TY - RPRT A1 - Hodoroaba, Vasile-Dan A1 - Fontanges, R. A1 - Delvallée, A. A1 - Deumer, J. A1 - Salzmann, C. A1 - Crouzier, L. A1 - Gollwitzer, C. A1 - Klein, T. A1 - Koops, R. A1 - Sebaihi, N. A1 - Pauw, Brian Richard A1 - Smales, Glen Jacob A1 - Artous, S. A1 - Taché, O. A1 - Feltin, N. T1 - Report on full algorithm sequences for nanoparticle detection and size measurement as developed on both a physical basis and by machine learning N2 - he main objective of the nPSize project is to improve the measurement capabilities for nanoparticle size based on both measurement methods traceable to SI units and new reference materials. Two basic approaches have been used in order to develop measurement procedures resulting in traceable results of the nanoparticle size distribution: physical modelling for the methods used in the project (TSEM, SEM, AFM and SAXS) and machine learning. Physical modelling: In this part, the physical models associated with different shape measurements for the techniques TSEM, SEM, AFM and SAXS have been collected and further developed with the aim to simulate the resulting signal as measured by the individual methods. Uncertainties and traceability associated with each model were investigated and evaluated. In the following, the progress on these physical models is reported for each individual method. Machine Learning modelling: The aim of this part is to use machine learning to enable automatic measurement of nanoparticle shape from expert a-priori information only. No physical model will be used as a-priori information in this task. The accuracy and traceability of the size results obtained by each technique will be analyzed and compared with the physical modelling. A machine learning database will then be used to create automatic detection algorithms. KW - Nanoparticles KW - Particle size distribution KW - SEM KW - TSEM KW - TEM KW - SAXS KW - AFM PY - 2022 UR - https://nbn-resolving.org/urn:nbn:de:kobv:b43-546531 DO - https://doi.org/10.5281/zenodo.5807864 SP - 1 EP - 20 PB - Zenodo CY - Geneva AN - OPUS4-54653 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER - TY - GEN A1 - Hodoroaba, Vasile-Dan A1 - Fontanges, Richard T1 - A new deep-learning AI tool for analysing images of complex nanoparticles N2 - A thousand times thinner than a human hair, nanoparticles (NPs) are finding applications in a range of modern products. However, as some can affect human health or the environment, knowing the types present is essential. Electron microscopy is the ‘gold standard’ for NP analysis, allowing identification based on manual size analysis, but a new method was required to analyse these particles quickly, accurately and in a consistent way. KW - Nanoparticles KW - Imaging KW - AI tool KW - Particle size and shape distribution PY - 2024 UR - https://www.euramet.org/casestudies/casestudiesdetails/news/a-new-deep-learning-ai-tool-for-analysing-images-of-complex-nanoparticles SP - 1 EP - 2 PB - EURAMET CY - Braunschweig AN - OPUS4-60095 LA - eng AD - Bundesanstalt fuer Materialforschung und -pruefung (BAM), Berlin, Germany ER -