Non-contact hemodynamic imaging reveals the jugular venous pulse waveform.

Figures

• 

  Figure 1. Histogram of the strongest signal correlation values relative to ground truth arterial waveform across all participants (n = 24). The distribution is strongly bimodal, indicating one group of signals that are highly correlated with the arterial waveform, and another group that is strongly negatively correlated with the arterial waveform.

• 

  Figure 2. Comparison of scale-normalized arterial and inverted pulse waveforms in a typical example. (a) The participant exhibited strong arterial pulsations, characterised by a sharp rise to systole and a dicrotic notch, as well as strong inverted pulsations, characterised by a phase-offset gradual rise and sharp drop. (b) The arterial and inverted waveforms are strongly linked through an inverse relationship (r = 0.96). Best viewed in colour.

• 

  Figure 3. Jugular venous pulse signal acquired for each participant (n = 24). The jugular venous pulse (black line) acquired using the non-contact coded hemodynamic imaging system is compared to the ground-truth arterial waveform (dotted gray). The two waveforms are strongly inversely proportional, some more strongly than others due to factors such as fat content and age-related skin inelasticity.

• 

  Figure 4. Three frames from a typical segment showing the timing of the blood volume pulse from Supplementary Video 1. (a) Upon ventricular ejection, the pulse travels through the carotid arterial track (red), reaching systole; there is no inverted pulsatile flow. (b) During the arterial descent towards diastole, the pulse continues to travel through the carotid track, and the start of the inverted pulse can be observed (blue). (c) No carotid pulse is observed during diastole, and the inverted pulse experiences its maximum absorbance. Best viewed in colour.

• 

  Figure 5. Locations exhibiting the strongest pulsing across all participants (n = 24) relative to the position of the carotid artery and jugular vein. Locations were determined by finding and marking the carotid and jugular track using ultrasound following video data capture. The pulsing locations were normalized and drawn relative to the individual’s marked anatomy. The data are presented for the right side only (marked with ‘o’), which is used clinically as the most direct conduit to the heart. The arterial pulse points were consistent with the anatomical location of the carotid artery (a), and the inverted waveform pulse points were consistent with the distal location of the jugular vein (b). Best viewed in colour.

• 

  Figure 6. Comparison of a typical “inverted” pulse to the jugular venous pulse waveform in the Wiggers diagram (adapted from ref. 24). The inverted pulse was consistent with the JVP waveform. The JVP waveform is biphasic whose waveform inflections are governed by differential cardiac pressures (see text). The variability between individual pulses is likely due to the effect of respiration on the intrathoracic pressure.

• 

  Figure 7. Example of a single pulse extracted from each participant. The characteristic JVP waves were manually identified in each waveform and annotated for clarity. Four of the major JVP waves (c, x, v, y waves) were observed in all participants. The subtle pre-systolic a wave was observed in 13/24 participants.

• 

  Figure 8. Inverse relationship between the carotid artery and jugular vein as seen with B-mode ultrasound in a typical participant. The jugular vein (below V) and carotid artery (below A) are located below the skin surface (top) with vessel diameters marked in millimetres. Between systole (a) and diastole (b), the carotid relaxes (10% diameter reduction), and the jugular expands (12% diameter expansion).

• 

  Figure 9. Study setup. Participants were supine for the duration of the study. The biophotonic imaging system was positioned above and slightly to the right of the participant, at a distance of 1.5 m. Illumination was provided by a spatially uniform 250 W tungsten-halogen illumination source. Imaging data were processed on a digital signal processing (DSP) unit. The participant wore a finger cuff which provided the ground truth arterial waveform for the analysis.

• 

  Figure 10. Data processing pipeline. Each frame was analysed in 0.25 × 0.25 mm regions. Each region’s temporal fluctuations were converted to absorbance using (2). Pearson’s linear correlation coefficient was computed for each waveform using the ground truth PPG waveform, yielding a spatial correlation map showing locations exhibiting strong forward and inverted pulsing. Best viewed in colour.