DistancePPG: Robust non-contact vital signs monitoring using a camera (1502.08040v2)

Abstract: Vital signs such as pulse rate and breathing rate are currently measured using contact probes. But, non-contact methods for measuring vital signs are desirable both in hospital settings (e.g. in NICU) and for ubiquitous in-situ health tracking (e.g. on mobile phone and computers with webcams). Recently, camera-based non-contact vital sign monitoring have been shown to be feasible. However, camera-based vital sign monitoring is challenging for people with darker skin tone, under low lighting conditions, and/or during movement of an individual in front of the camera. In this paper, we propose distancePPG, a new camera-based vital sign estimation algorithm which addresses these challenges. DistancePPG proposes a new method of combining skin-color change signals from different tracked regions of the face using a weighted average, where the weights depend on the blood perfusion and incident light intensity in the region, to improve the signal-to-noise ratio (SNR) of camera-based estimate. One of our key contributions is a new automatic method for determining the weights based only on the video recording of the subject. The gains in SNR of camera-based PPG estimated using distancePPG translate into reduction of the error in vital sign estimation, and thus expand the scope of camera-based vital sign monitoring to potentially challenging scenarios. Further, a dataset will be released, comprising of synchronized video recordings of face and pulse oximeter based ground truth recordings from the earlobe for people with different skin tones, under different lighting conditions and for various motion scenarios.

• The paper presents DistancePPG, a non-contact algorithm that accurately estimates vital signs from facial videos in challenging conditions.
• The paper employs weighted averaging of facial regions and automatic weight determination to improve signal quality across diverse skin tones and low-light scenarios.
• The paper achieves significant SNR improvements, with up to 7 dB gain for darker skin tones and 4.5 dB in motion scenarios, validated on a comprehensive dataset.

An Academic Review of "DistancePPG: Robust non-contact vital signs monitoring using a camera"

The paper "DistancePPG: Robust non-contact vital signs monitoring using a camera" by Kumar et al. introduces a novel algorithm named DistancePPG for non-contact monitoring of vital signs using video inputs. The focus of this research is the estimation of photoplethysmographic (PPG) signals from facial videos, tackling common challenges such as variations in skin tone, low lighting conditions, and subject movement.

Traditional methods for capturing vital signs like pulse and breathing rates rely extensively on contact-based sensors. While effective, these techniques can be invasive or impractical in some scenarios, underscoring the need for non-contact alternative methods. Previous endeavors into camera-based monitoring methods have demonstrated feasibility but fall short in scenarios involving poor lighting or subjects with darker skin tones.

Technical Contributions

1. Weighted Averaging of Regions: The DistancePPG algorithm improves the signal-to-noise ratio (SNR) of PPG estimates by employing a weighted average of skin-color signals from different facial regions. The novelty lies in the method of determining these weights, which considers the blood perfusion and light intensity incident in the region. This is essential for enhancing the quality of captured PPG signals across varying skin tones and lighting conditions.
2. Automatic Weight Determination: The paper proposes a new automatic technique to compute the optimal weights based on video recordings without external inputs. This self-contained approach allows DistancePPG to dynamically adapt to different subjects and environmental conditions.
3. Region-Based Motion Tracking: The authors address motion artifacts by implementing a region-specific tracking mechanism using deformable face tracking and Kanade-Lucas-Tomasi (KLT) feature trackers. This method assists in maintaining accurate PPG signal estimation despite motion by focusing on specific regions of the face that move less or maintain signal integrity.
4. Dataset and Evaluation: Kumar et al. also contribute a crucial dataset comprising synchronized video recordings and pulse oximeter ground truth data for a diverse subject group. This includes varying skin tones, lighting, and motion scenarios, thereby providing a comprehensive evaluation for the proposed methods.

Evaluation and Results

The authors present a detailed robustness evaluation of DistancePPG through static and dynamic scenarios varying in skin tone and light exposure. The improvements reported in SNR are particularly notable for darker skin tones, achieving up to a 6-7 dB increase compared to previous methods. In motion scenarios like reading or watching, the SNR improvements averaged to 4.5 dB. The paper also presents substantial reductions in estimation errors for pulse rate variability (PRV) and heart rate under these specified conditions.

Implications and Future Work

The proposed method expands the potential for in-situ health monitoring applications, particularly in environments where traditional sensors may be undesirable or impractical. DistancePPG could significantly enhance mobile health monitoring applications deployed on smartphones or computers, especially by providing more reliable readings in challenging scenarios or across diverse populations.

From a theoretical perspective, the automatic determination of weights for signal regions and enhanced region-based tracking can influence future work in remote sensing and biomedical signal processing. Improvements in non-contact methodologies could drive advancements in telemedicine and remote patient monitoring, particularly in under-resourced settings or during infectious disease outbreaks where non-contact solutions are paramount.

Looking forward, optimizing DistancePPG for real-time implementations and integrating it into existing telehealth platforms are natural progression paths. There is anticipation for further research to reduce errors in non-contact PR and PRV estimation under substantial motion, by leveraging advanced subjects' pose estimation and motion tracking frameworks.

In conclusion, the presented work on DistancePPG affords a significant step towards the broader adoption of remote physiological monitoring, bridging vital gaps in current technology and making strides toward inclusive healthcare solutions.