Enanced video heart rate and respiratory rate evaluation: Standard multiparameter monitor vs clinical confrontation in newborn patients

Vital signs in neonatal care are usually measured by a multiparameter monitor (MM). However, this kind of device is inaccurate on the assessment in newborns, which can present different respiratory patterns and superficial breathing, resulting hardly detectable by the MM. In some cases, the assessment of the physician is more effective on the evaluation of the patient. In this paper we present a non-contact solution to measure the respiration rate (RR) and heart rate (HR) using a commercially available web-camera (WeC) and a personal computer (PC). We use Eulerian Video Magnification (EVM) to amplify the color variations in the image sequence. By extracting the signal from the thorax portion of the patients we are able to measure the RR and HR by spectral analysis. The results are compared with the MM and the assessment of the physician. The measures of RR and HR correlate with the data from the MM and result even more accurate than the MM when compared with the physician's evaluation. We collect data on 40 patients demonstrating the feasibility of this method. The measure of RR and HR shows a root mean square error of 6.8 bpm for the HR and 2.1 bpm for the RR