High-resolution 3D analysis of nanostructures using super-resolution SEM images by deep learning

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K15139
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 28030:Nanomaterials-related
• Research Institution: Japan Fine Ceramics Center
• Principal Investigator: Yoshida Ryuji 一般財団法人ファインセラミックスセンター, その他部局等, 上級技師 (50595725)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: SEM / 深層学習 / SRGAN

Outline of Final Research Achievements

Super-resolution of SEM images was investigated using SRGAN, a type of deep learning. There were differences in the results obtained depending on the number of teacher images, the number of epochs, and the type and resolution of the teacher images, and we were able to clarify the advantages and disadvantages of applying deep learning to SEM images.

Free Research Field

電子顕微鏡

Academic Significance and Societal Importance of the Research Achievements

走査型電子顕微鏡(SEM)画像への深層学習を用いた超解像技術の適用事例の報告は少なく、本研究で得られた成果は今後のSEMおよび透過型電子顕微鏡(TEM)画像を含めた、顕微鏡画像全般に対する超解像技術の適用検討の一助となる。