Research on composite deep learning network for AI-based clinical image diagnosis

• Project/Area Number: 17K10385
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Radiation science
• Research Institution: National Institutes for Quantum Science and Technology
• Principal Investigator: Tachibana Yasuhiko 国立研究開発法人量子科学技術研究開発機構, 量子医科学研究所 分子イメージング診断治療研究部, 主幹研究員 (20749973)
• Co-Investigator(Kenkyū-buntansha): 相田 典子 地方独立行政法人神奈川県立病院機構神奈川県立こども医療センター(臨床研究所), 臨床研究所, 部長 (20586292)
• Project Period (FY): 2017-04-01 – 2023-03-31
• Project Status: Completed (Fiscal Year 2022)
• Budget Amount: ¥4,680,000 (Direct Cost: ¥3,600,000、Indirect Cost: ¥1,080,000); Fiscal Year 2019: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000); Fiscal Year 2018: ¥780,000 (Direct Cost: ¥600,000、Indirect Cost: ¥180,000); Fiscal Year 2017: ¥3,120,000 (Direct Cost: ¥2,400,000、Indirect Cost: ¥720,000)
• Keywords: 画像診断 / 人工知能 / 深層学習 / 機械学習 / ディープラーニング / AI / 分類 / 画像処理 / 可視化 / 分類問題 / バイアス / アノテーション / 画像変換問題 / データバイアス / MRI / 脳 / ニューラルネットワーク / 分類学習 / 放射線 / 画像

Outline of Final Research Achievements

As an achievement directly related to the submitted research proposal, we developed an original method for visualizing the basis of judgment in AI diagnosis (image classification task), and published the results at domestic and international conferences and in scientific journals.
In addition, we have launched several new research projects based on the findings obtained through this project, and are continuing to promote these projects and publishing the achievements.

Academic Significance and Societal Importance of the Research Achievements

Iの説明性を高めるために用いられる一般的な手法では画像上の位置など、画像から得られる医学的情報とは無関係な情報のみを対象とするが、提案手法では解剖学など医学的な情報を加味できるメリットが今後役立つ可能性がある。つまり、画像診断についてAIが医師に提案をするとき、医師がその提案をどのように解釈すべきか、信用できるかどうかを判断するために、より具体的な医学的根拠を利用することができる可能性がある。

Research Products

• [Journal Article] Dual-Energy Computed Tomography-Based Iodine Concentration Estimation for Evaluating Choroidal Malignant Melanoma Response to Treatment: Optimization and Primary Validation (2022)
  • Author(s): Tsuchiya Hiroki, Tachibana Yasuhiko, Kishimoto Riwa, Omatsu Tokuhiko, Hotta Eika, Tanimoto Katsuyuki, Wakatsuki Masaru, Obata Takayuki, Tsuji Hiroshi
  • Journal Title: Diagnostics Volume: 12 Issue: 11 Pages: 2692-2692
  • DOI: 10.3390/diagnostics12112692
  • Peer Reviewed / Open Access
• [Journal Article] Super-resolution generative adversarial networks with static T2*WI-based subject-specific learning to improve spatial difference sensitivity in fMRI activation (2022)
  • Author(s): Ota J, Umehara K, Kershaw J, Kishimoto R, Hirano Y, Tachibana Y, Ohba H, Obata T
  • Journal Title: Scientific Reports Volume: 12 Issue: 1 Pages: 10319-10319
  • DOI: 10.1038/s41598-022-14421-5
  • Peer Reviewed / Open Access
• [Journal Article] Exploring cell membrane water exchange in aquaporin-4-deficient ischemic mouse brain using diffusion-weighted MRI. (2021)
  • Author(s): Urushihata Takuya, Takuwa Hiroyuki, Manami Takahashi, Tachibana Yasuhiko, Jeffrey Kershaw, Nitta Nobuhiro, Shibata Sayaka, Masato Yasui, Higuchi Makoto, Obata Takayuki
  • Journal Title: European radiology experimental Volume: 44 Issue: 1 Pages: 44-44
  • DOI: 10.1186/s41747-021-00244-y
  • Peer Reviewed
• [Journal Article] Monte Carlo Simulator for Diffusion-weighted Imaging Sequences (2021)
  • Author(s): Yasuhiko Tachibana, Tanguy Duval, Takayuki Obata
  • Journal Title: Magnetic Resonance in Medical Sciences Volume: 20 Issue: 2 Pages: 222-226
  • DOI: 10.2463/mrms.bc.2020-0013
  • NAID: 130008046941
  • ISSN: 1347-3182, 1880-2206
  • Peer Reviewed / Open Access / Int'l Joint Research
• [Journal Article] The Utility of a Convolutional Neural Network for Generating a Myelin Volume Index Map from Rapid Simultaneous Relaxometry Imaging (2020)
  • Author(s): Yasuhiko Tachibana, Akifumi Hagiwara, Masaaki Hori, Jeffrey Kershaw, Misaki Nakazawa, Tokuhiko Omatsu, Riwa Kishimoto, Kazumasa Yokoyama, Nobutaka Hattori, Shigeki Aoki, Tatsuya Higashi, Takayuki Obata
  • Journal Title: Magnetic Resonance in Medical Sciences Volume: 19 Issue: 4 Pages: 324-332
  • DOI: 10.2463/mrms.mp.2019-0075
  • NAID: 130007948910
  • ISSN: 1347-3182, 1880-2206
  • Peer Reviewed / Open Access
• [Journal Article] New algorithm using L1 regularization for measuring electron energy spectra (2020)
  • Author(s): Hironao Sakaki, Tomohiro Yamashita, Akagi, Mamiko Nishiuchi, NicholasPeter Dover, HazelFrances Lowe, Kotaro Kondo, Akira Kon, Masaki Kando, Yasuhiko Tachibana, Takayuki Obata, Keiichiro Shiokawa, Tatsuhiko Miyatake, Yukinobu Watanabe
  • Journal Title: Review of Scientific Instruments Volume: 91 Issue: 7
  • DOI: 10.1063/1.5144897
  • Peer Reviewed
• [Journal Article] Multiple sclerosis lesions in motor tracts from brain to cervical cord: spatial distribution and correlation with disability (2020)
  • Author(s): Anne Kerbrat, Charley Gros, Atef Badji, Elise Bannier, Francesca Galassi, Benoit Combes, Raphael Chouteau, Pierre Labauge, Xavier Ayrignac, Yasuhiko Tachibana, Others
  • Journal Title: Brain Volume: 143 Issue: 7 Pages: 2089-2105
  • DOI: 10.1093/brain/awaa162
  • Peer Reviewed / Int'l Joint Research
• [Journal Article] A neural network model that learns differences in diagnosis strategies among radiologists has an improved area under the curve for aneurysm status classification in magnetic resonance angiography image series (2020)
  • Author(s): Yasuhiko Tachibana, Masataka Nishimori, Naoyuki Kitamura, Kensuke Umehara, Junko Ota, Takayuki Obata, Tatsuya Higashi
  • Journal Title: arXiv Volume: epub
  • Open Access
• [Journal Article] The Utility of Applying Various Image Preprocessing Strategies to Reduce the Ambiguity in Deep Learning-based Clinical Image Diagnosis (2020)
  • Author(s): Tachibana Yasuhiko、Obata Takayuki、Kershaw Jeff、Sakaki Hironao、Urushihata Takuya、Omatsu Tokuhiko、Kishimoto Riwa、Higashi Tatsuya
  • Journal Title: Magnetic Resonance in Medical Sciences Volume: 19 Issue: 2 Pages: 92-98
  • DOI: 10.2463/mrms.mp.2019-0021
  • NAID: 130007839525
  • ISSN: 1347-3182, 1880-2206
  • Peer Reviewed / Open Access
• [Journal Article] Improving the Quality of Synthetic FLAIR Images with Deep Learning Using a Conditional Generative Adversarial Network for Pixel-by-Pixel Image Translation (2019)
  • Author(s): Hagiwara A.、Otsuka Y.、Hori M.、Tachibana Y.、Yokoyama K.、Fujita S.、Andica C.、Kamagata K.、Irie R.、Koshino S.、Maekawa T.、Chougar L.、Wada A.、Takemura M.Y.、Hattori N.、Aoki S.
  • Journal Title: American Journal of Neuroradiology Volume: 40 Issue: 2 Pages: 224-230
  • DOI: 10.3174/ajnr.a5927
  • Peer Reviewed / Open Access
• [Journal Article] The utility of a convolutional neural network for generating a myelin volume index map from rapid simultaneous relaxometry imaging (2019)
  • Author(s): Yasuhiko Tachibana, Akifumi Hagiwara, Masaaki Hori, Jeff Kershaw, Misaki Nakazawa, Tokuhiko Omatsu, Riwa Kishimoto, Kazumasa Yokoyama, Nobutaka Hattori, Shigeki Aoki, Tatsuya Higashi, Takayuki Obata
  • Journal Title: arXiv Volume: arXiv:1904.10960
  • NAID: 130007948910
  • Open Access
• [Presentation] MRI画像診断の実態とAI応用 (2023)
  • Author(s): 立花 泰彦
  • Organizer: 総理工情報科学セミナー
  • Invited
• [Presentation] 放射線照射腫瘍モデルマウスの低酸素領域の描出におけるMRI の有用性の検討 (2022)
  • Author(s): 今泉 晶子, 平山 亮一, 長谷川 純崇, Ｋｅｒｓｈａｗ Ｊｅｆｆｒｅｙ, 立花 泰彦, 柴田 さやか, 新田 展大, 鵜澤 玲子, 東 達也, 小畠 隆行
  • Organizer: 第50回日本磁気共鳴医学会大会
• [Presentation] 乳腺ダイナミック造影MRIにおける血行動態の簡易指標マッピングと汎用ソフトウェア開発 (2022)
  • Author(s): 生駒 洋子, 尾松 徳彦, 岸本 理和, 立花 泰彦, 小畠 隆行
  • Organizer: 第50回日本磁気共鳴医学会大会
• [Presentation] The usefulness of 4D convolution in deep-learning-based noise reduction for low-SNR body DWI (2022)
  • Author(s): Tachibana Yasuhiko, Tsuchiya Hiroki, Kishimoto Riwa, Omatsu Tokuhiko, Mori Shinichiro, Obata Takayuki, Higashi Tatsuya
  • Organizer: ISMRM annual meeting 2022
• [Presentation] 画像解析のすすめ 臨床応用・ピットフォール・気を付けるべき点 (2022)
  • Author(s): 立花 泰彦
  • Organizer: 第50回日本磁気共鳴医学会大会
  • Invited
• [Presentation] The usefulness of 4D convolution in deep learning-based noise reduction for low-SNR body DWI (2022)
  • Author(s): Yasuhiko Tachibana, Hiroki Tsuchiya, Riwa Kishimoto, Tokuhiko Omatsu, Shinichiro Mori, Takayuki Obata, and Tatsuya Higashi
  • Organizer: ISMRM 2022
  • Int'l Joint Research
• [Presentation] Exploring the effect of aquaporin-4 suppression on cell membrane water exchange in a mouse model of brain ischemia using diffusion-weighted MRI (2021)
  • Author(s): Urushihata Takuya, Takuwa Hiroyuki, Manami Takahashi, Tachibana Yasuhiko, Jeffrey Kershaw, Nitta Nobuhiro, Shibata Sayaka, Masato Yasui, Higuchi Makoto, Obata Takayuki
  • Organizer: ISMRM 2021
  • Int'l Joint Research
• [Presentation] 拡散強調MRIによる腫瘍モデルマウスの低酸素領域の検出 (2021)
  • Author(s): 今泉 晶子, 小畠 隆行, Ｋｅｒｓｈａｗ Ｊｅｆｆｒｅｙ, 立花 泰彦, 柴田 さやか, 新田 展大, 長谷川 純崇, 東 達也
  • Organizer: 第49回日本磁気共鳴医学会大会
• [Presentation] instabilityを除いた深層学習によるノイズ低減法の評価 (2021)
  • Author(s): Hayato Nozaki, Yasuhiko Tachibana, Yujiro Otsuka, Wataru Uchida, Yuya Saito, Koji Kamagata
  • Organizer: 第49回日本磁気共鳴医学会大会
• [Presentation] Deep learning-based DWI Denoising method that suppressed the "instability" problem (2021)
  • Author(s): Hayato Nozaki, Tachibana Yasuhiko, Yujiro Otsuka, Wataru Uchida, Yuya Saito, Koji Kamagata, Shigeki Aoki
  • Organizer: ISMRM 2021
  • Invited
• [Presentation] Image Classification and Label Inhomogeneity (2021)
  • Author(s): Yasuhiko Tachibana
  • Organizer: ISMRM JPC 2021
  • Int'l Joint Research / Invited
• [Presentation] Deep Learning in Diagnostic Radiology (2021)
  • Author(s): Yasuhiko Tachibana
  • Organizer: Summer School for Molecular Biomedical Science and Engineering, Hokkaido University
  • Int'l Joint Research / Invited
• [Presentation] The neural network model that can consider the inhomogeneity of the judgements between different annotators: implementation for MRA diagnosis (2020)
  • Author(s): Yasuhiko Tachibana, Masataka Nishimori, Naoyuki Kitamura, Kensuke Umehara, Junko Ota, Takayuki Obata, and Tatsuya Higashi
  • Organizer: ISMRM 2020
  • Int'l Joint Research
• [Presentation] Deep-Brain: A Cutting-edge Concept for Outstanding Functional Resolution in fMRI (2020)
  • Author(s): Junko Ota, Kensuke Umehara, Yasuhiko Tachibana, Yoshiyuki Hirano, Hisateru Ohba, Takayuki Obata, Tatsuya Higashi
  • Organizer: RSNA2020
  • Int'l Joint Research
• [Presentation] 医用画像＋ディープラーニングな臨床研究をデザインするために知っておきたいこと (2020)
  • Author(s): 立花 泰彦
  • Organizer: 第48回日本磁気共鳴医学会大会
  • Invited
• [Presentation] Radiological application of Deep learning: what the radiologists need to know ? (2019)
  • Author(s): Yasuhiko Tachibana
  • Organizer: Neuroimaging summit 2019
  • Int'l Joint Research / Invited
• [Presentation] How to design an imaging research using deep learning? A beginner's guide (2019)
  • Author(s): Yasuhiko Tachibana
  • Organizer: NeuroImaging Summit 2019
  • Int'l Joint Research / Invited
• [Presentation] Deep learning in diagnostic imaging (2019)
  • Author(s): Yasuhiko Tachibana
  • Organizer: MRI アライアンス 第4回国際シンポジウム
  • Int'l Joint Research / Invited
• [Presentation] 医用画像へのディープラーニング応用 (2019)
  • Author(s): 立花 泰彦
  • Organizer: 第30回CCSEワークショップ
  • Invited
• [Presentation] A convolutional neural network-based approach for generating PET/CT image series in shorter scan time: A feasibility study (2019)
  • Author(s): Ritu Bhusal Chhatkuli, Yasuhiko Tachibana, Takayuki Obata, Yuto Kamitaka, Ryuichi Nishii, Tatsuya Higashi
  • Organizer: 第117回日本医学物理学会学術大会
  • Int'l Joint Research
• [Presentation] ディープラーニングを用いた課題時fMRIの高精度化に向けた初期的検討 (2019)
  • Author(s): 大田 淳子, 立花 泰彦, 平野 好幸, 梅原 健輔, 影山 肇, 生駒 洋子, 大場 久照, 小畠 隆行
  • Organizer: 第47回日本放射線技術学会秋季学術大会
• [Presentation] The utility of a convolutional neural network for generating a myelin volume index map from rapid simultaneous relaxometry imaging (2019)
  • Author(s): Yasuhiko Tachibana, Akifumi Hagiwara, Masaaki Hori, Jeffrey Kershaw, Misaki Nakazawa, Tokuhiko Omatsu, Riwa Kishimoto, Shigeki Aoki, Tatsuya Higashi, Takayuki Obata
  • Organizer: ISMRM2019
  • Int'l Joint Research
• [Presentation] Improving the Quality of Synthetic FLAIR Images with Deep Learning Using a Conditional Generative Adversarial Network for Pixel-by-Pixel Image Translation (2019)
  • Author(s): Akifumi Hagiwara, Yasuhiko Tachibana, Shigeki Aoki
  • Organizer: ISMRM2019
  • Int'l Joint Research
• [Presentation] Spatial distribution of multiple sclerosis lesions along the brain and spinal motor tracts and correlation with functional deficits (2019)
  • Author(s): Anne Kerbrat, Yasuhiko Tachibana, Julien Cohen-Adad
  • Organizer: ISMRM2019
  • Int'l Joint Research
• [Presentation] T2*WIを代替訓練データとして安静時fMRIを深層学習により高解像度化する手法の初期検討 (2019)
  • Author(s): 影山 肇, 立花 泰彦, 大田 淳子, 梅原 健輔, 平野 好幸, 小畠 隆行, 近藤 啓介, 志村 一男
  • Organizer: 第47回日本磁気共鳴医学会大会
• [Presentation] The utility of a convolutional neural network for generating a myelin volume index map from rapid simultaneous relaxometry imaging (2019)
  • Author(s): Yasuhiko Tachibana, Akifumi Hagiwara, Masaaki Hori, Jeff Kershaw, Misaki Nakazawa, Tokuhiko Omatsu, Riwa Kishimoto, Kazumasa Yokoyama, Nobutaka Hattori, Shigeki Aoki, Tatsuya Higashi, Takayuki Obata
  • Organizer: ISMRM2019
  • Int'l Joint Research
• [Presentation] 医用画像へのディープラーニング応用 (2019)
  • Author(s): 立花泰彦
  • Organizer: 第30回JAEA CCSEワークショップ
  • Invited
• [Presentation] 拡散強調像とDeep Learning (2018)
  • Author(s): 立花泰彦
  • Organizer: 第46回日本磁気共鳴医学会大会
  • Invited
• [Presentation] 医用画像に深層学習をどう使うか (2018)
  • Author(s): 立花泰彦
  • Organizer: 医用放射線技術研究会
  • Invited
• [Presentation] これでわかるはず、画像のためのディープラーニング入門 (2018)
  • Author(s): 立花泰彦
  • Organizer: 第1回日本医用画像人工知能研究会
  • Invited
• [Presentation] Synthetic MRIを利用したミエリンマップ推定における畳み込みニューラルネットワークの有用性 (2018)
  • Author(s): 立花泰彦, 萩原彰文, 堀正明, 中澤美咲, 尾松徳彦, 岸本理和, 青木茂樹, 東達也, 小畠隆行
  • Organizer: 第46回日本磁気共鳴医学会大会
• [Presentation] Deep learning を利用した画像分類において学習精度と判断過程の可視化を両立する手法の検討 (2017)
  • Author(s): 立花泰彦、小畠隆行、漆畑拓弥、尾松徳彦、岸本利和、東達也
  • Organizer: 第45回日本磁気共鳴医学会大会
• [Presentation] 医療画像を深層ニューラルネットワークに用いるために有用な前処置ソフトウェアの開発 (2017)
  • Author(s): 立花泰彦、小畠隆行、漆畑拓弥、生駒洋子、尾松徳彦、岸本利和、東達也
  • Organizer: 第36回日本医用画像工学会大会
• [Presentation] Deep-Neural-Network based image diagnosis: comparing various image preprocessing (2017)
  • Author(s): Y. Tachibana, T. Obata, J. Kershaw, Y. Ikoma, R. Kishimoto, T. Omatsu, T. Higashi
  • Organizer: ISMRM 2016 2016 Annual Meeting
  • Int'l Joint Research
• [Book] 脊髄脊椎ジャーナル Vol.35 No.10 (2023)
  • Author(s): 立花泰彦
  • Total Pages: 8
  • Publisher: 三輪書店
• [Book] INNERVISION 2017年9月号 (2017)
  • Author(s): 立花泰彦
  • Total Pages: 3
  • Publisher: 株式会社インナービジョン
• [Patent(Industrial Property Rights)] 情報処理装置、情報処理方法、情報処理プログラム、及び、記録媒体 (2023)
  • Inventor(s): 立花泰彦、橋本綾斗、岸本理和、森慎一郎
  • Industrial Property Rights Holder: 量子科学技術研究開発機構
  • Industrial Property Rights Type: 実用新案
  • Filing Date: 2023
• [Patent(Industrial Property Rights)] 画像診断装置、画像診断方法、画像診断プログラム、及び、記録媒体 (2022)
  • Inventor(s): 立花 泰彦
  • Industrial Property Rights Holder: 量子科学技術研究開発機構
  • Industrial Property Rights Type: 特許
  • Filing Date: 2022
• [Patent(Industrial Property Rights)] ノイズ除去装置、ノイズ除去方法、機械学習装置 、機械学習方法及びプログラム (2021)
  • Inventor(s): 立花 泰彦, 大塚 裕次朗, 鎌形 康司
  • Industrial Property Rights Holder: 量子科学技術研究開発機構、順天堂大学
  • Industrial Property Rights Type: 特許
  • Filing Date: 2021
• [Patent(Industrial Property Rights)] ノイズ除去装置、ノイズ除去方法、機械学習装置、機械学習方法及びプログラム (2021)
  • Inventor(s): 立花泰彦、大塚裕次朗
  • Industrial Property Rights Holder: 量子科学技術研究開発機構、順天堂大学
  • Industrial Property Rights Type: 特許
  • Filing Date: 2021