Infrared thermography based human respiration monitoring

Abstract

Infrared thermography (IRT) has evolved as an important biomedical tool in recent years. One major application of IRT is the reliable monitoring of human respiration rate (RR) in a contactless manner. This method is especially useful in case of babies with delicate skin. The present work reports the human RR monitoring using passive IRT, by observing the variation in nasal temperature, during breathing. The observed breathing signal has a low signal to noise ratio (SNR), hence it is denoised using the Infinite Impulse Response (IIR) filters. The IIR filters are compared based on their SNR and Mean Square Error values. The Butterworth filter shows the best filtering performance amongst all the IIR filters, which further improves with increasing filter order. A novel “Breath detection algorithm (BDA) is designed, that identifies the breaths in the acquired breathing signals as normal or abnormal, and yields the breaths per minute value, in an automated manner. The BDA is tested on 500 breathing signals under different scenarios like normal, slow and fast breathing, and with and without air conditioner and fan. The BDA performance is evaluated by calculating its sensitivity, precision, spurious cycle rate, and missed cycle rate values obtained as 98.4%, 99.19%, 0.80% and 1.6% respectively.