Non-invasive calibration method for pulse oximeters

Abstract

In case of a healthy subject the normal SpO₂ value is 97 ± 2% on see level. Modern, finger probe based pulse oximeters are measuring the SpO₂ level with 1-2% error. The dispersion between subjects can reach 4%, thus such accuracy is not really demanded by the majority of clinicians. Moreover, in case of fetal pulse oximetry 5% measuring error is accepted. Considering these factors we investigated the feasibility of a non-invasive calibration method with a self-developed pulse oximeter. This method is carried out without blood sampling. Pulse oximeters are measuring the R rate, which is proportional to the SpO₂ value. Calibrating an oximeter means finding the function between the R and SpO₂. A calibrated pulse oximeter was used as reference. In the case of every subject 15 minutes long measurements were performed. The reference device and our oximeter were attached to the subject at the same time, while artificial air with 14% oxygen content was inhaled by the subject for ten minutes. The SpO₂ was measured by the reference oximeter and the R rate by our oximeter. Based on 511 measured data pairs the relationship was determined between 86-100%. The relationship was estimated by linear regression. Although the original relation is non-linear, linear estimation can be used in this small range of SpO₂ with good accuracy. The average error of the calibrated device is 2.76%, which is appropriate in medical practice. This method is easier and cheaper as the invasive calibration, but the calibrated device will have slightly bigger measuring error.