Abstract
Blood pressure (BP) is an important parameter of human health, since hypertension is a major risk factor of the cardiovascular system. Nowadays, the continuous measurement of BP is only possible with an invasive measurement method using catheter as the gold standard. Diffuse correlation spectroscopy (DCS) is a powerful tool for assessing microvascular hemodynamics in deep tissues, which can reliably provide a blood flow index whose changes are proportional to relative changes in tissue blood flow (BF). This study presents a new solution that enables to use tissue BF measured by our developed DCS device for continuous BP monitoring via deep learning approach. We evaluated the utility of tissue BF for continuous BP estimation via a proposed BFBP model. In vivo experiments (12 subjects) were performed to collect tissue BF and continuous BP data simultaneously to verify the feasibility of BFBP model. The mean absolute errors of the continuous BP estimates were 5.54 ± 5.03 mm Hg for systolic BP, 1.71 ± 2.86 mm Hg for diastolic BP, indicating that DCS provides a novel way for continuous BP estimation. Moreover, we compared the proposed BFBP model with other models based on an open BP dataset from UC Irvine database. The experimental results indicated that the estimated continuous BP achieves grade A for both systolic blood pressure and diastolic blood pressure according to the British Hypertension Society standard. Ultimately, experimental results show that the proposed method enables feasible continuous estimation of BP using noninvasive tissue BF measured by our developed DCS device based on the proposed BFBP model.