Cardiovascular disease (CVD) is one of the main causes of death globally. CVD has been associated to pulse wave velocity (PWV), which is the measurement of artery wall stiffness or elasticity based on the speed at which pressure waves travel through arteries. Because PWV rises with arterial stiffness, it is a useful tool for early diagnosis, detection, and screening of CVD. Unfortunately, existing methods for measuring PWV are invasive and expensive, while non-invasive ultrasound and ECG techniques have fallen short as both are susceptible to errors due to noise or artifacts.
The experimental setup captured images of the subject's neck for analysis. Image Credit: Image courtesy of Toronto Metropolitan University
Scientists at Toronto Metropolitan University (Toronto, ON) are aiming to resolve the drawbacks of current PWV measurement techniques. They've developed a non-contact approach employing a Mikrotron high-speed camera for imaging of neck vessels. In doing so, they hope to expand the potential of remote photoplethysmography (rPPG), a non-invasive optical technique that measures changes in blood volume.
Experimental Setup Monitors Neck Vessels
Leveraging rPPG, the scientists measured PWV by capturing images of the carotid artery and jugular vein of multiple human test subjects. The Mikrotron EoSens 1.1CXP2 CoaXPress camera was placed close to subjects' necks without touching it, acquiring signals from the vessels via RGB channels. The experimental setup called for each subject to sit straight with their chin tilted to fully expose the left side of their necks. Two 600-lumen LED light panels ensured even illumination in the neck region. In addition, green, blue, and red light was balanced at a light temperature of 4600 K. Light intensity was set at 100% to prevent underexposure.
The Mikrotron camera model was equipped with a Canon EF 50 mm f/2.5 STM lens and mounted on a 60-inch tripod. The EoSens 1.1CXP2 camera features a four-lane CXP-12 interface capable of 12.5 Gbps per link for extremely fast data transfer rates. Imaging frequency used in this study was 2000 frames-per-second (fps) although the camera can achieve rates of up to 3660 fps at 1.1 megapixel resolution.
In the study, the neck of each subject was recorded in 10-second videos at a frame size of 512 × 312 pixels. The ten participants were between the ages of 18 and 60 with no known cardiovascular-related diseases. After being acquired, the videos were converted into a 4D dataset that contained time, color channels, frame height, and width.
MATLAB 2022b software applied its demosaic function to decode them into precise RGB color pictures. Gaussian blur further reduced noise for smoother images. The green channel was selected for PPG signal analysis due to its stronger signal strength and greater stability. Additionally, green light penetrates deeper into the skin than blue light, where blood vessels are more evenly distributed.
Signal peaks were chosen using MATLAB's "find peaks" function supporting pulse rates ranging from 50 to 110 beats per minute. The PPG signal's periodicity, which must coincide with the heart cycle for accurate analysis and guarantee that waveform peaks are constant, is the basis for signal selection. Reference signals were selected from the average signals of the 40 × 40 pixel clusters inside the ROI.
Conclusion
Using a Mikrotron camera and the rPPG approach, the Toronto Metropolitan University team developed a non-contact method of detecting PWV distribution in the neck vessel. The proposed technique can benefit clinical screening by identifying individuals at high risk of CVD. As a non-contact method, it also enhances patient comfort and holds promise for wider clinical use, offering a more accessible approach to managing long-term cardiovascular health.