2020 Fiscal Year Final Research Report

Development of Automated Cellular Video Image Processing Technology Using Reinforcement Learning via Simulation and GAN

Research Project

PDF

Project/Area Number	18K19842
Research Category	Grant-in-Aid for Challenging Research (Exploratory)
Allocation Type	Multi-year Fund
Review Section	Medium-sized Section 62:Applied informatics and related fields
Research Institution	Osaka University
Principal Investigator	Seno Shigeto 大阪大学, 情報科学研究科, 准教授 (30432462)
Co-Investigator(Kenkyū-buntansha)	間下以大大阪大学, サイバーメディアセンター, 准教授 (00467606)
Project Period (FY)	2018-06-29 – 2021-03-31
Keywords	バイオイメージインフォマティクス / 強化学習 / 深層学習
Outline of Final Research Achievements	With the recent development of microscopy technology, images and movies have been produced every day. The methods and purposes of bio-imaging are diverse, but in the case of analyzing images of moving cells, cell recognition and spatio-temporal tracking are universal tasks for extracting information from the images. Traditionally, these tasks have been handled by human tracking or by algorithms that have been developed respectively. In this study, we developed a method that uses deep learning and reinforcement learning methods to simulate a moving image from the correct solution and use the data to learn the task of cell tracking.
Free Research Field	バイオインフォマティクス
Academic Significance and Societal Importance of the Research Achievements	細胞の移動軌跡のシミュレーションと敵対的生成ネットワーク(GAN) を用いて、少数の正解付き訓練データから類似した疑似データを自動生成する方法の開発を行った。本研究では細胞動画像を対象とした研究を行ったが，シミュレーションとGANによる仮想動画の生成と、それを介した強化学習によって必要なタスクを獲得することができれば、様々な分野での応用が期待できる。