Statistical models of organs with diverse shape and size, and its application to organ recognition

• Project/Area Number: 18H03255
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Single-year Grants
• Section: 一般
• Review Section: Basic Section 61010:Perceptual information processing-related
• Research Institution: Tokyo University of Agriculture and Technology
• Principal Investigator: Shimizu Akinobu 東京農工大学, 工学(系)研究科(研究院), 教授 (80262880)
• Co-Investigator(Kenkyū-buntansha): 斉藤 篤 東京農工大学, 工学(系)研究科(研究院), 助教 (10781445)
• Project Period (FY): 2018-04-01 – 2021-03-31
• Project Status: Completed (Fiscal Year 2020)
• Budget Amount: ¥17,160,000 (Direct Cost: ¥13,200,000、Indirect Cost: ¥3,960,000); Fiscal Year 2020: ¥5,850,000 (Direct Cost: ¥4,500,000、Indirect Cost: ¥1,350,000); Fiscal Year 2019: ¥5,720,000 (Direct Cost: ¥4,400,000、Indirect Cost: ¥1,320,000); Fiscal Year 2018: ¥5,590,000 (Direct Cost: ¥4,300,000、Indirect Cost: ¥1,290,000)
• Keywords: 計測工学 / 解剖学 / 画像処理 / 情報工学 / 統計数理

Outline of Final Research Achievements

Most of the cases imaged at hospitals are abnormal cases whose organ’s shape and size are deviated from the mean organ. Conventional statistical models are not suitable to describe the organs whose shape and size are deviate from the mean organ. This study developed statistical models of organs with diverse shape and size. We also developed organ recognition algorithm by use of the statistical models.

Academic Significance and Societal Importance of the Research Achievements

本研究の学術的意義は，これまで正確に表現できなかった，分布の周辺にある形状やサイズの臓器を正確にモデル化できるようになった点である．また，そのモデルを用いた臓器の認識処理についてもこれまで例がなく，新規性が高い．社会的な意義としては，実際の病院で撮影されている異常な形状やサイズの症例の画像を対象とした，診断支援システムの精度向上が期待される点である．日々撮影されている膨大な数の医用画像の大部分を占める異常例に対する診断精度向上は，大きな社会的な意義がある．

Research Products

• [Int'l Joint Research] ESIEE(フランス)
• [Int'l Joint Research] Children's National Health System(米国)
• [Journal Article] Level set distribution model of nested structures using logarithmic transformation (2019)
  • Author(s): Saito A, Tsujikawa M, Takakuwa T, Yamada S, Shimizu A
  • Journal Title: Medical Image Analysis Volume: 56 Pages: 1-10
  • DOI: 10.1016/j.media.2019.05.003
  • Peer Reviewed
• [Journal Article] Liver segmentation from low-radiation-dose pediatric computed tomography using patient-specific, statistical modeling (2019)
  • Author(s): Koyo Nakayama, Atsushi Saito, Elijah Biggs, Marius George Linguraru, Akinobu Shimizu
  • Journal Title: International Journal of Computer Assisted Radiology and Surgery Volume: 印刷中 Issue: 12 Pages: 2057-2068
  • DOI: 10.1007/s11548-019-01929-x
  • Peer Reviewed / Int'l Joint Research
• [Presentation] A statistical intensity model of blood vessels in a CT volume using a flow based auto-encoder (2021)
  • Author(s): Ohzora Masuko
  • Organizer: The International Forum on Medical Imaging in Asia
  • Int'l Joint Research
• [Presentation] Liver segmentation from pediatric CT volumes using fully convolutional network with shape regularization by conditional statistical shape model (2020)
  • Author(s): Yumiko Tanaka
  • Organizer: Computer Assisted Radiology and Surgery
  • Int'l Joint Research
• [Presentation] Simultaneous process of skeleton segmentation and hot-spot extraction in a bone scintigram (2020)
  • Author(s): Morio Hara
  • Organizer: Computer Assisted Radiology and Surgery
  • Int'l Joint Research
• [Presentation] Purkinje cell somas segmentation in a two-photon microscopic volume of a mouse brain (2020)
  • Author(s): Ryo Kimizuka
  • Organizer: Computer Assisted Radiology and Surgery
  • Int'l Joint Research
• [Presentation] Statistical intensity model of lung vessels in a CT volume using beta-VAE (2019)
  • Author(s): Y. Saeki, A. Saito, J. Ueno, M. Harada, A. Shimizu
  • Organizer: Computer Assisted Radiology and Surgery
  • Int'l Joint Research
• [Presentation] Construction and Evaluation of a Statistical Shape Model of Liver Using Variational Autoencoder (2019)
  • Author(s): Zhihui Lu, Atsushi Saito, Shigeru Nawano, Akinobu Shimizu
  • Organizer: The IEEE International Symposium on Biomedical Imaging
  • Int'l Joint Research