Title:
IMAGE PROCESSING METHOD, MEDICINE SENSITIVITY TEST METHOD AND IMAGE PROCESSING DEVICE
Document Type and Number:
WIPO Patent Application WO/2019/230447
Kind Code:
A1
Abstract:
In order to enable the position and number of cells included in an image to be automatically measured, even for an image in which a marker is not used, this image processing method inputs (step S102) a test image to a first learning model, inputs (step S104) an output image therefrom into a second learning model, and outputs (step S105) an output image therefrom as a result image in which the position of a portion to be detected is indicated by a representative point. Here, the first learning model is constructed by deep learning using teacher data in which a first image having marker expression is associated with a second image not having marker expression, the first image and the second image being obtained by imaging the same cells. In addition, the second learning model is constructed by deep learning using teacher data in which a third image having marker expression is associated with information indicating the position of the representative point included in the third image.
Inventors:
MIZUKAMI TAMIO (JP)
KISHIMOTO KATSUMI (JP)
KISHIMOTO KATSUMI (JP)
Application Number:
PCT/JP2019/019666
Publication Date:
December 05, 2019
Filing Date:
May 17, 2019
Export Citation:
Assignee:
FRONTIER PHARMA INC (JP)
International Classes:
C12Q1/06; C12M1/00; G01N33/48; G06T7/00; G16H30/40
Domestic Patent References:
| WO2017027380A1 | 2017-02-16 |
Foreign References:
| JP2017085966A | 2017-05-25 | |||
| JP2017519985A | 2017-07-20 | |||
| JP2017045341A | 2017-03-02 | |||
| JP2015521037A | 2015-07-27 |
Other References:
NIIOKA, H. ET AL.: "Classification of C2C12 cells at differentiation by convolutional neural network of deep learning using phase contrast images", HUMAN CELL, vol. 31, no. 1, 13 December 2017 (2017-12-13), pages 87 - 93, XP036389083, DOI: 10.1007/s13577-017-0191-9
TAKAHASHI, SORA ET AL.: "3P-0777] Deep learning- aided label-free cell counting technology: prediction of fluorescent labels from unlabeled cell images", THE 41ST ANNUAL CONFERENCE OF THE MOLECULAR BIOLOGY SOCIETY OF JAPAN, 9 November 2018 (2018-11-09), XP009524212
MIYAKI, AKIRA: "P7. Development of live/death identification and counting technique for label-free and noninvasive cells by deep learning", THE 77TH ANNUAL MEETING OF THE JAPANESE CANCER ASSOCIATION, 27 September 2018 (2018-09-27), XP009524210
ONO, KOSUKE: "P8. Live/death identification and counting technique developed by deep learning for label-free and noninvasive cells, and its application to hepatocyte toxicity assessment", THE 77TH ANNUAL MEETING OF THE JAPANESE CANCER ASSOCIATION, 27 September 2018 (2018-09-27), XP009524213
PHILLIP ISOLA ET AL.: "Image-to-image Translation with Conditional Adversarial Networks", CVPR, 21 November 2016 (2016-11-21), Retrieved from the Internet
TAKAHASHI, SORA ET AL.: "3P-0777] Deep learning- aided label-free cell counting technology: prediction of fluorescent labels from unlabeled cell images", THE 41ST ANNUAL CONFERENCE OF THE MOLECULAR BIOLOGY SOCIETY OF JAPAN, 9 November 2018 (2018-11-09), XP009524212
MIYAKI, AKIRA: "P7. Development of live/death identification and counting technique for label-free and noninvasive cells by deep learning", THE 77TH ANNUAL MEETING OF THE JAPANESE CANCER ASSOCIATION, 27 September 2018 (2018-09-27), XP009524210
ONO, KOSUKE: "P8. Live/death identification and counting technique developed by deep learning for label-free and noninvasive cells, and its application to hepatocyte toxicity assessment", THE 77TH ANNUAL MEETING OF THE JAPANESE CANCER ASSOCIATION, 27 September 2018 (2018-09-27), XP009524213
PHILLIP ISOLA ET AL.: "Image-to-image Translation with Conditional Adversarial Networks", CVPR, 21 November 2016 (2016-11-21), Retrieved from the Internet
Attorney, Agent or Firm:
FURIKADO, Shoichi et al. (JP)
Download PDF: