in Computer-aided Detection and Classification of Focal Lesions: Lung Nodules in Thoracic CT and Colorectal Polyps in CT Colonography.2017

    • Author(s)
      Tajbakhsh N., Zarshenas A., Liu J., and Suzuki K.
    • Organizer
      Scientific Assembly and Annual Meeting of Radiological Society of North America (RSNA)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Two Deep-Learning Models for Lung Nodule Detection and Classification in CT: Convolutional Neural Network (CNN) vs Neural Network Convolution (NNC).2017

    • Author(s)
      Tajbakhsh N., Zarshenas A., Liu J., and Suzuki K.
    • Organizer
      Scientific Assembly and Annual Meeting of Radiological Society of North America (RSNA)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Radiation Dose Reduction in Thin-Slice Chest CT at a Micro-Dose (mD) Level by Means of 3D Deep Neural Network Convolution (NNC).2017

    • Author(s)
      Zarshenas A., Zhao Y., Liu J., Higaki T., Awai K., and Suzuki K.
    • Organizer
      Scientific Assembly and Annual Meeting of Radiological Society of North America (RSNA)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Detection of Solid Pulmonary Nodules in Micro-Dose CT (mDCT) with "Virtual" Higher-Dose (vHD) CT Technology: An Observer Performance Study.2017

    • Author(s)
      Fukumoto W., Suzuki K., Higaki T., Zhao Y., Zarshenas A., and Awai K.
    • Organizer
      Scientific Assembly and Annual Meeting of Radiological Society of North America (RSNA)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Computer-Based Interactive Demonstration and Comparative Study: Virtual Full-Dose (VFD) Digital Breast Tomosynthesis (DBT) Images Derived From Reduced-Dose Acquisitions versus Clinical Full-Dose DBT Images.2017

    • Author(s)
      Liu J., Zarshenas A., Wei Z., Yang L., Fajardo L. L., and Suzuki K.
    • Organizer
      Scientific Assembly and Annual Meeting of Radiological Society of North America (RSNA)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] What Was Changed in Machine Learning (ML) in Medical Image Analysis After the Introduction of Deep Learning?2017

    • Author(s)
      Suzuki K., Zarshenas A., Liu J., Zhao Y., and Luo Y.
    • Organizer
      Scientific Assembly and Annual Meeting of Radiological Society of North America (RSNA)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Virtual High-Dose (VHD) Technology: Radiation Dose Reduction in Digital Breast Tomosynthesis (DBT) by Means of Supervised Deep-Learning Image Processing (DLIP).2017

    • Author(s)
      Liu J., Zarshenas A., Wei Z., Yang L., Fajardo L. L., and Suzuki K.
    • Organizer
      Scientific Assembly and Annual Meeting of Radiological Society of North America (RSNA)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Presentation] How Deep Should We Go with Deep Learning in Medical Image Analysis?2017

    • Author(s)
      Tajbakhsh N., Zarshenas A., Liu J., and Suzuki K.
    • Organizer
      Scientific Assembly and Annual Meeting of Radiological Society of North America (RSNA)
    • Related Report
      2017 Annual Research Report
    • Int'l Joint Research
  • [Book] Artificial Intelligence in Decision Support Systems for Diagnosis in Medical Imaging2018

    • Author(s)
      Suzuki K., Chen Y.
    • Total Pages
      387
    • Publisher
      Springer
    • ISBN
      9783319688428
    • Related Report
      2017 Annual Research Report
  • [Book] Emerging Developments and Practices in Oncology2018

    • Author(s)
      Xu J., Zarshenas A., Chen Y., and Suzuki K.
    • Total Pages
      305
    • Publisher
      IGI Global
    • ISBN
      9781522530855
    • Related Report
      2017 Annual Research Report
  • [Book] Artificial Intelligence in Decision Support Systems for Diagnosis in Medical Imaging2018

    • Author(s)
      Tajbakhsh N. and Suzuki K.
    • Total Pages
      387
    • Publisher
      Springer-Verlag
    • ISBN
      9783319688435
    • Related Report
      2017 Annual Research Report
  • [Book] Machine Learning in Medical Imaging (MLMI)2017

    • Author(s)
      Wang Q., Shi Y., Suk H., Suzuki K.
    • Total Pages
      391
    • Publisher
      Springer International Publishing
    • ISBN
      9783319673899
    • Related Report
      2017 Annual Research Report
  • [Book] Image-based Computer-assisted Radiation Therapy2017

    • Author(s)
      Suzuki K
    • Total Pages
      375
    • Publisher
      Springer-Nature
    • ISBN
      9789811029431
    • Related Report
      2017 Annual Research Report

URL: 

Published: 2017-08-25   Modified: 2022-05-23  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi