Pulse Oximeter Signal Modeling and Fusion for Hypoxia Monitoring Sayandeep Acharya * , Arjun Rajasekar * , Barry S. Shender † , Leonid Hrebien * and Moshe Kam * * Department of Electrical and Computer Engineering Drexel University, Philadelphia, Pennsylvania 19104, USA Email:{sa427@, ar924@, lhrebien@ece., kam@minerva.ece.}drexel.edu † Human Systems Department Naval Air Warfare Center Aircraft Division, Patuxent River, MD 20670, USA Email: barry.shender@navy.mil Abstract—We develop models and fusion rules for oximeters that detect the onset of hypoxia. Hypoxia is a medical condition affecting portions of the body that are deprived of oxygen supply. Prolonged exposure to cerebral oxygen deficiency can lead to unconsciousness or even death. The onset of hypoxia in humans is of concern for those operating in high altitudes, and in military flights characterized by high-acceleration maneuvers. Using oximeters for measuring blood oxygen saturation levels is a common means to detect hypoxia in real time. Many types of oximeters can be used for this task but all are prone to complicated noise characteristics and bias inaccuracies. It may therefore be advisable to collect and combine data streams from multiple oximeters for more reliable Hypoxia/No Hypoxia decisions (compared to decisions made by a single oximeter). Here we develop statistical noise models for three popular types of oximeters (Respironics Novametrix 515B, Nonin forehead pulse oximeter 9847, and Masimo Rad-87). We also combine data streams from these oximeters using a Kalman filter. The result is a smooth and reliable estimate of blood oxygen saturation level which can be used to detect the onset of Hypoxia. Keywords: Hypoxia Monitoring, Sensor Fusion, Kalman Filter, Colored Noise, Pulse Oximeters. I. I NTRODUCTION Hypoxia is diminished availability of oxygen to the cells of the body [1]. It can occur due to inadequate oxygenation of the lungs for extrinsic reasons, deficiency of oxygen in atmosphere, venous-to-arterial shunts (intrapulmonary or intra cardiac), inadequate transport and delivery of oxygen, or inadequate tissue oxygenation or oxygen use. Exposure to severe hypoxia can lead to death of cells and depressed mental activity. Sometimes it culminates in coma and reduced work capacity of the muscles. Hypoxia occurs most commonly in people traveling to high altitude, performing strenuous exercise or work for prolonged periods of time at high altitudes. Another population at risk is combatants such as fighter pilots who undertake high G maneuvers. Measuring the blood oxygen saturation (SpO2) is the most common and easiest way to instrumentally determine the presence of hypoxia. A healthy human has on average a SpO2 value of 95-100%. SpO2 values below 90% are considered low, and are taken as a possible indication of onset of hypoxia. The most common non-invasive device used to measure blood oxygen saturation levels is the pulse oximeter. The device uses a photo detector to measure the difference in the extinction curves of hemoglobin and oxygenated hemoglobin using light of different wavelengths [2], [3]. The common types of oxime- ters are applied either on the finger or on the forehead of the subject being monitored. Hypoxia monitoring has been reported in several previous studies. A hypoxia detection and warning system was patented as a Aviation Hypoxia Monitor [4], which has a single pulse oximeter attached to the ear and provides a visual and audio signal if the blood level of a subject decreases significantly. The Hypoxia Detection and Warning System in [5] is com- posed of an electrochemical oxygen sensor located within the breathing mask of a pilot. It provides a vibratory warning within the mask when partial pressure of oxygen in the system falls below a set point. In [6], a personal hypoxia monitoring system is proposed which uses the cross-correlation between heart rate, respiratory rate, blood flow velocity and blood oxygen saturation levels to identify the onset of hypoxia. Even though pulse oximeters are very popular in oper- ating rooms, emergency medical aids, and ambulatory use by heart and respiratory-system patients, oximeters are prone to inaccuracies due to several sources, most notably light scattering inside blood tissues. They are also affected by noise artifacts due to motion, ambient light interference, respiratory maneuvers, and pooling of blood at the point of measurement due to body orientation. In situations where fast and reliable hypoxia detection is required, a single pulse oximeter may not be sufficient, and it may be advantageous to use a combination of several such devices. In a study conducted at the Naval Air Warfare Center Aircraft Division (NAWCAD) [7], three oximeters from dif- ferent manufacturers were used simultaneously. These were Respironics Novametrix 515B (transmittance type on finger), and two reflectance type oximeters - Nonin pulse oximeter 9847 and Masimo Rad-87 (used on the forehead). The current study describes an attempt to fuse their observations using Kalman filtering so as to obtain a smoother and more reliable estimate of the blood oxygen saturation level than what one can get from a stand-alone single oximeter. The algorithm we propose can be executed in real time and has moderate computational requirements (computations can be carried out