Illumination Invariant Non-Invasive Heart Rate And Blood Pressure Estimation From Facial Thermal Images using Deep Learning

Nair, Kavya S; Sarath, S

doi:10.1109/icccnt51525.2021.9579621

Cited by 3 publications

(6 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As a result, the model with Boruta as the variable selection method and SVR as the modeling method had the highest accuracy in resting blood pressure estimation. The set number of ICs was 60, the number of explanatory variables chosen by Boruta was 8, and the hyperparameters used were 2 6 for C, 2 0 for ε, and 2 −2 for γ . The mean and standard deviation of the RMSE for this model are smaller than for the stepwise and MRA models, the same methods used in previous studies, which indicating improved generalization performance.…”

Section: Resultsmentioning

confidence: 99%

“…The objective of this study is to optimize the variable selection and the modeling method to improve the estimation accuracy of the general model for resting BP estimation based on facial thermal images. As a result, the model constructed with 60 ICs, Boruta as the variable selection method, SVR as the modeling method, and the hyperparameters used were 2 6 for C, 2 0 for , and 2 −2 for γ , which had the highest accuracy. The RMSE of the best BP estimation model was 6.80 mmHg and the correlation coefficient was 0.543.…”

Section: Discussionmentioning

confidence: 99%

“…Therefore, it is thought that BP fluctuations could be spatially captured as facial skin temperature fluctuations from facial thermal images (FTIs) measured by a thermography device. One study extracts PPG from FTIs and uses deep learning model to estimate BP [6]. Since this technique extracts PPG from FTIs, it is necessary to take a series of FTIs.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

An Optimization for Resting Blood Pressure Estimation Model Based on Independent Components of Facial Thermal Images

Ito,

Nagumo,

Oiwa

et al. 2024

IEEJ Transactions Elec Engng

View full text Add to dashboard Cite

The increasing number of patients with hypertension is a problem worldwide. Daily blood pressure monitoring is important for the prevention of this disease, and a non‐contact measurement method is desirable. A previous study applied independent components analysis to facial skin temperatures measured remotely using a thermography device for extracting features related to resting blood pressure. Then, a blood pressure estimation model was constructed and obtained an accuracy of 7.73 mmHg root mean square error. However, this previous study did not consider the variables selection methods and models to be constructed. Considering these factors is expected to improve the accuracy of estimation. The objective of this study is to optimize the resting blood pressure estimation model. Two variable selection methods and three modeling methods were applied to independent components of facial thermal images, and the accuracy was compared. The result showed an accuracy of 6.80 mmHg root mean square error with a pattern of support vector regression model using independent components selected by Boruta. In addition, eight explanatory variables were selected by Boruta, and suggesting that the lip, orbital, and lateral nasal regions were related to resting blood pressure. © 2024 Institute of Electrical Engineer of Japan and Wiley Periodicals LLC.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

An Optimization for Resting Blood Pressure Estimation Model Based on Independent Components of Facial Thermal Images

Ito,

Nagumo,

Oiwa

et al. 2024

IEEJ Transactions Elec Engng

View full text Add to dashboard Cite

show abstract

“…For pulse rate estimation, it may use an invasive PPG sensor placed on the index finger [59]. The same apparent for estimated BP, which may favor the need for an invasive PPG sensor [63] and a sphygmomanometer [64] to analyze the level of BP. In general, standard biochemical invasive methods such as glucometer can be used to collect blood glucose [44].…”

Section: B Biosensors Signal Processing Methodsmentioning

confidence: 99%

“…Another paper showed that estimated HR and BP from thermal images can be done by transforming the raw PPG signals to the frequency domain using FFT and applying Inverse FFT (IFFT) to transform frequency domain to time domain [63]. The preprocessed PPG signals are then filtered using the Short Time Fourier Transform (STFT) for HR and BP using the Infinite Impulse Response (IIR) Chebyshev Type 2 Filter (CT2F).…”

Section: ) Multi-parameter Vital Signs Extraction A) Deep Learning Me...mentioning

confidence: 99%

A Review of Facial Thermography Assessment for Vital Signs Estimation

2022

View full text Add to dashboard Cite

Estimated vital signs might include a variety of measurements that can be used in detecting any abnormal conditions by analyzing facial images from continuous monitoring with a thermal video camera. To overcome the limitless human visual perceptions, thermal infrared has proven to be the most effective technique for visualizing facial colour changes that could have been reflected by changes in oxygenation levels and blood volume in facial arteries. This study investigated the possibility of vital signs estimation using physiological function images converted from the thermal infrared images in the same ways that visible images are used, with a need for an efficient extractor method as correction procedures that have used datasets that include images with and without wearing glasses or protective face masks. This paper, summarize thermal images using advanced machine learning and deep learning methods with satisfactory performance. Also, we presented the evaluation matrices that were included in the assessment based on statistical analysis, accuracy measures and error measures. Finally, to discuss future gaps and directions for further evaluations.INDEX TERMS Thermal images, features extractions, vital signs estimation, evaluation matrices.

show abstract

LVT Face Database: A Benchmark Database for Visible and Hidden Face Biometrics

Mirabet-Herranz,

Dugelay

2023

2023 International Conference of the Biometrics Special Interest Group (BIOSIG)

View full text Add to dashboard Cite

Although the estimation of eHealth parameters from face visuals (images and videos) has grown as a major area of research in the past years, deep-learning-based models are still challenged by RGB lack of robustness, for instance with changing illumination conditions. As a means to overcome these limitations and to unlock new opportunities, thermal imagery has arisen as a favorable alternative to solidify different technologies such as heart rate estimation from faces. However, the reduced number of databases containing thermal imagery and the lack of health annotation of the subjects in them limits the exploration of this spectrum. Motivated by this, in this paper, we present our Label-EURECOM Visible and Thermal (LVT) Face Database for face biometrics. This database is the first that contains paired visible and thermal images and videos from 52 subjects with metadata of 22 soft biometrics and health parameters. Moreover, we establish the first study introducing the potential of thermal images for weight estimation from faces on our database.

show abstract

Illumination Invariant Non-Invasive Heart Rate And Blood Pressure Estimation From Facial Thermal Images using Deep Learning

Cited by 3 publications

References 22 publications

An Optimization for Resting Blood Pressure Estimation Model Based on Independent Components of Facial Thermal Images

An Optimization for Resting Blood Pressure Estimation Model Based on Independent Components of Facial Thermal Images

A Review of Facial Thermography Assessment for Vital Signs Estimation

LVT Face Database: A Benchmark Database for Visible and Hidden Face Biometrics

Contact Info

Product

Resources

About