Deep learning-based method for the continuous detection of heart rate in signals from a multi-fiber Bragg grating sensor compatible with magnetic resonance imaging

Mariusz Krej; Tomasz Osuch; Alicja Anuszkiewicz; Stanisław Stopinski; Krzysztof Anders; Krzysztof Matuk; Andrzej Weigl; Eugeniusz Tarasow; Ryszard Piramidowicz; Lukasz Dziuda

doi:10.1364/BOE.441932

Deep learning-based method for the continuous detection of heart rate in signals from a multi-fiber Bragg grating sensor compatible with magnetic resonance imaging

Biomed Opt Express. 2021 Nov 24;12(12):7790-7806. doi: 10.1364/BOE.441932. eCollection 2021 Dec 1.

Authors

Mariusz Krej¹, Tomasz Osuch^{2

3}, Alicja Anuszkiewicz^{2

4}, Stanisław Stopinski⁵, Krzysztof Anders⁵, Krzysztof Matuk⁶, Andrzej Weigl⁶, Eugeniusz Tarasow^{6

7}, Ryszard Piramidowicz⁵, Lukasz Dziuda¹

Affiliations

¹ Military Institute of Aviation Medicine, Department of Psychophysiological Measurements and Human Factor Research, Krasinskiego 54/56, 01-755 Warsaw, Poland.
² Warsaw University of Technology, Faculty of Electronics and Information Technology, Institute of Electronic Systems, Warsaw University of Technology, Nowowiejska 15/19, 00-665 Warsaw, Poland.
³ National Institute of Telecommunications, Szachowa 1, 04-894 Warsaw, Poland.
⁴ Lukasiewicz Research Network - Institute of Microelectronics and Photonics, Photonic Materials Group, al. Lotnikow 32/46, 02-668 Warsaw, Poland.
⁵ Warsaw University of Technology, Faculty of Electronics and Information Technology, Institute of Microelectronics and Optoelectronics, Koszykowa 75, 00-662 Warsaw, Poland.
⁶ TMS Diagnostyka Sp. z o.o., Wiertnicza 84, 02-952 Warsaw, Poland.
⁷ Medical University of Bialystok, Faculty of Medicine, Department of Radiology, Kilinskiego 1, 15-089 Bialystok, Poland.

Abstract

A method for the continuous detection of heart rate (HR) in signals acquired from patients using a sensor mat comprising a nine-element array of fiber Bragg gratings during routine magnetic resonance imaging (MRI) procedures is proposed. The method is based on a deep learning neural network model, which learned from signals acquired from 153 MRI patients. In addition, signals from 343 MRI patients were used for result verification. The proposed method provides automatic continuous extraction of HR with the root mean square error of 2.67 bpm, and the limits of agreement were -4.98-5.45 bpm relative to the reference HR.