Neonatal non-contact respiratory monitoring based on real-time infrared thermography
- Abbas K Abbas†1Email author,
- Konrad Heimann†2,
- Katrin Jergus2,
- Thorsten Orlikowsky2 and
- Steffen Leonhardt1
© Abbas et al; licensee BioMed Central Ltd. 2011
Received: 2 June 2011
Accepted: 20 October 2011
Published: 20 October 2011
Monitoring of vital parameters is an important topic in neonatal daily care. Progress in computational intelligence and medical sensors has facilitated the development of smart bedside monitors that can integrate multiple parameters into a single monitoring system. This paper describes non-contact monitoring of neonatal vital signals based on infrared thermography as a new biomedical engineering application. One signal of clinical interest is the spontaneous respiration rate of the neonate. It will be shown that the respiration rate of neonates can be monitored based on analysis of the anterior naris (nostrils) temperature profile associated with the inspiration and expiration phases successively.
The aim of this study is to develop and investigate a new non-contact respiration monitoring modality for neonatal intensive care unit (NICU) using infrared thermography imaging. This development includes subsequent image processing (region of interest (ROI) detection) and optimization. Moreover, it includes further optimization of this non-contact respiration monitoring to be considered as physiological measurement inside NICU wards.
Continuous wavelet transformation based on Debauches wavelet function was applied to detect the breathing signal within an image stream. Respiration was successfully monitored based on a 0.3°C to 0.5°C temperature difference between the inspiration and expiration phases.
Although this method has been applied to adults before, this is the first time it was used in a newborn infant population inside the neonatal intensive care unit (NICU). The promising results suggest to include this technology into advanced NICU monitors.
Basically, vital signals are physical quantities measured from the body and can be used to determine the physiological status and functioning. Examples of these signals include heart rate, breathing rate, body temperature and blood pressure. The normal range of vital signs varies with age, sex, weight, exercise tolerance and body conditions [1, 2]. Nasal inspiration, the way neonates acquire air and hence oxygen, is important for maintaining the internal milieu of the lung, since ambient air is conditioned to nearly alveolar conditions (i.e. body temperature and fully saturated with water vapor) upon reaching the nasopharynx cavity. Essentially, respiration measurement can be performed by using nasal thermocouples, respiratory-effort belt transducer, piezoelectric transducer, optical sensor (pulse oximetry) and electrocardiography ECG. However, all these techniques are inconvenient to take in at home and they may bring discomfort and soreness to the patient [2–4]. Apnoea (abrupt stopping of respiration) and bradycardia (rapid decrease of heart rate) are common and serious problems in premature infants. One of the methods to quantify respiratory rate in these infants is to use a thermistor that is fixed above the upper lip directly in front of the nares. This by itself can induce apnoeas because of upper respiratory airway obstruction. Therefore, one of important field in such monitoring system is neonatal intensive care unit (NICU), where the patients (neonates) need continuous monitoring of such vital signs (e.g. respiration rate) without creating a discomfort or irritation to them. In principle, optical, electromagnetic, acoustic, and pneumatic techniques can be employed to realize noncontact measurement of physiological quantities. Wang et al.  performed a study on non-contact detection of breathing and heart beat based on radar principles. Similarly, Droitcour et al.  developed a respiratory rate monitoring system using a non-contact, low power 2.4 GHz Doppler radar system and obtained good results when monitoring breathing activities for hospitalized patients. De Chazal et al.  modified a biomotion sensing technique for respiratory activity detection based on 5.8 GHz Doppler radar. Hafner et al.  developed non-contact cardiopulmonary sensing with a baby monitor for premature infants inside neonatal intensive care unit (NICU) by using simple Doppler radars operating in continous wave (CW) mode. Moreover, Zito et al.  developed a wearable system-on-chip (SoC) ultra wide band (UWB) radar for contactless cardiopulmonary monitoring. Matusi  has proposed a novel approach for touchless measurement of heart rate variability (HRV) by using a combination of microwave radar and infrared thermography to analyze the exhaled CO/CO2 gas concentrations. Furthermore, Mathews et al.  also prototyped a contactless vital signal monitor which uses very low power, high frequency Doppler radar to detect the respiration and heart rates. Ling et al.  introduced the OxyArm module, which is a new minimal contact oxygen delivering system for mouth or nose breathing. Moreover, Hoffmann et al.  developed a capacitive textile force sensor for detecting respiration activity rate in the human body. Additionally, Nakajim et al.  employed a real-time image sequence analysis of CCD video camera for evaluating posture changes and respiratory rate of a subject in bed. Moreover, Heimann et al.  investigate a new non-contact monitoring method of heart and lung activity using magnetic induction measurement. In premature infants, a thermistor was used to quantify respiratory flow during inspiration and expiration with the disadvantage of a possible obstruction of the upper airways.
In contrast to the infrared (IR) detectors, which measure the radiation energy emitted from any object containing solid matter, that may represent an option for passive non-contact measurement of vital signs including respiration activity . Resume to that, the skin is the largest organ of the human body and helps maintain the thermal equilibrium of the body and the environment through a heat transfer process.
Infrared thermographic imaging
Target's surface radiation heat exchange
This means that the other two physical quantities (absorbed and reflected radiation) will not be considered in IR thermal imaging.
Material and Methods
Clinical Acquisition of IR Thermograms
All measurements were conducted at the Department of Neonatology (RWTH Aachen University Hospital). This has been approved by the Medical Ethics committee of the RWTH Aachen University Hospital, issued on 19 August 2009 (EK032/09). We examined seven premature infants with a median gestational age of 29 weeks. They were all consecutively admitted directly after birth to our Department of Neonatology. We excluded infants with additional risk factors apart from prematurity, e.g. chromosomal abnormalities or brain haemorrhage. Study design and protocol were approved by the Ethics Committee of Aachen University Hospital and parental consent was obtained prior to enrollment. None of the infants was mechanically ventilated. They all had respiratory support via CPAP (Continuous Positive Airway Pressure) directly after birth because of respiratory distress syndrome, a very common disease in these infants. One of them still had a CPAP during the study. While five infants were handled in an incubator, two infants were positioned in an IR radiant warmer bed. Cardiorespiratory stability was a precondition to be included into the study to get a reliable signal over the whole time period of the IRT. During IRT, vital parameters including oxygen saturation were continuously monitored to make sure that there were no negative side effects. The authors are aware that this setup and this heterogenous patient population can not be a basis for a rigid clinical validation study. Instead, with this paper we are presenting a method for automated data analysis which may eventually lead to a valid ventilation monitoring technique.
IR thermography post-processing
The acquired thermographic images were exported to MATLAB® for post-processing and pre-filtering. The mean value was removed by a moving average filter. Motion compensation was applied using the trend-remove function of the MATLAB System Identification toolbox. The distance between the camera and the subject was kept at less than 150 cm in order to attenuate background IR radiation from the surrounding objects, and to eliminate geometrically induced disturbances. Initially, raw thermal data was used to construct time-varying signals for each thermographic pixel in the area of interest in order to build temperature time profile. However, this approach makes the signal extremely noisy due to variation in the ambient temperature of the surrounding region of interest (ROI) [18–21]. In our work, the thermal images were examined at different points in time. All frames that carry information relevant to respiration in the ROI were selected. Following these steps, the time-varying signals from each point in the ROI were averaged and continuous wavelet transform (CWT) was applied. The resulting waveforms yield excellent results to identify infrared thermography respiration (IRTR) signals [22, 23]. The preliminary tests were conducted in our neonatal intensive care unit (NICU) to identify IRTR signals in neonates; stability over the measurement time intervals has been proven.
Respiration thermal signature
Infrared Thermography Respiration (IRTR) signature detection
The overall equation for heat transfer inside the nasal cavity at the nostril level will be equal to the summation of the following heat flow components:
: rate of heat dissipated by radiation between the air flow and the nasal surface
: rate of heat dissipated by convection between the nasal inner lining (skin and mucosa) and air flow
: rate of heat dissipated by evaporation at the nasal surface (mucosal thin film)
: rate of heat dissipated by blood perfusion
: rate of heat dissipated by latent heat loss through the respiration air flow, where the convective component is negligible in the inspiration thermal signal, due to outflow air with a temperature equal to the tissue temperature [20, 27].
where T 0 is the temperature at initial time, where temperature is minimal and T d is the temperature at final time, when temperature is maximal and α cal is the thermal camera calibration coefficient, which is adjusted according to the clinical thermographic setting.
IRTR signal wavelet analysis
Neonatal respiration monitoring in the NICU
small air-flow jet in the neonatal respiration cycle and small lung volume
the possibility that part of neonatal IRTR signature located at other infrared spectral band, in which the current camera system is unable to detect the full variation in thermal energy
the geometry of the nasal aperture, i.e. the inner surface area of the nasal aperture (nostrils), is small (about 0.08 cm2 as compared with about 1.07 cm2 in the average adult).
the small amount of mucous secretion from the nasal cavity compared to adults
low humidity level in the nostrils region of the neonate
In fact, all factors mentioned above may reduce the amplitude of the IRTR signal. Therefore, some remaining problems need to be addressed in future investigations and further improvements should be performed taking into account the following: a) extending patient population (number of neonates considered in the study) to make more quantitative analysis, b) using more precise IR detector with high temperature resolution.
Results and Discussion
Additional file 1: Neonatal IRTR video sample 1 - Neonatal respiration detection with IR thermography-Video 1. This file is an audivisual file which illustrates the infrared thermography frame-sequence for detecting respiration thermal siganture from the nostrils region. This video file is taken through the preliminary clinical study on neonates cared in radiant warmer (open therapy). (WMV 5 MB)
Additional file 2: Neonatal IRTR video sample 2 -- Neonatal respoiration detection with IR thermography-Video 2. This file is an audivisual file which illustrates the infrared thermography frame-sequence for detecting respiration thermal siganture from the nostrils region. This video file is taken through the preliminary clinical study on neonates cared inside convective incubator system (closed therapy). (WMV 5 MB)
Comparison of mean respiratory rate during 5-minute measurements derived from the IRTR method and from conventional ECG measurement
IRTR respiration derived parameters
RR (IRTR) [bpm]
RR (ECG) [bpm]
Minimum ROI temp [°C]
Maximum ROI temp [°C]
In conclusion, whereas respiration jet and nostril temperature monitoring has been applied to adult volunteers, in this work IR thermography was shown to allow non-contact respiratory monitoring in neonates inside the NICU. Physically, the work is based on changes of convective heat transfer at the infra-nasal region, induced by breathing in dedicated ROIs. Until now the method seems more effective in adults than in newborns, due to the larger lung volumes in adults. Both the IR imaging device and the method itself still face drift problems due to variation in background temperatures. This requires improvements in image processing and boundary detection of the nasal region separated from the rest of the imaging scenario (e.g. incubator internal wall, mattress, and other facial regions). Moreover, the presented results are preliminary and need further studies in a larger number of neonates and under different care setups. The mathematical method needs further improvement, such as automatic ROI definition and automatic calibration. Furthermore, the IRTR monitoring may assist in the estimation of a possible temperature loss as a part of thermoregulation, and may be also considered as a first step to evaluate non-invasive respiratory behaviour of premature infants . In comparison to the ECG derived respiration rate, the IRTR signal is correlated to this acquired signal from bedside monitor, while there is a slight difference in respiration rate estimated from each method (see Table 1). The main impediments to high resolution IRTR signature detection are the IR camera physical coverage and the thermal detector's resolution. In spite of the calibration mechanism in modern thermal cameras, most of medical IR imaging setups face calibration drift; and this needs enhancement in order to avoid erroneous measurement. To deal with this problem, the proposed solution is based on a virtual sensing mechanism to track the ROI over a defined anatomical part. Moreover, the information on intensity was extracted and transformed into a corresponding color-coded space of IR thermographic images.
However, despite the limited number of measurements, these preliminary results provide a good basis for further investigation of the neonatal thermal respiration signature. More studies performed under standardized clinical conditions are needed, so that the method can be applied to examine the symmetrical pattern of the IRTR signature. Moreover, clinical investigations should explore a range of upper respiratory tract diseases. Furthermore, this method may be an effective quantitative technique to measure the nasal symmetrical air-flow pattern in preterm infants. This possibly can give information about the depth and the frequency of each breath cycle to get an early sign of changes in the infant's behavior. Also, the temperature difference up to 0.66°C can be interpreted as a part of the infant's thermoregulation and gives an interference of heat loss through expiration. Despite the remaining problems, the authors feel that the presented technique is a promising and effective step toward establishing cable-free monitoring of infants under intensive care conditions.
Statement of consent
An oral consent was gained from the parents of the patient for publication of images and related files.
We hereby express our thanks to Prof. V. Blazek and Prof. V. J. Kumar for their valuable recommendation and reviewing of this paper. All experimental device (IR camera and temperature sensors were provided by MedIT, RWTH Aachen University. In addition, we thank all the medical staff in the Departement of Neonatology at University Hospital, RWTH Aachen University, for their tolerance and support during the study and clinical measurements.
- Murthy JN, van Jaarsveld J, Fei J, Pavlidis I, Harrykissoon R, Lucke : Thermal infrared imaging: A novel method to monitor airflow during polysomnography. SLEEP 2009, 32: 15211527.Google Scholar
- Wang JQ, Wang HB, Jin XJ, Yang GS, Yang B, Dong XZ, Qiu LJ: The study on non-contact detection of breathing and heartbeat based on radar principles. Fourth Medical Military conference 2001, 25: 132–135.Google Scholar
- de Chazal P, O'Hare E, Fox N, Heneghan C: Assessment of sleep/wake patterns using a non-contact biomotion sensor. In Conf Proc IEEE Eng Med Biol Soc 2008, 33: 514–517. IEEEGoogle Scholar
- Matthews G, Sudduth B, Burrow M: A non-contact vital signs monitor. Crit Rev Biomed Eng 2000, 28(1–2):173–178.View ArticleGoogle Scholar
- Droitcour AD, Seto TB, Park BK, Yamada S, Vergara A, Hourani CE, Shing T, Yuen A, Lubecke VM, Boric-Lubecke O: Non-contact respiratory rate measurement validation for hospitalized patients. Conf Proc IEEE Eng Med Biol Soc 2009 2009, 24: 4812–4815.View ArticleGoogle Scholar
- Hafner N, Mostafanezhad I, Lubecke VM, Boric-Lubecke O, Host-Madsen A: Non-contact cardiopulmonary sensing with a baby monitor. Conf Proc IEEE Eng Med Biol Soc 2007, 12: 2300–2302.Google Scholar
- Zito D, Pepe D, Mincica M, Zito F, Rossi DD, Lanata A, Scilingo EP, Tognetti A: Wearable system-on-a-chip UWB radar for contact-less cardiopulmonary monitoring: present status. Conf Proc IEEE Eng Med Biol Soc 2008, 14: 5274–5277. IEEEGoogle Scholar
- Matsui T, Hattori H, Takase B, Ishihara M: Non-invasive estimation of arterial blood pH using exhaled CO/CO2 analyzer, microwave radar and infrared thermography for patients after massive hemorrhage. J Med Eng Technol 2006, 30: 97–101. 10.1080/03091900500062158View ArticleGoogle Scholar
- Ling E, McDonald L, Dinesen TJR, DuVall D: The OxyArm - a new minimal contact oxygen delivery system for mouth or nose breathing. Can J Anaesth 2002, 49(3):297–301. 10.1007/BF03020531View ArticleGoogle Scholar
- Hoffmann T, Eilbrecht B, Leonhardt S: Respiratory monitoring system on the basis of capacitive textile force sensor. IEEE transaction of sensors 2010, 27.Google Scholar
- Nakajim K, Matsumoto Y, Tamura T: Development of real-time image sequence analysis for evaluating posture change and respiratory rate of a subject in bed. Physiological Measurement 2001, 22(2):N21-N28.View ArticleGoogle Scholar
- Heimann K, Steffen M, Bernstein N, Heerich N, Stanzel S, Cordes A, Leonhardt S, Wenzl TG, Orlikowsky T: Non-contact monitoring of heart and lung activity using magnetic induction measurement in a neonatal animal model. Biomed Tech (Berl) 2009, 6(54):337–345.View ArticleGoogle Scholar
- Holst G: Common Sense Approach to Thermal Imaging. SPIE Press, JCD Publishing; 2000.Google Scholar
- Maldague X: Theory and practice of infrared technology for nondestructive testing. Volume 2. Wiley; 2001.Google Scholar
- Kaplan H: Practical applications of infrared thermal sensing and imaging equipment. Tutorial texts in optical engineering, SPIE Press; 2007.View ArticleGoogle Scholar
- Hudson R: Infrared Systems Engineering. John Wiley & Sons, Inc, New York; 1969.Google Scholar
- F K: The CRC Handbook of Thermal Engineering. CRC Press, Boca Raton, FL; 2000.Google Scholar
- Pavilidis I, Levine J: Thermal image analysis for polygraph testing. IEEE Engineering in Medicine and Biology Magazine 2002, 21(6):56–64. 10.1109/MEMB.2002.1175139View ArticleGoogle Scholar
- Tarkov MS, Vainer BG: Evaluation of a Thermogram Heterogeneity Based on the Wavelet Haar Transform. Proc Siberian Conference on Control and Communications SIBCON '07 2007, 145–152.Google Scholar
- Murthy R, Pavlidis I: Non-contact monitoring of breathing function using infrared imaging. Technical report in Biomedical engineering Number UH-CS-05–09, University of Houston; 2005.Google Scholar
- Fei J, Pavlidis I, Murthy J: Thermal vision for sleep apnea monitoring. Med Image Comput Comput Assist Interv 2009, 12(Pt 2):1084–1091.Google Scholar
- Caniou J: Passive infrared detection: theory and applications. Volume 2. Kluwer Academic Publishers; 1999.View ArticleGoogle Scholar
- Fei J, Pavlidis I: Virtual thermistor. Conf Proc IEEE Eng Med Biol Soc 2007, 250–253.Google Scholar
- Fei J, Pavlidis I: Thermistor at a distance: unobtrusive measurement of breathing. IEEE Trans Biomed Eng 2010, 57(4):988–998.View ArticleGoogle Scholar
- Naftali S, Schroter RC, Shiner RJ, Elad D: Transport Phenomena In The Human Nasal Cavity: A Computational Model. Annals of Biomedical Engineering 1998, 26(5):831–839.View ArticleGoogle Scholar
- Abbas AK, Heimann K, Orlikowsky T, Leonhardt S: Non-Contact Respiratory Monitoring Based on Real-Time IR-Thermography. In World Congress on Medical Physics and Biomedical Engineering,, IFMBE Proceedings Edited by: IFMBE. 2010., 25: Google Scholar
- Fei J, Pavlidis I: Thermistor at a distance: unobtrusive measurement of breathing. IEEE Trans Biomed Eng 2010, 57(4):988–998. [http://dx.doi.org/10.1109/TBME.2009.2032415]View ArticleGoogle Scholar
- Danjoux R: The evolution in spatial resolution. InfraMation Magazine 2001, 2(12):1–3.Google Scholar
- Akay M, Mello C: Wavelets for biomedical signal Processing. In Proceedings of the 19th annual international conference of the IEEE Edited by: engineering in medicine, biology society. 1997., 6: Google Scholar
- Stark H: Wavelets and signal processing: an application-based introduction. Springer; 2005. [http://books.google.com/books?id=IYUMyVRdJjkC]Google Scholar
- Bohnhorst B, Heyne T, Peter C, Poet sC: Skin-to-skin (kangaroo) care, respiratory control, and thermoregulation. J Pediatr 2001, 138: 193–197. 10.1067/mpd.2001.110978View ArticleGoogle Scholar
- Bohnhorst B, Gill D, Drdelmann M, Peter C, Poets C: Bradycardia and desaturation during skin-to-skin care: no relationship to hyperthermia. J Pediatrics 2004, 145: 499–502. 10.1016/j.jpeds.2004.06.019View ArticleGoogle Scholar
- Heimann K, Vaen P, Peschgens T, Stanzel S, Wenzl T, Orlikowsky T: Impact of skin-to-skin care, prone and supine positioning on cardio respiratory parameters and thermoregulation in premature infants. Neonatology 2010 2010, 97: 311–317.Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.