Study for performance improvement of automated lesion detection system in medical images using weakly-labeled data

• Project/Area Number: 18K12096
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 90130:Medical systems-related
• Research Institution: The University of Tokyo
• Principal Investigator: Nomura Yukihiro 東京大学, 医学部附属病院, 特任講師 (60436491)
• Project Period (FY): 2018-04-01 – 2021-03-31
• Project Status: Completed (Fiscal Year 2020)
• Budget Amount: ¥3,900,000 (Direct Cost: ¥3,000,000、Indirect Cost: ¥900,000); Fiscal Year 2020: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000); Fiscal Year 2019: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000); Fiscal Year 2018: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
• Keywords: 医用画像 / 診断支援システム / ディープラーニング / セグメンテーション / 異常検知 / 機械学習 / ラベル

Outline of Final Research Achievements

In this research, we constructed a methodology for performance improvement of an automated lesion detection system using weakly-labeled data. We constructed two types of lesion shape label estimation methods using the location of a lesion center and the measured size. In addition, we prepared to implement the constructed lesion shape estimation method in the web-based image database (CIRCUS DB). We also constructed automated lesion detection systems based on anomaly detection trained using data where only the existence of lesions is known.

Academic Significance and Societal Importance of the Research Achievements

本研究の成果として得られた、病変ラベル推定方法および病変の存在のみが既知な症例を用いた異常検知手法を用いることで、症例データ収集における医師の負担が軽減され、かつ臨床現場に多数ある弱ラベル付症例が利用可能となる。このため、病変自動検出システムの研究のさらなる推進が図れると考える。病変自動検出システムの研究が進めば多くの高性能なシステムが臨床現場で使用されるようになり、臨床画像診断の質的向上に寄与すると考える。

Research Products

• [Journal Article] Unsupervised Deep Anomaly Detection in Chest Radiographs (2021)
  • Author(s): Nakao Takahiro、Hanaoka Shouhei、Nomura Yukihiro、Murata Masaki、Takenaga Tomomi、Miki Soichiro、Watadani Takeyuki、Yoshikawa Takeharu、Hayashi Naoto、Abe Osamu
  • Journal Title: Journal of Digital Imaging Volume: n/a Issue: 2 Pages: 418-427
  • DOI: 10.1007/s10278-020-00413-2
  • Peer Reviewed / Open Access
• [Journal Article] Development of training environment for deep learning with medical images on supercomputer system based on asynchronous parallel Bayesian optimization (2020)
  • Author(s): Nomura Y, Sato I, Hanawa T, Hanaoka S, Nakao T, Takenaga T, Hoshino T, Sekiya Y, Miki S, Yoshikawa T, Hayashi N, Abe O
  • Journal Title: Journal of Supercomputing Volume: － Issue: 9 Pages: 7315-7332
  • DOI: 10.1007/s11227-020-03164-7
  • Peer Reviewed
• [Journal Article] Novel platform for development, training, and validation of computer-assisted detection/diagnosis software (2020)
  • Author(s): Nomura Y, Miki S, Hayashi N, Hanaoka S, Sato I, Yoshikawa T, Masutani Y, Abe O
  • Journal Title: Int J Comput Assist Radiol Surg Volume: 15 Issue: 4 Pages: 661-672
  • DOI: 10.1007/s11548-020-02132-z
  • Peer Reviewed / Open Access
• [Journal Article] Can the spherical gold standards be used as an alternative to painted gold standards for the computerized detection of lesions using voxel-based classification? (2018)
  • Author(s): Nomura Y, Hayashi N, Hanaoka S, Takenaga T, Nemoto M, Miki S, Yoshikawa T, Abe O
  • Journal Title: Japanese Journal of Radiology Volume: 37 Issue: 3 Pages: 264-273
  • DOI: 10.1007/s11604-018-0784-6
  • NAID: 210000187082
  • Peer Reviewed
• [Presentation] ディープラーニングに基づく3次元汎用半自動病変セグメンテーションの初期検討 (2020)
  • Author(s): 野村行弘, 花岡昇平, 竹永智美, 中尾貴祐, 柴田寿一, 三木聡一郎, 吉川健啓, 渡谷岳行, 林直人, 阿部修
  • Organizer: 第39回日本医用画像工学会大会
• [Presentation] Deep generative model-based unsupervised detectionof inappropriate images in a chest X-ray dataset (2019)
  • Author(s): Nakao T, Hanaoka S, Nomura Y, Murata M, Takenaga T, Miki S, Watadani T, Yoshikawa T, Hayashi N, Abe O
  • Organizer: CARS 2019
  • Int'l Joint Research
• [Presentation] 新たなwebベースの統合的CAD開発プラットフォームの構築 (2019)
  • Author(s): 野村行弘、三木聡一郎、林直人、花岡昇平、吉川健啓、増谷佳孝、阿部修
  • Organizer: 電子情報通信学会医用画像研究会
• [Presentation] Preliminary study of automated pulmonary mass detection in chest radiography using U-Net (2019)
  • Author(s): Nomura Y, Hayashi N, Hanaoka S, Yoshikawa T, Murata M, Nakao T, Takenaga T, Miki S, Watadani T, Abe O
  • Organizer: 第78回日本医学放射線学会総会
• [Presentation] Preliminary study of automated detection of pulmonary nodule in ultrashort echo time MR images (2019)
  • Author(s): Nomura Y, Hanaoka S, Yoshikawa T, Sato I, Nakao T, Murata M, Takenaga T, Koshino S, Miki S, Watadani T, Hayashi N, Abe O
  • Organizer: CARS 2019
  • Int'l Joint Research
• [Presentation] MRIにおける脳転移検出に対するConvolutional Neural Network の応用：線量分布を正解データとして利用 (2018)
  • Author(s): 村田仁樹、花岡昇平、野村行弘、竹永智美、中尾貴祐、高橋渉、名和要武、吉川健啓、林直人、阿部修
  • Organizer: 第37回日本医用画像工学会大会