Reconfigurable Point-of-Care System for Hemoglobin Estimation From Photoplethysmogram

Abstract

Hemoglobin is an integral part of blood, and its abnormality indicates various diseases. Different noninvasive methods are developed to predict the concentration of hemoglobin. Previous studies verified the potential of photoplethysmogram (PPG) signals in estimating the health parameter. Although different hardware tools have been used to develop digital systems over the years, they lack the reconfigurability feature needed to develop a point-of-care (POC) system. In this study, a field programmable gate array (FPGA)-based reconfigurable hardware system, including preprocessor, memory and control, feature extractor and classifier subsystems, is designed targeting Zynq 7000 Zedboard. The system utilizes six features extracted from the PPG signals collected using DCM08 PPG sensor and linear regression classifier model for prediction. PPG signals based on four different wavelengths of light are tested, and the best result has been achieved with infrared light having a wavelength of 940 nm, which will help to design PPG sensors for wearable and medical devices. The mean absolute error with this wavelength is 2.55 g/L with an error rate of 1.78%. The power consumption analysis validates the designed system to be a low-power device. The designed processor can be used as a POC system, and due to its reconfigurable advantage, the system can be further improved by adding other health parameter predictions and disease detection.