Number of the records: 1
Artificial intelligence-aided single-molecule bioaffinity assays with photon-upconversion labels for microfluidic applications
- 1.0576913 - ÚIACH 2024 CZ eng A - Abstract
Hlaváček, Antonín - Křivánková, Jana - Weisová, Julie
Artificial intelligence-aided single-molecule bioaffinity assays with photon-upconversion labels for microfluidic applications.
12th International Conference Analytical Cytometry : Book of abstracts. Praha: AMCA, spol. s. r. o., 2023. s. 1-1.
[International Conference Analytical Cytometry /12./. 02.09.2023-05.09.2023, Praha]
R&D Projects: GA ČR(CZ) GA21-03156S
Institutional support: RVO:68081715
Keywords : photon-upconversion nanoparticles * microfluidics * microfluidic analysis
OECD category: Analytical chemistry
https://www.conference.csac.cz/Amca-CSAC/media/content/2023/docs/Book_of_abstracts_CSAC_2023.pdf
Microfluidics and femtoliter arrays became essential for single-molecule enzyme-linked immunosorbent assays with femtomolar limits of detection. Although successful commercially, a key impediment is shared with other heterogeneous assays – the separation of immunochemical complexes on solid surfaces. In contrast, homogeneous bioaffinity assays perform all assay steps in a free dispersion – very useful for automated microfluidic analysis. Here we present a new approach for homogeneous assays, which utilizes background-free luminescence of photon-upconversion nanoparticles, customized optical instrumentation, and artificial intelligence. As a model system, the formation of bioaffinity complexes between nanoparticle-conjugated streptavidin and biotin is investigated, and a homogeneous single-molecule competitive assay for biotin is developed. The instrumentation utilizes a laboratory-built epiphoton-upconversion microscope. High-intensity near-infrared excitation (976 nm, 14 kW cm-2) enables the imaging of freely diffusing photon-upconversion labels as diffraction-limited spots (5 ms exposition time). The detection of bioaffinity interactions from micrographs is automated by using convolutional neural networks.
Permanent Link: https://hdl.handle.net/11104/0346311
Number of the records: 1