1. Introduction

Changes in fluid pressures within the circulatory system's blood vessels reflect the heart's mechanical function and the circulatory system's compliance.

Central venous pressure (CVP) is an estimate of right atrial pressure and is often used to assess hemodynamic status. A primary problem with assessing CVP lies in the current gold standard measurement method: catheterization using a central line catheter (CLC). Catheterization requires surgically inserting a central line via the jugular vein into the superior vena cava or right atrium. CLC is an invasive procedure requiring advanced expertise and skill. Complications include pain at the cannulation site, local hematoma, infection (both at the site as well as bacteremia), misplacement into another vessel (possibly causing arterial puncture or cannulation), vessel laceration or dissection, air embolism, thrombosis, and hemothorax or pneumothorax requiring a possible chest tube [1]. Therefore, although CVP can provide important clinical insights, the insertion of CLC is reserved primarily for Emergency Department and Intensive Care Unit settings [2]. Thus, less invasive alternatives to the CVP are required. Additionally, since catheter monitoring is limited to measuring at a single location, spatial flow perfusion characteristics cannot be assessed, which may encode important clinical information [3].

It may be possible to use venous system blood volume as a surrogate for CVP. For example, Pelluri and Kotamraju [4] found a statistically significant (p < 0.01) correlation between inferior vena cava diameter and CVP response to fluid resuscitation. Similarly, in a prospective observational study, Kumar et al. [5] found that right subclavian vein respiratory variation can predict fluid responsiveness in a spontaneously breathing patient in circulatory shock and correlates with the inferior vena cava collapsibility index.

However, the subclavian vein is a deep vein, and similarly to the vena cava, it requires ultrasound to measure its diameter (a proxy for the blood volume).

Thus, developing an approach for simpler venous system blood volume assessment can be of great clinical significance, with a myriad of potential applications, including remote patient monitoring. Potentially, the assessment can be based on certain superficial veins (e.g., external jugular vein) or easily accessible deep veins (e.g., internal jugular vein).

In particular, the potential approach can be based on the jugular vein pressure, which has well-established clinical utility. The jugular vein (JV) is a major venous extension of the heart's right atrium, so right atrial pressure changes can be indirectly extracted from jugular vein observations.

The JV pressure can be considered a parameter highly correlated with CVP but can be measured non-invasively [6]. For example, JV pressure and the corresponding jugular venous pulse (JVP) waveform is a powerful diagnostic tool for assessing venous filling. Alterations from normality provide insight into cardiac function associated with the right heart chambers in pulmonary hypertension, tricuspid stenosis [6]), mechanical diseases (e.g., tricuspid regurgitation [6]), electrical diseases (e.g., atrial fibrillation, heart block, atrioventricular dissociation [6]), abnormal external forces (e.g., tamponade, tension pneumothorax, constrictive pericarditis [6]), and heart failure. The literature review identified several critical shortcomings associated with current blood flow motion assessment methods in jugular veins.

Firstly, there are no convenient, reliable, and objective methods for doing so – assessment can be performed non-invasively during a physical examination [6], however, such a bedside manual measurement represents a significant clinical challenge. It is crucial that the examiner distinguishes between venous and arterial pulsations and that the top of the venous column is recognized. According to Applefeld [6], evaluating the JVP is perhaps one of the most challenging skills to train in mastering physical diagnosis techniques. Brennan et al. [7] found that medical residents were unable to identify the JV in 37% of the patients. In estimating the JV pressure, the residents assigned the patients to the normal pressure group 64% of the time. In contrast, only 36% of the patients were normal by echo and pressure measurements, indicating a high false negative rate and, consequently, a low sensitivity. Thus, more objective clinical tools are required to measure JV pressure and assess the waveform accurately.

Ultrasound has recently been proposed to measure the JVP waveform through Doppler velocity imaging [8], [9]. Even though this method is becoming more affordable, advancing towards handheld devices, such technology requires stable probe skin contact and trained ultrasound technicians, and is limited to axial hemodynamic information. Moreover, tissue alteration may occur due to the application of pressure required for sufficient contact with the skin, and specifically may obstruct the venous flow and affect results, which generally motivates exploring non-contact techniques.

To date, a few non-contact approaches have been proposed to measure the deformation of the JV and the carotid artery as an indication of the blood pressure in the corresponding vessel using PPG imaging by a color camera [10], nonuniform light by a color camera [11], subpixel image registration [12], and Specular Reflection Vascular Imaging (SRVI) [13]. All modalities possess the desirable characteristic of absence of any pressure for signals acquisition, therefore not altering tissue properties, unlike ultrasound. However, the PPG imaging study of Amelard et al. [10] reported that the JVP waveform peak was visible in only half of the subjects, which indicates the lack of sensitivity of these methods. Further, the proposed methods required a stand to hold the camera [10], nonuniform illumination of the neck [11], subpixel image registration [12], or special imaging geometry [14], which can be impractical in many clinical applications. Ideally, remote measurement would be possible with a consumer-grade camera, such as that available in a mobile phone – the use of that platform facilitates both data acquisition and transmission, in addition to being a low-cost, broadly accessible solution. These characteristics are critical for success of applications such as remote patient monitoring, which necessitate low-cost avenues for providing robust diagnostic-grade data to the physician.

In addition to rPPG, the contact PPG (cPPG) was shown to successfully extract the JVP waveform from the anterior jugular vein [15]. However, anterior JV morphology displays significant intersubject variability, which may prevent translating this technology into clinical practice. Using larger vessels with less variable morphology, like internal jugular veins, can help with clinical translation and adoption.

The second shortcoming found in the literature is that the absence of realistic experimental models limits the method development for venous system blood volume analysis. For example, while the possibility of remote detection of blood motion in the JV has been reported using several methods [10–13], all these results were reported in healthy volunteers. Moreover, these measurements were taken either in a sitting or supine position. To the authors’ best knowledge, the possibility of noninvasive detection of venous blood motion in abnormal conditions was not explored so far.

The best approach would be collecting data on a realistic clinical cohort, i.e., with elevated JV pressure. However, measurement on a realistic population is challenging due to multiple factors, including scarcity of such patients, heterogeneity of relevant factors (e.g., skin tone, severity), and ethical concerns. All these factors raise the issue of statistical validity and reproducibility. Thus, developing a reproducible experimental model is important as a foundation for further advancement.

To address this issue, we can consider that venous return can be modulated using certain physiological maneuvers. In particular, venous return to the heart decreases during the straining phase of the Valsalva maneuver and the squatting-to-standing maneuver. Venous return increases during passive leg elevation and the standing-to-squatting maneuver.

The aim of this preliminary study was two-fold: a) to investigate whether photoplethysmography can be used to characterize the fluid motion in the venous system, particularly in internal JV (IJV), and b) to propose a model for venous system blood volume modulations. To this aim, we monitored the IJVs using contact and rPPG techniques (acquired with a consumer-grade camera embedded within a standard smartphone) during clinically validated physiological tests: abdominojugular test and breath holding.

2. Material and methods

As the neck contains multiple large blood vessels, which can complicate the signal interpretation, we have selected the IJV as a primary target.

While the external JV (IJV’s superficial counterpart) collects blood from the face and neck, the IJV drains blood from the brain. Further, while cerebral venous outflow occurs predominantly through the IJVs in the supine position, flow distribution is modified in different body positions. For instance, in the upright position, the IJVs partially or fully collapse due to the atmospheric pressure exceeding the intraluminal pressure, diverting cerebral venous outflow to the vertebral veins and vertebral plexus [16]. However, in most cases, the IJVs are not completely occluded in the up-right posture, and fluid communication is present between the cerebral and central venous systems [17].

The IJV is located in the same sheath as the carotid artery. However, the JVP is very distinct from the carotid artery pulse, differing in shape and timing. The JVP contains three positive waves labeled as “a,” “c,” and “v”. These positive deflections occur, respectively, before the carotid upstroke and just after the P wave of the ECG (a wave), simultaneous with the upstroke of the carotid pulse, just after the S wave of the ECG (c wave); and during ventricular systole until the tricuspid valve opens, after the T wave of the ECG (v wave).

To modulate venous return, we experimented with two well-established clinical tests: abdominojugular and breath holding.

2.1 Abdominojugular test (AJT)

The venous return can be artificially raised by applying pressure to the liver (hepatojugular reflux). Hepatojugular reflux is based on negative intrathoracic pressure, which leads to increased venous return. This technique is used to locate and distinguish the JVP from the carotid pulse. Unlike the carotid pulse, the JVP is impalpable.

To perform the abdominojugular test (AJT), the patient lies at 30 degrees with their head tilted 45 degrees to the left. Then, an oblique light is used to illuminate the jugular region to identify the IJV. The clinician presses firmly on either the right upper quadrant of the abdomen (i.e., over the liver) or over the center of the abdomen [18] for 10 seconds with a pressure of 20-35mmHg while observing the swelling of the IJV in the neck (ensuring that the patient does not perform a Valsalva maneuver) [19].

In a healthy individual, the jugular venous pressure remains constant or temporarily rises for a heartbeat or two before returning to normal, presenting as a lack of jugular vein swelling and therefore a negative abdominojugular reflux test. Negative abdominojugular reflux is also seen in Budd-Chiari syndrome.

A positive result is defined as either a sustained rise in the JVP of at least 3 cm or more [19], or a fall of 4 cm or more [6] after the examiner releases pressure. The test has a wide range in reported sensitivity of 24% [20] to 72% [19], at a specificity of 93% to 96%. The large range in sensitivity may be explained by the optimal conditions of a cardiac lab (72%), as compared to suboptimal in the emergency department (24%).

2.2 Breath-holding

Venous pressure is also impacted by inspiration. Deep inspiration generates negative intrathoracic pressure, leading to an increased venous return. In healthy adults, this phenomenon enhances blood flow to the right heart chambers and causes decreased JV pressure. Thus, the mean venous pressure normally falls during passive inspiration as phasic blood flow occurs in the superior vena cava, and the right ventricle accommodates this increased venous return. When constrictive pericarditis is present, phasic blood flow does not occur in the superior vena cava. Thus, during inspiration, the mean venous pressure rises (Kussmaul's sign) [6].

The human physiological response to breath-holding (BH) is called the “diving reflex,” and the main effects are bradycardia, decreased cardiac output, and increased arterial blood pressure [21]. Bradycardia is induced by increased vagal activity, whereas the peripheral vasoconstriction of selected vascular beds is linked to increased sympathetic discharge [22].

2.3 Protocol

The study consisted of the simultaneous acquisition of rPPG and cPPG waveforms over IJVs during physiological tests. ECG was also acquired as a reference signal.

Two positions (“head up” and “head down”) were explored. In the “head up” (HU) position, the volunteer was lying supine with a 20-degree tilt (head above legs). In the “head down” (HD) position, the volunteer was lying supine with a negative 15 degrees tilt (legs above head).

Two experiments took place in each position: AJT and BH. AJT was performed by a medical doctor for 15 seconds. For BH, the volunteers were instructed to hold their breath for 15 seconds. The experimental duration of 15 seconds was set to be the same for each of BH and AJT for simplicity, and given that that 15 seconds is an appropriate duration for BH (generally well tolerated) and just slightly longer than the standard AJT test (10 seconds).

Each experiment consisted of three segments: baseline, for 15 seconds; experimental, for 15 seconds, during which the physiological test of either AJT or BH were administered; recovery, for 15 seconds. Transition between baseline and experimental took place at the 15-second time point, from experimental to recovery at the 30-second time point, and recovery concluded the data acquisition at the 45-second. Therefore, the total duration of each experiment was 45 seconds. The duration of the baseline and recovery sections were set to the be same as the experimental section.

In total, each participant underwent four experiments (each clinical test in each head position): AJT HU, AJT HD, BH HU, BH HD.

Experimentation took place on two healthy volunteers.

2.4 Data collection

In each position, prior to experimentation, the location of the IJV and carotid arteries on both the left and right sides of the neck were established with ultrasound, and orthogonal diameters measured. The diameters were used to calculate the cross-sectional area of the vessel, assuming an ellipsoid shape. Statistical testing (paired t-test) was used to assess the change in area across relevant scenarios.

rPPG: During each experiment, the left side of the neck was recorded using the camera embedded within a standard smartphone (iPhone). The video was collected at 120 frames per second (maximum supported frame rate) with locked autofocus and auto-white balance. The camera was in a constant position with respect to the subject placement across all acquisitions.

cPPG: The contact PPG sensor was placed on the right (i.e., opposite from rPPG) side of the neck along the IJV, and acquired data at 250 samples per second.

The sampling rates of rPPG (120 samples per second) and cPPG (250 samples per second) were configured to the maximum supported for their respective devices, and are appropriate for the use within this work, as well as enabling future work. In support of that, 50 samples per second has been found to be sufficient for heart rate variability, a typical PPG application [23]. Further, a review from 2020 described the state of the art in smartphone-measured PPG for prediction of blood pressure, either in a contact modality (placing finger against the LED and camera) or contactless, similar to our method herein. The sampling rates ranged from 20-100 frames per second, but given the rate of advancement of smartphone technology, it was possible to exceed even the state of the art with a more recently released device and acquire data our rPPG at 120 frames per second [24].

Simultaneous with cPPG, the ECG signal was collected using a 2-electrode schema, also at 250 samples per second.

The ECG and PPG signals were synchronized through acquisition with a common clinical monitor, passing through filtering on-board the monitor (no additional filtering applied in post-processing). The rPPG acquisition was manually synchronized with the ECG and PPG by initiating the acquisitions at the same time (and therefore, perfect synchronization cannot be guaranteed). After all experiments, the data was extracted as a csv file for further processing.

2.5 Image processing

Two positions of IJV on the left side were found using ultrasound and marked by a marker (see Fig. 1(A)). Blurring was applied to reduce image noise [25]. We selected a band between these 2 points (but excluding them) (see Fig. 1(B)) and extracted average signal from the green channel (selected for highest rPPG signal strength [26] [27]) using this mask. We will refer to it as the rPPG signal thereafter.

 

Fig. 1. A: position of the left IJV found using ultrasound marked on the volunteer's neck with two markings: inferior (left marking) and superior (right marking), respectively. In digital post-processing, inferior has been annotated with a green circle, superior with a red circle. Micromotion is extracted a 30-pixel Euclidean radius from the digital annotations. The annotations have been connected with a line that defines the center of the rPPG mask. B: corresponding rPPG video mask, containing every pixel within a Euclidean distance of 7 from the line.

Download Full Size | PDF

We also extracted the average green-channel (matching rPPG) signals from two marks (inferior and superior signals with respect to position on the neck, respectively). As marks experience micromotions due to neck blood vessel distensions, we refer to these signals as micromotion.

The sole post-processing applied to the rPPG and micromotion signals was the removal of the mean, to facilitate direct comparison of AC information across signal types.

The complete data acquisition and processing flow is shown below in Fig. 2.

 

Fig. 2. Data Acquisition and Processing Flow.

Download Full Size | PDF

2.6 Statistical reliability of remote signals

To quantify the reliability of the extracted rPPG and micromotion signals, the dominant cardiac frequency was compared to cPPG as the ground truth. cPPG was chosen over ECG because cPPG is more similar in origin and morphology to the remote signals than ECG, which measures the electrical activity of the heart.

For cPPG and the remote signals, the continuous wavelet transform was used to derive the time-frequency information from 0.6-10 Hz, with the lower bound set to exclude baseline originating from respiration and other sources not of interest, and the upper bound set to exclude higher-frequency information not relevant to determining the dominant cardiac frequency. Wavelet transform was used to manage the inherent non-stationarity instrinsic in physiological signals [28]. Averaging was then performed over the time dimension to transform to a typical power spectrum, from which the dominant frequency was found as the peak power.

The mean relative error in the detected dominant frequency was calculated across all signals between cPPG, and each of rPPG, micromotion in the superior position and micromotion in the inferior position. Further, paired t-tests were performed to determine if the mean of the differences of the detected dominant frequencies between cPPG and each of the remote signals was significantly different than zero.

3. Results

Diameters of IJV and carotid artery on both sides were measured for each position (head up and down) before starting experimentation in a particular position, from which the cross-sectional area was calculated (assuming ellipsoid shape, product of diameters and π/4). The results of the measurements are presented in Table 1. Across all measurements (subject, position, side), the IJV had a significantly greater area than the carotid (p = 0.006). Further, the carotid area was not significantly impacted by head position (p = 0.33), while the IJV area was significantly increased in the head-down position (p = 0.025).

Table 1. Cross-sectional areas (mm²), calculated from the two diameters (mm) of Carotid Artery (CA) and Internal Jugular Vein (IJV), measured with ultrasound at head-up and head-down positions

View Table

The measured signals displayed changes during the physiological testing and recovery period, with the raw cPPG data collected during experiments depicted in Fig. 3 (Volunteer 1) and Fig. 4 (Volunteer 2). For example, the amplitude and the shape of the cPPG signal changed significantly during Segment 2 (physiological test) of the AJT/HU test in both volunteers.

 

Fig. 3. Contact PPG signal captured from Volunteer 1 during abdominojugular test head up (a) and down (b), breath-holding head up (c) and down (d).

Download Full Size | PDF

 

Fig. 4. Contact PPG signal captured from Volunteer 2 during abdominojugular test head up (a) and down (b), breath-holding head up (c) and down (d). Note the scale on (d) – the recovery segment is absent.

Download Full Size | PDF

Similarly, the raw rPPG and micromotion data collected during the eight experiments is depicted in Fig. 5 (Volunteer 1) and Fig. 6 (Volunteer 2). Each graph contains three sets of data: rPPG (blue) and two sets of micromotion data: inferior (yellow) and superior (red). The amplitude and shape of the cPPG signal changed significantly during Segment 2 (physiological test) of the AJT/HU test in both volunteers.

 

Fig. 5. rPPG (blue), superior (red), and inferior (yellow) micromotion signals captured from Volunteer 1 during abdominojugular test head up (a) and down (b), breath-holding head up (c) and down (d).

Download Full Size | PDF

 

Fig. 6. rPPG (blue), superior (red), and inferior (yellow) micromotion signals captured from Volunteer 1 during abdominojugular test head up (a) and down (b), breath-holding head up (c) and down (d).

Download Full Size | PDF

The comparison of the extracted dominant frequency across cPPG and the remote signals, using cPPG as the ground truth, showed relative errors of 13%, 3% and 14% for rPPG, micromotion in the superior position, and micromotion in the inferior position, respectively. Further, no significant differences were detected using paired t-test: cPPG vs rPPG, p = 0.62; cPPG vs micromotion superior, p = 0.92; cPPG vs micromotion inferior, p = 0.92.

4. Discussion

In this pilot study to identify the possibility of venous blood motion detection in healthy volunteers, we have observed several interesting phenomena.

Firstly, we have found that the cPPG signal over IJV was significantly affected by physiological testing. Moreover, cPPG waveforms exhibited primary features of the JVP waveform (Fig. 7). For example, “a,” “c,” and “v” peaks are clearly identifiable. Moreover, the R peak measured by ECG is located between the “a” and “c” peaks, closer to the “a” peak. These results are in agreement with results observed on the superficial anterior JV [15].

 

Fig. 7. Contact PPG waveform (red) and ECG waveform (blue) during baseline period breath-holding test/head down position/volunteer #2 during baseline (A) and breath-holding exercise (B).

Download Full Size | PDF

However, a closer look at Fig. 7 reveals an even more interesting pattern. Figure 7(a) shows the subject prior to initiation of breath-holding; the characteristic JVP “a” and “c” waves are present (aligned approximately with the P-wave and T-wave, respectively), but the “v"-wave is absent. Figure 7(b) shows the subject during breath holding; the “a” and “c” waves are still present, and the “v"-wave has now appeared in the TP interval. Breath-holding has been demonstrated to have a myriad of physiological effects, including elevation in blood pressure [29], [30]. Further, the “v"-wave represents the passive change in right atrial filling pressure [6], and therefore, it is unsurprising that the appearance of the “v"-wave is provoked by breath-holding, as is observed in Fig. 7.

Thus, it is very plausible that this signal can be attributed to the changes in blood volume in IJV. In fact, our cPPG data collection setup was optimized to capture the signal from the target vessel (IJV). In particular, it had a long source-detector separation (approximately 3 cm). With a known source-detector separation, the sampling depth of the spatially-resolved spectroscopy can be estimated by a common clinical practice rule-of-thumb, which evaluates the sampling depth as ½ of the source-distance separation. Based on this, our sampling depth ∼1.5 cm, and therefore larger than the IJV depth, which was approximately 1 cm based on our ultrasound measurements.

Our video-capturing setup allowed us to study several different imaging methodologies. The most noticeable result is the utility of micromotions detected using marks on the skin, which allowed for reliable data collection. Moreover, the waveforms of these signals are very similar to those captured by cPPG. In fact, the micromotion from the superior marking showing the greatest agreement to cPPG from the perspective of dominant frequency, with a 3% relative error. Thus, it is very plausible that micromotion represent skin displacement caused by IJV distensions.

The first noticeable distinction regarding the micromotion signals is the relative strength of the superior as compared to the inferior. The second noticeable distinction is the difference between head-up and head-down positions. In particular, in the head-up position, one can expect that the IJV diameter around the inferior marking is relatively constant. However, the amount of blood in IJV (and its diameter accordingly) around superior marking significantly oscillates during the cardiac cycle. As cPPG samples a large part of IJV, it also displays similar significant oscillations.

As such, we can conclude that the presented methodology of micromotion detection represents a novel approach for capturing JVP signals. While the approach is limited to capturing just at several locations, it significantly simplifies the data acquisition setup. Thus, it represents a practical trade-off, which can be suitable for certain experiments. Further, the usage of the low-cost camera embedded with a smartphone enables future exploration of remote monitoring applications.

The rPPG signal can most likely be attributed primarily to the microcirculatory bed. In particular, our video-capturing geometry was not optimized to capture the specular reflectance signal (see, for example, [14]). Therefore, the setup captured the rPPG signal primarily. Indirectly, it is confirmed by the small ac/dc ratio of the signal, which is typical for rPPG signals. Thus, it has a minimal sampling depth, limited to the microcirculatory bed. A further confirmation of the microcirculation hypothesis is the noticeable time shift between the rPPG and micromotion signals during the recovery phase in the AJT/HU experiment on volunteer #2 and the BH/HD experiment on volunteer #1, which is typical for time shift between arterial and venous compartments [13]. As such, one can expect that our rPPG and cPPG/micromotion signals correspond to microcirculatory and IJV signals, respectively. Such as these systems are not directly linked; their correlation is relatively weak, as observed in our experiments.

Our findings can be analyzed in the context of physiology of performed tests. In particular, breath-holding is typically preceded by a deep inhalation. Both deep inspiration and AJT are known to cause the negative intrathoracic pressure, resulting in an increased venous return. In healthy adults, this phenomenon enhances blood flow in the right heart chambers and causes decreased JVP. Thus, in the head up (HU) position for both AJT and BH tests, IJV filling is expected to be smaller, which results in smaller IJV diameter. There, since the same volume of blood is drained during the cardiac cycle, a larger relative intracycle variations of IJV diameter is required to accommodate the volume given the reduced average diameter, resulting in higher amplitude cPPG and micromotion signals. This was indeed observed in our experiments.

In the head down (HD) position, it is expected that the IJV will be filled with blood, which was confirmed by measuring the IJV diameters (see Table 1, which show significantly greater cross-sectional area in the head down position). Moreover, this filling (and corresponding diameters) is expected to be minimally affected by the right heart chambers filling. Thus, it is expected that the IJV diameter (and cPPG and micromotion signal, respectively) will be affected by the heart activity and physiological tests to a substantially smaller degree than in the HU position. This was observed in our experiments.

Finally, significant oscillations across the remote signals are apparent in the recovery phase, which can be most likely attributed to CVP changes due to modulations in intrathoracic pressure caused by deep breathing, as the subjects catch their breath after the physiological tests (particularly BH). Note that the absence of similar breathing artifacts in the cPPG signal is caused by cPPG signal filtration in the clinical monitor.

The primary limitation of the current work is the small number of subjects in this pilot study, which does not allow any extrapolation of results. A crucial anticipated effect of this small sample size is that the range of physiological responses to breath-holding and the abdominojugular test cannot be expected to represent the true population. Further, statistical testing revealed significant finding despite the small number of samples, indicating the robustness of the changes detected. However, no a-priori sample size estimation nor powering analysis was performed, and doing so given a larger pool of subjects would enhance the findings. Thus, the immediate focus of future work will be to confirm these findings on a larger pool of subjects to derive statistically significant results. Further, while the acquisition of cPPG and ECG were synchronized (using a common clinical monitor), as were the rPPG and micromotion signals (using the camera), the contact and contactless modalities were not synchronized with each other – we aim to address that limitation in future work. A further enhancement that may be explored in future work is the number and placement of the markings for micromotion detection – given the results herein, an optimal configuration may involve placing markings at very specific locations along the IJV – for instance, the first inferior location that the IVJ is detectible at a specific ultrasound penetration depth, and then following the path IJV to place markings at a defined interval (such as 1 cm) until the final superior location that the IVJ is detectible. With a set of marking such as that outlined, it may be possible to observe a series of micromotion signals. Given that the spacing between the signals is known, it may be possible to calculate wave propagation speed as well as other characteristics.

There are other possible optimizations to consider in the next iteration of data acquisition. Firstly, the use of an adaptive mask that adjusts the size of rPPG and micromotion extraction regions. Secondly, integration of more complex pre-processing and filtering that may enable the observation of subtle signals contained within the data.

5. Conclusion

We proposed a new approach for capturing micromotions caused by neck vessel distensions. Contact PPG and micromotion analysis were able to extract internal jugular vein waveforms and, therefore, have great potential for venous blood flow monitoring. Abdominojugular and breath-holding tests in the head-up position can be used as the model for the modulation of the venous flow. While breath-holding tests display more pronounced results, extending them to animal models will be difficult.