Loading...
Publication
Facial video based physiological variables estimation in dark environments
Authors
Abstract(s)
As estimativas de parâmetros fisiológicos desempenham um papel relevante na de terminação do estado de saúde de um indivíduo. Entre esses parâmetros, a fre quência cardíaca e a saturação de oxigénio têm sido amplamente utilizadas para
monitorização da saúde durante exames médicos, cirurgias, diagnóstico de distúr bios do sono e em unidades de cuidados intensivos. As técnicas de referência para
estimar esses parâmetros são a eletrocardiografia e a fotopletismografia. Ambas são
técnicas baseadas em contacto e, portanto, podem causar desconforto ao paciente
em cenários como monitorização prolongada e pele sensível ou queimada. Assim, a
fotopletismografia remota foi introduzida como uma variante sem contacto da foto pletismografia. Esta técnica extrai o sinal de pulso do volume sanguíneo das sequên cias espaço-temporais da região de interesse, seguida pela estimativa da frequência
cardíaca. Por outro lado, as estimativas da saturação de oxigénio são realizadas us ando o método de razão de razões usando os canais vermelho e azul. Os métodos
existentes sem contacto foram projetados para condições de luz ambiente. Alguns
métodos desenvolvidos para ambientes escuros usaram câmaras infravermelhas, que
são caras, e os espectros resultantes têm força pulsátil inferior aos espectros visíveis.
Portanto, esta tese investiga o potencial dos espectros visíveis para medidas fisiológ icas em ambientes escuros (iluminância 1,0 lux). Especificamente, esta tese tem
três contribuições principais: primeiro, um novo método de estimativa da frequência
cardíaca baseado na análise de componentes independentes subcompleta, que foi de senvolvido e testado sob diferentes condições em tempo real, e segundo, um conjunto
de dados "Dark-Video" abrangendo participantes de diferentes etnias e, finalmente,
uma nova arquitetura de aprendizagem profunda para aprimoramento de imagens
escuras que também foi proposta para facilitar medições fisiológicas nos ambientes
escuros mencionados acima (ou seja, métodos de estimativa em cascata pelo apri moramento de imagens). Diversas experiências foram conduzidas para a análise de
desempenho usando métricas de desempenho selecionadas criticamente provaram
a superioridade dos métodos desenvolvidos e também exibiram o seu potencial de
serem clinicamente viáveis. A direção futura desta trabalho visa implementar esses
métodos para cenários como monitorização do sono sem contacto ou monitorização
durante a condução noturna.
Physiological parameter estimations play a significant role in determining an indi vidual’s health status. Among these parameters, heart rate and oxygen saturation have been extensively used for health monitoring during medical checkups, surgery, sleep disorders diagnosis, and intensive care units. The gold standard techniques for estimating these parameters are electrocardiography and photoplethysmogra phy. Both are contact-based techniques and, therefore, can cause discomfort to the subject in scenarios such as prolonged monitoring and sensitive or burnt skin. Thus, remote photoplethysmography was introduced as a non-contact variant of photo plethysmography. It extracts the blood volume pulse signal from the spatiotemporal sequences of the region of interest, followed by heart rate estimation. On the other hand, oxygen saturation estimations are being performed using the ratio-of-ratios method using red and blue channels. Existing non-contact methods were designed for ambient light conditions. A few methods developed for dark environments used infrared cameras, which are expensive, and the resulting spectra have poorer pul satile strength than visible spectra. Therefore, this thesis investigates the potential of visible spectra for physiological measurements in dark environments (illuminance ≤ 1.0 lux). Specifically, this thesis has three key contributions: first, a novel heart rate estimation algorithm (U-LMA) based on undercomplete independent compo nent analysis, which was developed and tested under different real-time conditions, and second, a "Dark-Video" dataset encompassing participants of different ethnici ties, and finally a novel deep learning architecture for dark image enhancement that was also proposed to facilitate physiological measurements in the above mentioned dark environments (i.e., estimation methods cascaded by image enhancement). Di verse experiments conducted for the performance analysis using critically selected performance metrics not only proved the superiority of the developed methods but also exhibited their potential of being clinically viable. The future direction of this research aims to implement these methods for scenarios such as non-contact sleep monitoring or monitoring during nighttime driving.
Physiological parameter estimations play a significant role in determining an indi vidual’s health status. Among these parameters, heart rate and oxygen saturation have been extensively used for health monitoring during medical checkups, surgery, sleep disorders diagnosis, and intensive care units. The gold standard techniques for estimating these parameters are electrocardiography and photoplethysmogra phy. Both are contact-based techniques and, therefore, can cause discomfort to the subject in scenarios such as prolonged monitoring and sensitive or burnt skin. Thus, remote photoplethysmography was introduced as a non-contact variant of photo plethysmography. It extracts the blood volume pulse signal from the spatiotemporal sequences of the region of interest, followed by heart rate estimation. On the other hand, oxygen saturation estimations are being performed using the ratio-of-ratios method using red and blue channels. Existing non-contact methods were designed for ambient light conditions. A few methods developed for dark environments used infrared cameras, which are expensive, and the resulting spectra have poorer pul satile strength than visible spectra. Therefore, this thesis investigates the potential of visible spectra for physiological measurements in dark environments (illuminance ≤ 1.0 lux). Specifically, this thesis has three key contributions: first, a novel heart rate estimation algorithm (U-LMA) based on undercomplete independent compo nent analysis, which was developed and tested under different real-time conditions, and second, a "Dark-Video" dataset encompassing participants of different ethnici ties, and finally a novel deep learning architecture for dark image enhancement that was also proposed to facilitate physiological measurements in the above mentioned dark environments (i.e., estimation methods cascaded by image enhancement). Di verse experiments conducted for the performance analysis using critically selected performance metrics not only proved the superiority of the developed methods but also exhibited their potential of being clinically viable. The future direction of this research aims to implement these methods for scenarios such as non-contact sleep monitoring or monitoring during nighttime driving.
Description
Keywords
Separação cegas de fontes Volume sanguíneo Aprendizagem profunda Métodos sem contacto Estimação de parâmetros fisiológicos Blind source separation Blood volume pulse Deep learning Non-contact approaches Physiological parameters estimation Informatics Engineering . Faculty of Exact Sciences and Engineering