TY - GEN
T1 - Non-contact continuous blood pressure monitoring from video
AU - Liu, Haojie
AU - Kong, Lingqin
AU - Liu, Han
AU - Wang, Yisheng
AU - Liu, Zihan
AU - Wang, Huiying
AU - Li, Jinmei
AU - Dong, Liquan
AU - Liu, Ming
AU - Chu, Xuhong
AU - Li, Cuiling
N1 - Publisher Copyright:
© 2025 SPIE.
PY - 2025
Y1 - 2025
N2 - This paper presents a non-contact continuous blood pressure detection method based on video. This method integrates the correlation between systolic blood pressure and diastolic blood pressure, and realizes non-contact continuous blood pressure monitoring of individuals. In this method, finger videos are collected by ordinary industrial cameras, and imaging photoplethysmography (IPPG) signals are extracted. De-trending and band-pass filtering are used to remove the baseline drift and noise of original IPPG signal. And an effective single IPPG signal is screened by limiting the number of sampling points, the amplitude changes of the starting point and the ending point and whether there is a dicrotic wave, and the characteristics of signal such as peak value, systolic time, diastolic time and pulse width are extracted, and the regression models of systolic blood pressure and diastolic blood pressure are established respectively, finally the noncontact detection of individual blood pressure is realized. The experimental results show that this method has high accuracy, the mean absolute error (MAE) of systolic blood pressure is 1.34 mmHg, the mean error (ME) is 0.46 mmHg, the standard deviation (STD) is 1.95 mmHg; the mean error (MAE) of diastolic blood pressure is 1.59 mmHg, the mean error (ME) is 0.36 mmHg, and the standard deviation (STD) is 2.01 mmHg.
AB - This paper presents a non-contact continuous blood pressure detection method based on video. This method integrates the correlation between systolic blood pressure and diastolic blood pressure, and realizes non-contact continuous blood pressure monitoring of individuals. In this method, finger videos are collected by ordinary industrial cameras, and imaging photoplethysmography (IPPG) signals are extracted. De-trending and band-pass filtering are used to remove the baseline drift and noise of original IPPG signal. And an effective single IPPG signal is screened by limiting the number of sampling points, the amplitude changes of the starting point and the ending point and whether there is a dicrotic wave, and the characteristics of signal such as peak value, systolic time, diastolic time and pulse width are extracted, and the regression models of systolic blood pressure and diastolic blood pressure are established respectively, finally the noncontact detection of individual blood pressure is realized. The experimental results show that this method has high accuracy, the mean absolute error (MAE) of systolic blood pressure is 1.34 mmHg, the mean error (ME) is 0.46 mmHg, the standard deviation (STD) is 1.95 mmHg; the mean error (MAE) of diastolic blood pressure is 1.59 mmHg, the mean error (ME) is 0.36 mmHg, and the standard deviation (STD) is 2.01 mmHg.
KW - Blood pressure monitoring
KW - non-contact
KW - regression modelling
UR - http://www.scopus.com/inward/record.url?scp=85219388347&partnerID=8YFLogxK
U2 - 10.1117/12.3056993
DO - 10.1117/12.3056993
M3 - Conference contribution
AN - SCOPUS:85219388347
T3 - Proceedings of SPIE - The International Society for Optical Engineering
BT - Tenth Symposium on Novel Optoelectronic Detection Technology and Applications
A2 - Ping, Chen
PB - SPIE
T2 - 10th Symposium on Novel Optoelectronic Detection Technology and Applications
Y2 - 1 November 2024 through 3 November 2024
ER -