A Comparison of Image Analysis Tools for Segmentation on SEM Micrographs - Zeiss ZEN Intellesis vs. Thermofisher AVIZO

0575302 - ÚPT 2024 GB eng A - Abstrakt
Jozefovič, Patrik - Ambrož, Ondřej - Čermák, Jan - Man, Jiří - Mikmeková, Šárka
A Comparison of Image Analysis Tools for Segmentation on SEM Micrographs - Zeiss ZEN Intellesis vs. Thermofisher AVIZO.
Microscopy and Microanalysis. Cambridge University Press. Roč. 29, S1 (2023), s. 1889-1891. ISSN 1431-9276. E-ISSN 1435-8115.
[Microscopy & Microanalysis 2023. 23.07.2023-27.07.2023, Minneapolis]
Grant ostatní: AV ČR(CZ) LQ100652201
Program: Prémie Lumina quaeruntur
Institucionální podpora: RVO:68081731 ; RVO:68081723
Klíčová slova: machine learning * image analysis * segmentation * metalography * scanning electron microscopy
Obor OECD: Materials engineering
https://academic.oup.com/mam/article/29/Supplement_1/1889/7228906

AISI301LN, a low-carbon variant of austenitic stainless steel renowned for its superior corrosion resistance, strength, and versatility, finds widespread applications across industries. However, its susceptibility to martensitic transformation during plastic deformation poses challenges to its mechanical properties. To address this, in our study we are focusing on two key aspects. Firstly, we aim to develop accelerated imaging methods for detecting martensite in the microstructure, to unlock the possibility of understanding the behaviour of the AISI301LN, as well as detecting and mapping the proportion of martensite phase in microstructure. Secondly, we explore automated image analysis techniques for precise segmentation of phases in scanning electron microscope (SEM) micrographs. We assess the performance of commercial solutions such as Thermo Fisher Amira-Avizo and ZEISS ZEN Intellesis, comparing them with custom-trained neural network models based on established architectures for image segmentation. Our research not only contributes to a deeper understanding of AISI301LN behavior but also evaluates the effectiveness of machine learning tools in enhancing its properties.
Trvalý link: https://hdl.handle.net/11104/0345089