Design of Optical Skin Phantoms for Reflection Photoplethysmography

Abstract

Optical interrogation of tissues can be used for monitoring physiological parameters, such as heart rate or heart rate variability, or for diagnosis of disease. One such measurement modality is the photoplethysmogram (PPG), which is common in medical settings but has recently become popular in the general population in the form of fitness trackers. Unlike medical devices which use red and near-infrared light to obtain PPGs, wearable electronics predominantly use green illumination. These optical heart rate monitors are not subject to United States Food and Drug Administration (FDA) regulations because they are classified as low-risk general wellness products. This presents an opportunity for products to be released without being sufficiently tested. Optical phantoms provide a means to test optical systems under controlled conditions without the unpredictability that physiology can impart. Physical phantom models of blood and skin soft tissue were made with optical agents mixed into water and polydimethylsiloxane (PDMS), respectively, and characterized. Several optical heart rate monitors intended for fitness tracking then underwent testing by being affixed to a skin soft tissue phantom, which had a hollow channel through which the blood phantom was pumped to emulate pulsatile blood flow. The frequency of the pumping waveform was controlled, and the readings from each monitor were compared to this ground truth. A three-axis motion stage was used to test the monitors’ abilities to reject motion artifact. Differences in performance between the monitors were observed, which further highlighted the need for in vitro testing platforms before sending products to market. A second generation of skin phantoms was designed to account for the cutaneous microvasculature, which plays an important role in a reflection PPG measurement due to green light’s shallow penetration depth in tissue. Anatomical models describing the layers of the skin and their individual blood content values were analyzed using Monte Carlo simulations. Then, a model with a simplified layered geometry was described and simulated to determine if it could yield a comparable response to that of the anatomical models when considering various epidermis tones.