0557881 - ÚPT 2023 RIV CA eng C - Konferenční příspěvek (zahraniční konf.)
Venglovskyi, Iurii
Magnetic Resonance Spectroscopy Signal Analysis Based on Fingerprinting Dictionary Approaches.
Proceedings of the 7th World Congress on Electrical Engineering and Computer Systems and Science (EECSS’21). Orléans: International ASET, 2021, č. článku ICBES114. ISBN 978-1-927877-92-0.
[International Conference on Biomedical Engineering and Systems (ICBES’21) /8./. online (CZ), 29.07.2021-31.07.2021]
Grant CEP: GA MŠMT(CZ) EF16_013/0001775
GRANT EU: European Commission(XE) 813120 - INSPiRE-MED
Institucionální podpora: RVO:68081731
Klíčová slova: magnetic resonance spectroscopy * fingerprinting dictionary * spectroscopic imaging * artificial intelligence
Obor OECD: Medical engineering
Web výsledku:
https://avestia.com/EECSS2021_Proceedings/files/paper/ICBES/ICBES_114.pdf

DOI: https://doi.org/10.11159/icbes21.114

Magnetic resonance spectroscopy (MRS) is a technique applicable in medical diagnosis or research, which has the unique capability to give non-invasive access to the biochemical content (metabolites) of scanned organs. Up to recent times, all the proposed methods solved metabolite quantification as an optimization problem attempting to minimize the difference between the data and a given parameterized model function. This paper proposes quantification of metabolites in MR spectroscopic imaging using a fingerprinting method, whose function is based on the creation of a dictionary of linear combinations of metabolite signals. Experimental results demonstrate the accuracy of the proposed method, compared to data obtained by a standard quantification method (QUEST), on concentration estimates of 8 metabolites from signals with macromolecule background and noise. The prototype results indicate that the concept of MR fingerprinting dictionary, useful also for preparing data for machine learning, can serve as an alternative method for metabolite quantification by NMR signal analysis.
Trvalý link: http://hdl.handle.net/11104/0331780