Camera-based Blood Oxygen Imaging for Lower Limb Arterial Stenosis Diagnosis

Peripheral arterial disease (PAD) is a significant health problem that adversely affects the quality of life for many elders. Lower limb arterial stenosis (LLAS) is the primary manifestation of PAD. In this study, we applied pressures to the calves for PAD simulation, recording the plantar responses through camera-based photoplethysmography (PPG) signals. Subsequently, we explored camera measured blood oxygen saturation (SPO2) ("Ratio of Ratios" (RR) as surrogate) as an indicator to investigate the feasibility of classifying different PAD severities. RR is an intermediate feature for SpO2 calibration, which is calculated from two wavelength setups: (1) R and G channels, and (2) R and infrared channels. With RR as features, and perfusion index (PI) as the benchmarking candidate, we trained an LLAS detection model with two classification levels and an LLAS severity grading model with five classification levels. The detection and grading models trained by RR achieved the accuracies of 82% and 51%, respectively. These results highlight the potential of SpO2 as an indicator for fast camera-based PAD screening.