Imaging photoplethysmography is a technique through which the morphology of the blood volume pulse can be obtained through non-contact video recordings of exposed skin with superficial vasculature. The acceptance of such a convenient modality for use in everyday applications may well depend upon the availability of consumer-grade imagers that facilitate ease-of-adoption. Multiple imagers have been used previously in concept demonstrations, showing improvements in quality of the extracted blood volume pulse signal. However, the use of multi-imager sensors requires synchronization of the frame exposures between the individual imagers, a capability that has only recently been available without creating custom solutions. In this work, we consider the use of multiple, commercially-available, synchronous imagers for use in imaging photoplethysmography. A commercially-available solution for adopting multi-imager synchronization was analyzed for 21 stationary, seated participants while ground-truth physiological signals were simultaneously measured. A total of three imagers were used, facilitating a comparison between fused data from all three imagers versus data from the single, central imager in the array. The within-subjects design included analyses of pulse rate and pulse signal-to-noise ratio. Using the fused data from the triple-imager array, mean absolute error in pulse rate measurement was reduced to 3.8 as compared to 7.4 beats per minute with the single imager. While this represents an overall improvement in the multi-imager case, it is also noted that these errors are substantially higher than those obtained in comparable studies. We further discuss these results and their implications for using readily-available commercial imaging solutions for imaging photoplethysmography applications.
Imaging photoplethysmography uses image sensors to measure changes in light absorption resulting from skin microvascular blood volume pulsations throughout the cardiac cycle. Imaging photoplethysmography has been demonstrated as an effective, non-contact means of assessing pulse rate, pulse rate variability, and respiration rate. Other potential uses include measuring spatial blood perfusion, oxygenation, and flow dynamics. Herein we demonstrate the development of a multispectral testbed for imaging photoplethysmography consisting of 12 monochromatic, 120fps imagers with 50nm, bandpass filters distributed from 400-750nm and contained in a 3D-printed, 4x3 grid housing mounted on a tripod positioned orthogonal to the subject. A co-located dual-CCD RGB/near-infrared imager records conventional RGB and NIR images expanding the spectral window recorded. After image registration, a multispectral image cube of the 13, partially overlapping bands is created. A spectrometer records high (spectral) resolution data from the participant’s right cheek using a collimating lens attached to the measurement fiber. In addition, a spatial array of 5 RGB imagers placed at 0°, ±20° and ±40° positions with respect to the subject is employed for motion and spatial robustness. All imagers are synchronized by a hardware trigger source synchronized with a reference, physiological measurement device recording the subject’s electrocardiography, bilateral fingertip and/or ear lobe photoplethysmography, bilateral galvanic skin response, and respiration signals. The development of the testbed and pilot data is presented. A full-scale evaluation of the spectral components of the imaging photoplethysmographic signal, optimization of iPPG SNR, and spatial perfusion and blood flow dynamics is currently underway.
Imaging photoplethysmography, a method using imagers to record absorption variations caused by microvascular blood volume pulsations, shows promise as a non-contact cardiovascular sensing technology. The first long-range imaging photoplethysmography measurements at distances of 25, 50, and 100 meters from the participant was recently demonstrated. Degraded signal quality was observed with increasing imager-to-subject distances. The degradation in signal quality was hypothesized to be largely attributable to inadequate light return to the image sensor with increasing lens focal length. To test this hypothesis, a follow-up evaluation with 27 participants was conducted outdoors with natural sunlight illumination resulting in ~5-33 times the illumination intensity. Video was recorded from cameras equipped with ultra-telephoto lenses and positioned at distances of 25, 50, 100, and 150 meters. The brighter illumination allowed high-definition video recordings at increased frame rates of 60fps, shorter exposure times, and lower ISO settings, leading to higher quality image formation than the previous indoor evaluation. Results were compared to simultaneous reference measurements from electrocardiography. Compared to the previous indoor study, we observed lower overall error in pulse rate measurement with the same pattern of degradation in signal quality with respect to increasing distance. This effect was corroborated by the signal-to-noise ratio of the blood volume pulse signal which also showed decreasing quality with respect to increasing distance. Finally, a popular chrominance-based method was compared to a blind source separation approach; while comparable in measurement of signal-to-noise ratio, we observed higher overall error in pulse rate measurement using the chrominance method in this data.
Non-contact, imaging photoplethysmography uses photo-optical sensors to measure variations in light absorption, caused by blood volume pulsations, to assess cardiopulmonary parameters including pulse rate, pulse rate variability, and respiration rate. Recently, researchers have studied the applications and methodology of imaging photoplethysmography. Basic research has examined some of the variables affecting data quality and accuracy of imaging photoplethysmography including signal processing, imager parameters (e.g. frame rate and resolution), lighting conditions, subject motion, and subject skin tone. This technology may be beneficial for long term or continuous monitoring where contact measurements may be harmful (e.g. skin sensitivities) or where imperceptible or unobtrusive measurements are desirable. Using previously validated signal processing methods, we examined the effects of imager-to-subject distance on one-minute, windowed estimates of pulse rate. High-resolution video of 22, stationary participants was collected using an enthusiast-grade, mirrorless, digital camera equipped with a fully-manual, super-telephoto lens at distances of 25, 50, and 100 meters with simultaneous contact measurements of electrocardiography, and fingertip photoplethysmography. By comparison, previous studies have usually been conducted with imager-to-subject distances of up to only a few meters. Mean absolute error for one-minute, windowed, pulse rate estimates (compared to those derived from gold-standard electrocardiography) were 2.0, 4.1, and 10.9 beats per minute at distances of 25, 50, and 100 meters, respectively. Long-range imaging presents several unique challenges among which include decreased, observed light reflectance and smaller regions of interest. Nevertheless, these results demonstrate that accurate pulse rate measurements can be obtained from over long imager-to-participant distances given these constraints.
Non-contact, imaging photoplethysmography uses cameras to facilitate measurements including pulse rate, pulse rate variability, respiration rate, and blood perfusion by measuring characteristic changes in light absorption at the skin’s surface resulting from changes in blood volume in the superficial microvasculature. Several factors may affect the accuracy of the physiological measurement including imager frame rate, resolution, compression, lighting conditions, image background, participant skin tone, and participant motion. Before this method can gain wider use outside basic research settings, its constraints and capabilities must be well understood. Recently, we presented a novel approach utilizing a synchronized, nine-camera, semicircular array backed by measurement of an electrocardiogram and fingertip reflectance photoplethysmogram. Twenty-five individuals participated in six, five-minute, controlled head motion artifact trials in front of a black and dynamic color backdrop. Increasing the input channel space for blind source separation using the camera array was effective in mitigating error from head motion artifact. Herein we present the effects of lower frame rates at 60 and 30 (reduced from 120) frames per second and reduced image resolution at 329x246 pixels (one-quarter of the original 658x492 pixel resolution) using bilinear and zero-order downsampling. This is the first time these factors have been examined for a multiple imager array and align well with previous findings utilizing a single imager. Examining windowed pulse rates, there is little observable difference in mean absolute error or error distributions resulting from reduced frame rates or image resolution, thus lowering requirements for systems measuring pulse rate over sufficient length time windows.