Physiological Measurement
Related content
PAPER
Response characteristics for thermal and pressure devices commonly used for monitoring nasal and oral airflow during sleep studies To cite this article: J M Gehring et al 2014 Physiol. Meas. 35 455
View the article online for updates and enhancements.
- Sensitivity and specificity of hypopnoea detection using nasal pressure in the presence of a nasal expiratory resistive device (Provent®) Stephen Milne, Terence C Amis, John R Wheatley et al. - Quantitative flowmeter for polysomnography J P Kirkness, M Verma, B M McGinley et al. - Sleep apnea: a review of diagnostic sensors, algorithms, and therapies Mehdi Shokoueinejad, Chris Fernandez, Emily Carroll et al.
Recent citations - Lala Li et al - Sensitivity and specificity of hypopnoea detection using nasal pressure in the presence of a nasal expiratory resistive device (Provent®) Stephen Milne et al
This content was downloaded from IP address 134.157.146.58 on 16/05/2018 at 15:19
Institute of Physics and Engineering in Medicine Physiol. Meas. 35 (2014) 455–470
Physiological Measurement
doi:10.1088/0967-3334/35/3/455
Response characteristics for thermal and pressure devices commonly used for monitoring nasal and oral airflow during sleep studies J M Gehring 1 , J-G Cho 1,2 , J R Wheatley 1,2 and T C Amis 1,2 1 Ludwig Engel Centre for Respiratory Research, Westmead Millennium Institute, Westmead, NSW, Australia 2 University of Sydney at Westmead Hospital, Westmead, NSW, Australia
E-mail: [email protected] Received 5 November 2013, revised 17 January 2014 Accepted for publication 28 January 2014 Published 20 February 2014 Abstract
We examined thermocouple and pressure cannulae responses to oral and nasal airflow using a polyester model of a human face, with patent nasal and oral orifices instrumented with a dual thermocouple (F-ONT2A, Grass) or a dual cannula (0588, Braebon) pressure transducer ( ± 10 cm H2O, Celesco) system. Tidal airflow was generated using a dual compartment facemask with pneumotachographs (Fleisch 2) connected to the model orifices. During nasal breathing: thermocouple amplitude = 0.38 Ln [pneumotachograph amplitude] + 1.31 and pressure cannula amplitude = 0.93 [pneumotachograph amplitude]2.15; during oral breathing: thermocouple amplitude = 0.44 Ln [pneumotachograph amplitude] + 1.07 and pressure cannula amplitude = 0.33 [pneumotachograph amplitude]1.72; (all range ∼0.1–∼4.0 L s−1; r2 > 0.7). For pneumotachograph amplitudes 0.7. For the pressure cannula, the data were best fitted using a power function; for nasal breathing P = 0.93 V2.15 (figure 3(C)) and for oral breathing P = 0.33 V1.72 (figure 3(D)); both r2 > 0.9. In comparison with a theoretical linear response (line of identity; see figure 3) both nasal and oral TS overestimated V at lower V ( 0.7). For the nasal pressure cannula: P = 1.11 V − 0.33, while for oral breathing: P = 0.29 V − 0.04 (both r2 > 0.7). In this case the response values (i.e. dTS/dV and dP/dV) are the fitted line slopes. Thus, while the response for nasal and oral V is similar for the thermal based system, the response of the pressure based system for the detection of nasal V was almost four times that for oral V. 461
J M Gehring et al
Physiol. Meas. 35 (2014) 455
(A)
(B)
Figure 5. Scatter plots showing breath-by-breath values (filled circles represent data from nasal sensors, open circles data from oral sensors) for a restricted ( 0.25 (figure 8(D)), but not for nasal breathing or for pressure cannulae signals. ∗ p < 0.05, r > 0.9.
2.0) were performed for both the nasal and oral routes with target flow amplitudes between 0.2 to 1.0 L s−1. Breath-by-breath amplitudes for TS and P were plotted against the corresponding values for V. Differentiation of mathematical expressions fitted to TS or P versus V was used to calculate dP/dV and dTS/dV and device sensitivities were calculated at a flow amplitude of 0.5 L s−1 (figure 8). For both oral and nasal routes of breathing, neither dP/dV nor dTS/dV correlated with TI/TE at any Ttot (Pearson’s correlation, all p > 0.05). Similarly, no significant correlations were found between dP/dV or dTS/dV and Ttot for any TI/TE except for dTS/dV for oral breathing at TI/TE > 0.25 (r2 > 0.9, p < 0.05). For example, at TI/TE = 1.0, dTS/dV was 1.1 volts/L s−1 for Ttot = 5 s and 1.6 volts/L s−1 for Ttot = 10 s, i.e. a lower Ttot resulted in lower dTS/dV. 4. Discussion In this study of the response characteristics of both thermal and pressure based sensor systems for the detection of nasal and oral route respiratory airflow, the major findings were: (1) the thermocouple system was more responsive at lower airflows than at higher airflow, while the reverse applied to the dual cannula pressure based sensor system, (2) there was no cross contamination from one breathing route to the other for either sensor systems, (3) nasal 465
J M Gehring et al
Physiol. Meas. 35 (2014) 455
and oral route sensor response characteristics differed for both systems, (4) oro-nasal airflow partitioning may not be accurately reflected by either system, (5) changes in oral orifice crosssectional area had opposite effects on the response characteristics of thermal versus pressure based oral sensors, and (6) altering the pattern of breathing only affected the oral thermocouple sensitivity. Several previous studies have examined the relative response characteristics of nasal pressure cannulae and thermally responsive sensors for the detection of apneas and hypopneas (respiratory events) during sleep (S´eri`es and Marc 1999, Norman et al 1997, Thurnheer et al 2001, Heitman et al 2002). In general these studies have found that thermistors/thermocouples do not detect as many events as are identified by nasal pressure recordings. Moreover, reductions in thermal based signals associated with falls in tidal volume are less than those recorded with other monitoring systems such as inductive plethysmography (S´eri`es and Marc 1999, Pasto et al 1996). In addition, thermally responsive monitoring systems typically have long time constants making them insensitive to the shape of the inspiratory airflow profile and, therefore, are not suitable for evaluating inspiratory airflow limitation (Montserrat et al 1997). These findings have led to recommendations that non-invasive nasal airflow monitoring during sleep is best accomplished using nasal pressure measurements rather than thermistors/thermocouples (American Academy of Sleep Medicine 1999). Heitman et al (2002) provided evidence validating the detection of apneas and hypopneas during sleep with nasal pressure cannulae in comparison with ‘gold standard’ measurement using a pneumotachograph. Most studies that have compared thermal and pressure based airflow monitoring systems for polysomnography have concentrated on the detection of sleep disordered breathing events (S´eri`es and Marc 1999, Norman et al 1997, Heitman et al 2002). Few studies have quantitatively examined the relationship between ‘gold standard’ measures and surrogate variable signal magnitude. Pneumotachograph measurement of airflow is such a ‘gold standard’ methodology but requires the use of a face mask. Since such instrumentation of the face is likely to alter the conditions of operation for the surrogate systems this approach has not been favored, especially for the thermal based systems. Attempts have been made to compare surrogate signals with other measures of airflow. Pennock (1992) compared thermal device responses with those from a ‘piezo-electric belt’ showing that disagreement between these airflow monitoring systems was related to individual device design. The relative response characteristics for the quantification of oral versus nasal airflow were not examined. Berg et al (1997) compared minute ventilation measured directly by means of a head-out body plethysmograph with measurements of nasal ventilation as detected by thermal, pressure and inductive plethysmographic monitoring systems. They concluded that all these indirect methods were in poor agreement with plethysmograph measured ventilation in awake subjects. Other attempts to quantitatively evaluate surrogate variable measures of airflow have been performed in physical models of the upper airway. Farr´e et al (1998) described a heated chamber model through which cyclical airflow (measured via a pneumotachograph) could be generated by a servo-controlled bellows. This model was used to test the response characteristics of a thermistor placed in front of an artificial nose (two cylindrical tubes built into the front wall of the chamber). Tested over a peak-to-peak airflow amplitude (sinusoidal airflow) of 0 to 1.0–1.5 L s−1 the response of the themistor was nonlinear and depended on the airflow pattern, the distance of the thermal probes from the ‘nose’ and the cross-sectional area of the ‘nose’. Fleury et al (1996) describe a model consisting of a polystyrene ‘head’ with two tubes for nostrils but do not provide any data. We used a physical model of the face in an attempt to reproduce the operating conditions for the thermistor/thermocouple and pressure monitoring systems when used on an unencumbered 466
J M Gehring et al
Physiol. Meas. 35 (2014) 455
human face. Our physical model was constructed to reflect the anatomical relationships of the human face and to reproduce the relative dimensions of the nasal passages, especially the nasal valve and nasal vestibule region. Unlike previous models our model also incorporates an oral pathway. In addition, the use of modeling compound to define the mouth of the model allowed the effect of oral orifice geometry to be studied. The approach of having a human subject, connected to the model via a dual compartment face mask and breathing with feedback from dual pneumotachographs, allowed target peak airflows to be achieved and exposed the sensor systems to human exhaled gas. Limitations of the model include the rigid nature of its surface, the lack of a background body temperature (thermal systems applied to the human face operate with a background of skin surface temperature), potential loss of heat and water content of exhaled gas in the face mask and tubing systems and the lack of any upper airway muscle activity. In addition, although the lips and oral orifice were modeled, the oral cavity and pharyngeal regions were not. It is well recognized that measurement of total respiratory airflow using nasal route monitoring is subject to error if mouth breathing occurs (Norman et al 1997, Thurnheer et al 2001). For this reason, polysomnographic sleep monitoring systems often include a thermal or pressure based sensor positioned at the lips (S´eri`es and Marc 1999). While there has been considerable attention paid to the accuracy of surrogate variable measurements of nasal airflow very little attention has been paid to the accuracy of thermal and pressure based systems in the assessment of oral airflow. There are grounds to suspect that there might be differences in the sensitivity of these systems to mouth versus nose breathing. For example, oral exhaled gas may be higher in water content and temperature than nasal exhaled gas, and oral geometry (degree of mouth opening) can vary over a much wider range for the mouth than for the nose. Differences in airflow regimes (e.g. turbulence) will influence local pressure measurements. Additionally, when such measurements are being made to assess the degree of oro-nasal airflow partitioning the relative response characteristics of the same sensor system applied to the nasal versus the oral pathway become critical. For example, even if the surrogate variable under study does not accurately reflect airflow, if the response characteristics for mouth and nose airflow are similar the system may still be capable of accurately reflecting the relationship between nasal and oral airflow (e.g. nasal/oral signal ratio). However, no previous studies have examined the relative nasal versus oral response characteristics for thermal or pressure based surrogate variable airflow detection systems. As has been demonstrated previously (Farr´e et al 1998), and in the present study, over the entire range of flow rates studied both the nasal and oral thermocouple and pressure sensing systems were not linearly related to the pneumotachograph measurements. The thermocouple signals were best represented by a logarithmic function characterized by overestimation of changes in pneumotachograph signal amplitude (when compared with a linear response) for amplitudes below ∼1.4 L s−1 for nasal breathing and ∼1.1 L s−1 for oral breathing (see figure 3). In contrast the pressure cannulae response was best represented by a power function characterized by similar changes in surrogate variable and pneumotachograph signals at low signal amplitudes but overestimation of amplitude changes at larger amplitudes. There was no cross contamination detected for either sensing system. When surrogate variable sensor system sensitivity to change in airflow amplitude was examined by differentiating the fitted mathematical expressions, the thermal system response characteristics were hyperbolic. Thus, the thermocouple system was very responsive to changes in airflow amplitude at low airflow amplitude but this sensitivity rapidly decreased as airflow amplitude increased, reaching a low, but relatively constant, sensitivity at higher airflow amplitudes (figure 4). For the thermocouple system, nasal and oral route response characteristics were similar. In contrast, the response of both the nasal and oral pressure cannula 467
J M Gehring et al
Physiol. Meas. 35 (2014) 455
systems increased with increasing airflow amplitude in a slightly curvilinear fashion. However, for the pressure based system, the nasal and oral pressure cannula response characteristics were dramatically different with the oral cannula response reduced across all the range of airflow amplitudes compared with the nasal cannula response. When the responses of the two surrogate systems were analyzed, using a linear model applied across a restricted pneumotachograph amplitude airflow range (
Related content
PAPER
Response characteristics for thermal and pressure devices commonly used for monitoring nasal and oral airflow during sleep studies To cite this article: J M Gehring et al 2014 Physiol. Meas. 35 455
View the article online for updates and enhancements.
- Sensitivity and specificity of hypopnoea detection using nasal pressure in the presence of a nasal expiratory resistive device (Provent®) Stephen Milne, Terence C Amis, John R Wheatley et al. - Quantitative flowmeter for polysomnography J P Kirkness, M Verma, B M McGinley et al. - Sleep apnea: a review of diagnostic sensors, algorithms, and therapies Mehdi Shokoueinejad, Chris Fernandez, Emily Carroll et al.
Recent citations - Lala Li et al - Sensitivity and specificity of hypopnoea detection using nasal pressure in the presence of a nasal expiratory resistive device (Provent®) Stephen Milne et al
This content was downloaded from IP address 134.157.146.58 on 16/05/2018 at 15:19
Institute of Physics and Engineering in Medicine Physiol. Meas. 35 (2014) 455–470
Physiological Measurement
doi:10.1088/0967-3334/35/3/455
Response characteristics for thermal and pressure devices commonly used for monitoring nasal and oral airflow during sleep studies J M Gehring 1 , J-G Cho 1,2 , J R Wheatley 1,2 and T C Amis 1,2 1 Ludwig Engel Centre for Respiratory Research, Westmead Millennium Institute, Westmead, NSW, Australia 2 University of Sydney at Westmead Hospital, Westmead, NSW, Australia
E-mail: [email protected] Received 5 November 2013, revised 17 January 2014 Accepted for publication 28 January 2014 Published 20 February 2014 Abstract
We examined thermocouple and pressure cannulae responses to oral and nasal airflow using a polyester model of a human face, with patent nasal and oral orifices instrumented with a dual thermocouple (F-ONT2A, Grass) or a dual cannula (0588, Braebon) pressure transducer ( ± 10 cm H2O, Celesco) system. Tidal airflow was generated using a dual compartment facemask with pneumotachographs (Fleisch 2) connected to the model orifices. During nasal breathing: thermocouple amplitude = 0.38 Ln [pneumotachograph amplitude] + 1.31 and pressure cannula amplitude = 0.93 [pneumotachograph amplitude]2.15; during oral breathing: thermocouple amplitude = 0.44 Ln [pneumotachograph amplitude] + 1.07 and pressure cannula amplitude = 0.33 [pneumotachograph amplitude]1.72; (all range ∼0.1–∼4.0 L s−1; r2 > 0.7). For pneumotachograph amplitudes 0.7. For the pressure cannula, the data were best fitted using a power function; for nasal breathing P = 0.93 V2.15 (figure 3(C)) and for oral breathing P = 0.33 V1.72 (figure 3(D)); both r2 > 0.9. In comparison with a theoretical linear response (line of identity; see figure 3) both nasal and oral TS overestimated V at lower V ( 0.7). For the nasal pressure cannula: P = 1.11 V − 0.33, while for oral breathing: P = 0.29 V − 0.04 (both r2 > 0.7). In this case the response values (i.e. dTS/dV and dP/dV) are the fitted line slopes. Thus, while the response for nasal and oral V is similar for the thermal based system, the response of the pressure based system for the detection of nasal V was almost four times that for oral V. 461
J M Gehring et al
Physiol. Meas. 35 (2014) 455
(A)
(B)
Figure 5. Scatter plots showing breath-by-breath values (filled circles represent data from nasal sensors, open circles data from oral sensors) for a restricted ( 0.25 (figure 8(D)), but not for nasal breathing or for pressure cannulae signals. ∗ p < 0.05, r > 0.9.
2.0) were performed for both the nasal and oral routes with target flow amplitudes between 0.2 to 1.0 L s−1. Breath-by-breath amplitudes for TS and P were plotted against the corresponding values for V. Differentiation of mathematical expressions fitted to TS or P versus V was used to calculate dP/dV and dTS/dV and device sensitivities were calculated at a flow amplitude of 0.5 L s−1 (figure 8). For both oral and nasal routes of breathing, neither dP/dV nor dTS/dV correlated with TI/TE at any Ttot (Pearson’s correlation, all p > 0.05). Similarly, no significant correlations were found between dP/dV or dTS/dV and Ttot for any TI/TE except for dTS/dV for oral breathing at TI/TE > 0.25 (r2 > 0.9, p < 0.05). For example, at TI/TE = 1.0, dTS/dV was 1.1 volts/L s−1 for Ttot = 5 s and 1.6 volts/L s−1 for Ttot = 10 s, i.e. a lower Ttot resulted in lower dTS/dV. 4. Discussion In this study of the response characteristics of both thermal and pressure based sensor systems for the detection of nasal and oral route respiratory airflow, the major findings were: (1) the thermocouple system was more responsive at lower airflows than at higher airflow, while the reverse applied to the dual cannula pressure based sensor system, (2) there was no cross contamination from one breathing route to the other for either sensor systems, (3) nasal 465
J M Gehring et al
Physiol. Meas. 35 (2014) 455
and oral route sensor response characteristics differed for both systems, (4) oro-nasal airflow partitioning may not be accurately reflected by either system, (5) changes in oral orifice crosssectional area had opposite effects on the response characteristics of thermal versus pressure based oral sensors, and (6) altering the pattern of breathing only affected the oral thermocouple sensitivity. Several previous studies have examined the relative response characteristics of nasal pressure cannulae and thermally responsive sensors for the detection of apneas and hypopneas (respiratory events) during sleep (S´eri`es and Marc 1999, Norman et al 1997, Thurnheer et al 2001, Heitman et al 2002). In general these studies have found that thermistors/thermocouples do not detect as many events as are identified by nasal pressure recordings. Moreover, reductions in thermal based signals associated with falls in tidal volume are less than those recorded with other monitoring systems such as inductive plethysmography (S´eri`es and Marc 1999, Pasto et al 1996). In addition, thermally responsive monitoring systems typically have long time constants making them insensitive to the shape of the inspiratory airflow profile and, therefore, are not suitable for evaluating inspiratory airflow limitation (Montserrat et al 1997). These findings have led to recommendations that non-invasive nasal airflow monitoring during sleep is best accomplished using nasal pressure measurements rather than thermistors/thermocouples (American Academy of Sleep Medicine 1999). Heitman et al (2002) provided evidence validating the detection of apneas and hypopneas during sleep with nasal pressure cannulae in comparison with ‘gold standard’ measurement using a pneumotachograph. Most studies that have compared thermal and pressure based airflow monitoring systems for polysomnography have concentrated on the detection of sleep disordered breathing events (S´eri`es and Marc 1999, Norman et al 1997, Heitman et al 2002). Few studies have quantitatively examined the relationship between ‘gold standard’ measures and surrogate variable signal magnitude. Pneumotachograph measurement of airflow is such a ‘gold standard’ methodology but requires the use of a face mask. Since such instrumentation of the face is likely to alter the conditions of operation for the surrogate systems this approach has not been favored, especially for the thermal based systems. Attempts have been made to compare surrogate signals with other measures of airflow. Pennock (1992) compared thermal device responses with those from a ‘piezo-electric belt’ showing that disagreement between these airflow monitoring systems was related to individual device design. The relative response characteristics for the quantification of oral versus nasal airflow were not examined. Berg et al (1997) compared minute ventilation measured directly by means of a head-out body plethysmograph with measurements of nasal ventilation as detected by thermal, pressure and inductive plethysmographic monitoring systems. They concluded that all these indirect methods were in poor agreement with plethysmograph measured ventilation in awake subjects. Other attempts to quantitatively evaluate surrogate variable measures of airflow have been performed in physical models of the upper airway. Farr´e et al (1998) described a heated chamber model through which cyclical airflow (measured via a pneumotachograph) could be generated by a servo-controlled bellows. This model was used to test the response characteristics of a thermistor placed in front of an artificial nose (two cylindrical tubes built into the front wall of the chamber). Tested over a peak-to-peak airflow amplitude (sinusoidal airflow) of 0 to 1.0–1.5 L s−1 the response of the themistor was nonlinear and depended on the airflow pattern, the distance of the thermal probes from the ‘nose’ and the cross-sectional area of the ‘nose’. Fleury et al (1996) describe a model consisting of a polystyrene ‘head’ with two tubes for nostrils but do not provide any data. We used a physical model of the face in an attempt to reproduce the operating conditions for the thermistor/thermocouple and pressure monitoring systems when used on an unencumbered 466
J M Gehring et al
Physiol. Meas. 35 (2014) 455
human face. Our physical model was constructed to reflect the anatomical relationships of the human face and to reproduce the relative dimensions of the nasal passages, especially the nasal valve and nasal vestibule region. Unlike previous models our model also incorporates an oral pathway. In addition, the use of modeling compound to define the mouth of the model allowed the effect of oral orifice geometry to be studied. The approach of having a human subject, connected to the model via a dual compartment face mask and breathing with feedback from dual pneumotachographs, allowed target peak airflows to be achieved and exposed the sensor systems to human exhaled gas. Limitations of the model include the rigid nature of its surface, the lack of a background body temperature (thermal systems applied to the human face operate with a background of skin surface temperature), potential loss of heat and water content of exhaled gas in the face mask and tubing systems and the lack of any upper airway muscle activity. In addition, although the lips and oral orifice were modeled, the oral cavity and pharyngeal regions were not. It is well recognized that measurement of total respiratory airflow using nasal route monitoring is subject to error if mouth breathing occurs (Norman et al 1997, Thurnheer et al 2001). For this reason, polysomnographic sleep monitoring systems often include a thermal or pressure based sensor positioned at the lips (S´eri`es and Marc 1999). While there has been considerable attention paid to the accuracy of surrogate variable measurements of nasal airflow very little attention has been paid to the accuracy of thermal and pressure based systems in the assessment of oral airflow. There are grounds to suspect that there might be differences in the sensitivity of these systems to mouth versus nose breathing. For example, oral exhaled gas may be higher in water content and temperature than nasal exhaled gas, and oral geometry (degree of mouth opening) can vary over a much wider range for the mouth than for the nose. Differences in airflow regimes (e.g. turbulence) will influence local pressure measurements. Additionally, when such measurements are being made to assess the degree of oro-nasal airflow partitioning the relative response characteristics of the same sensor system applied to the nasal versus the oral pathway become critical. For example, even if the surrogate variable under study does not accurately reflect airflow, if the response characteristics for mouth and nose airflow are similar the system may still be capable of accurately reflecting the relationship between nasal and oral airflow (e.g. nasal/oral signal ratio). However, no previous studies have examined the relative nasal versus oral response characteristics for thermal or pressure based surrogate variable airflow detection systems. As has been demonstrated previously (Farr´e et al 1998), and in the present study, over the entire range of flow rates studied both the nasal and oral thermocouple and pressure sensing systems were not linearly related to the pneumotachograph measurements. The thermocouple signals were best represented by a logarithmic function characterized by overestimation of changes in pneumotachograph signal amplitude (when compared with a linear response) for amplitudes below ∼1.4 L s−1 for nasal breathing and ∼1.1 L s−1 for oral breathing (see figure 3). In contrast the pressure cannulae response was best represented by a power function characterized by similar changes in surrogate variable and pneumotachograph signals at low signal amplitudes but overestimation of amplitude changes at larger amplitudes. There was no cross contamination detected for either sensing system. When surrogate variable sensor system sensitivity to change in airflow amplitude was examined by differentiating the fitted mathematical expressions, the thermal system response characteristics were hyperbolic. Thus, the thermocouple system was very responsive to changes in airflow amplitude at low airflow amplitude but this sensitivity rapidly decreased as airflow amplitude increased, reaching a low, but relatively constant, sensitivity at higher airflow amplitudes (figure 4). For the thermocouple system, nasal and oral route response characteristics were similar. In contrast, the response of both the nasal and oral pressure cannula 467
J M Gehring et al
Physiol. Meas. 35 (2014) 455
systems increased with increasing airflow amplitude in a slightly curvilinear fashion. However, for the pressure based system, the nasal and oral pressure cannula response characteristics were dramatically different with the oral cannula response reduced across all the range of airflow amplitudes compared with the nasal cannula response. When the responses of the two surrogate systems were analyzed, using a linear model applied across a restricted pneumotachograph amplitude airflow range (
