«PEKKA TIIHONEN Novel Portable Devices for Recording Sleep Apnea and Evaluating Altered Consciousness Doctoral dissertation To be presented by ...»
Figure 2.2: The diagnostic criteria of obstructive sleep apnea syndrome in adults (AASM 2005).
The RERA is a respiratory effort-related arousal.
Previously, overnight laboratory polysomnography including EEG has been the only acceptable diagnostic method for evaluation of suspected obstructive sleep apnea.
However, portable monitoring or respiratory polygraphy has now become an acceptable alternative (Thurnheer 2007), and the AASM has published clinical practice guidelines to help clinicians in the use of PM (Collop 2007). The monitoring devices can be classified into four different types (Ferber 1994) (Table 2.6).
Table 2.6: The classification (Ferber 1994) and advantages and disadvantages (Collop 2009) of devices designed for the diagnostics of sleep apnea.
PM has several advantages: increased accessibility, patient acceptance, convenience of home recording, low cost and applicability to telemedicine. However, it also has several disadvantages including potential data loss, risk of misinterpretation of the results due to limited data such as misinterpretation of wake time as sleep, and possibility for inappropriate use of automatic scoring (Douglas 2003, Collop 2009) (Table 2.7).
Pioneering work for portable monitoring was done in the late 1980s (Penzel 1990).
Since then, numerous portable monitor studies have been performed to evaluate the ability of PM to record sleep apnea (Whittle 1997, Ballester 2000, Portier 2000, Lloberes 2001, Gagnadoux 2002, Dingli 2003, Douglas 2003, Ahmed 2007, Collop 2008a, Collop 2008b, Kayyali 2008). In addition, two recent studies, which
concentrated on the outcomes of treatment of obstructive sleep apnea, have demonstrated that in comparison to polysomnography, portable monitoring could effectively diagnose patients with OSA (Whitelaw 2005, Mulgrew 2007).
Table 2.7: The advantages and disadvantages of PM in diagnostics of OSA (Collop 2009).
Advantages Increased accessibility Patient acceptance May be done in the home Convenience Relatively low costs Disadvantages Absence of a trained technician to correct artefacts and make equipment adjustments, sensor failures Inability to intervene in medically unstable patients Potential data loss or distortion Potential misinterpretation of the results due to limited data Inability to perform subsequent multiple sleep latency testing according to the standard protocol Varied sensor technology No measurement of sleep
2.4.2 Methods for recording of sleep apnea
Sensitivity, specificity, reproducibility and patient safety are important general criteria for any sensor for monitoring respiration (Johansson 2004). Small size, minimal interference and obstruction, rapid response, and low cost are also desirable features.
Type 3 PM devices do not contain channels and transducers for recording fast signals (e.g. EEG, ECG). However, there are several transducers for recording nasal and oral flow, respiratory movements, snoring, oxygen saturation, heart rate and body position.
There is a number of possible detection modalities to distinguish respiratory airflow:
differential pressure measurements, hot wire anemometry, passive temperature sensing, humidity sensing and CO2 measurements (Johansson 2004). The gold-standard device for measuring respiratory airflow is the pneumotachograph (AASM 1999, Farre 2004).
Although pneumotachographs are used in sleep laboratories, they are not suitable for routine use with PM devices at home when patients are breathing freely without masks.
A novel method for detecting respiratory airflow is to use a polyvinylidene fluoride (PVDF) thermal sensor on the skin above the upper lip and below the nose (Berry 2005, Nakano 2007).
Detecting respiratory (and other) movements wirelessly (e.g., using static charge sensitive bed) is a tempting method, because the patient can sleep naturally without any immediate contact to a sensor (Alihanka 1981, Erkinjuntti 1984). Unfortunately, this type of sensor detects only movements and does not measure the airflow or oxygen saturation and thus, additional contact type transducers (e.g. pressure sensors using nasal cannulas and pulse oxymeters) are needed.
A single transcutaneous carbon dioxide sensor can accurately assess both ventilation and oxygen saturation (Senn 2005). However, these sensors need regular remembraning and calibration (Eberhard 2007) and while they are applicable for use in sleep laboratories they are not convenient for portable monitoring at home. End-tidal CO2 measures the concentration of carbon dioxide of exhaled gas that is presumed to
originate in the alveoli (Weese-Mayer 2000). During spontaneous breathing, the capnogram is often distorted due to aspiration of expired gas. Furthermore, mouth breathing is a major problem if only the nasal CO2 is measured. Recently, a flowthrough capnometer has been introduced for recording capnograms with minimum distortion and this device can detect apnea reliably during sleep (Yamamori 2008).
Future testing and practical use will show the value of this method in the polysomnography.
The AASM manual (AASM 2007) specifies both the desirable and minimal sampling rates for polysomnographic recordings and the preferred sensors and transducers. The recommendations (excluding those for EEG, ECG, EMG and ECG signals) are presented in Table 2.8. AASM rates an esophageal pressure sensor as an important transducer for respiratory effort detection. In a sleep laboratory it is useful, but in PM conducted at home it is not applicable, especially if the patient has to attempt to fix the transducers without professional help. Table 2.8 can be used as a good starting point for the development of a type 3 PM device. In the next sections potential and useable transducers are elaborated more precisely.
Table 2.8: The specifications for sampling rates of signals applicable to PM recording, adapted from AASM (2007).
Recording of oronasal airflow with a thermistor Thermistor is a special type of resistor whose resistance varies with temperature (U.S.
Patent No: 2,021,491). Thermistors are nonlinear sensors and they usually consist of sintered mixtures of oxides (NiO, Mn2O3, or Co2O3) (Millman 1979).
For accurate temperature measurements, the resistance-temperature curve can be described with the Steinhart-Hart equation
Thus, ln(R) depends linearly on 1/T.
The thermal time constant for a thermistor is the time required for the thermistor to change its body temperature by 63.2% of a specific temperature span when the measurements are conducted under zero-power conditions in thermally stable environments. The thermal time constant affects considerably the recorded signals (Farre 2004). Large time constant values dampen fast changes and should be avoided.
When selecting a thermistor, initial resistance, size, surface material, cost and usability must be taken into account. It should be noted that transducer self-heating will increase the temperature of a transducer above the ambient temperature causing a positive bias to the real temperature value. In order to remove the bias from the signal, a high-pass filtering must be applied before the AD conversion.
A thermistor (or thermoelement) records respiratory airflow indirectly by sensing the temperature changes during inspiration and expiration. Unfortunately, temperature
changes are not linearly correlated to volume flow changes (Farre 2004). The amplitude of the signal produced by a thermistor sensor depends largely on the surrounding ambient temperature in front of the mouth and nostrils. There is no signal if the ambient temperature is the same as the temperature of the exhaled air. Fortunately, this is rarely a problem in clinical recordings. However, a thermistor can give only a qualitative estimate of the airflow change which can lead to uncertainty in detecting flow limitation events (i.e. hypopneas) (Berg 1997, Farre 1998, Redline 2007). Despite this uncertainty, the AASM manual for scoring sleep (AASM 2007) recommends the oronasal thermal sensor as the sensor of choice for detecting the absence of airflow in the identification of apnea events.
Recording of nasal airflow with a pressure sensor Since the speed of airflow in the upper airways (Figure 2.3) is much lower than the speed of sound, it can be assumed that the air flow is locally incompressible. Further, assuming that the flow is quasi-steady and neglecting viscous effects, the Bernoulli law can be applied (Bernoulli 1738, Newcombe 1997, Payan 2003).
Figure 2.3: Sagittal section of the upper airways in the supine position.
where A1 is the cross sectional area of the nasal cavity at the level of the cannula head and v1 is the velocity of the infinitesimal volume element at the level of cannula head.
Applying Bernoulli’s equation to the situation described in Figure 2.4 gives
Figure 2.4: The principle of the nasal flow recording with a pressure transducer.
The pressure transducer measures the differential pressure p2 – p1. According to Equation (10) the volume flow rate is proportional to the square root of the differential pressure. In the transducer, the measured differential pressure is converted to a voltage signal, which is further high-pass filtered and digitized. The high-pass filtering is needed for practical reasons, because transducer offsets are large.
Although the Bernoulli’s principle is used when recording the nasal air flow, the air flow distribution in the nasal cavity is not necessarily uniform or laminar, the density of air is not exactly the same in the cavity and ambient air, and the elevations are not equal. Thus, Equation (10) is not precise. Furthermore, area A1 is not constant from patient to patient and the cannula opening may be far from the centre of the cavity. For this reason, a calibration factor has been suggested to be added on the right side of equation (10). It has been noted that this calibration factor shows a considerable intrasubject and inter-night variation because it depends on the exact position of the prongs in the nostrils and, thus, the device cannot quantitatively estimate the flow throughout
the night (Farre 2004). In some patients, nasal prongs with an inadequate size can induce an increase in airway resistance (Lorino 2000). For these reasons, the measured flow rate values cannot be absolutely correct and due to these uncertainties, mathematical and physical models for detecting inspiratory flow limitation during sleep have been developed (Mansour 2002, Mansour 2004).
Obviously, if only the nasal airflow is recorded and oral breathing is neglected, apnea and hypopnea events cannot be reliably detected. Estimation of oral breathing based on the nasal airflow signal is very difficult because of individual differences in the geometry of the upper airways. Logically, any airflow through the mouth lowers the pressure amplitude recorded from the nostrils. The importance of detecting absence of airflow through both nose and mouth is stressed in the AASM recommendation (AASM
2007) which state that apnea events should be detected with a thermal sensor.
However, the AASM (AASM 2007) also recommends that hypopneas should be detected by a nasal pressure sensor. This recommendation is based on a number of studies showing that nasal pressure recording is an accurate method for recording nasal airflow during sleep (Norman 1997, Ballester 1998, Series 1999, Hernandez 2001, Thurnheer 2001, Heitman 2002, Teichtahl 2003, BaHammam 2004). In particular, the pressure method is more sensitive than the use of a thermistor in detecting hypopneas. A square-root linearization of pressure signal recorded with nasal prongs has been shown to increase greatly the accuracy of detecting hypopneas and flow limitation (Farre 2001). The square-root linearization of the pressure signal to obtain the flow signal is calculated automatically in Somnologica 3.2 (Embla Co., Broomfield, CO, USA). The calculation algorithm is proprietary information of the manufacturer.
Recording of abdominal and thoracic respiratory efforts with strain gauges Body surface movements can be used to study breathing (Gribbin 1983). The instruments which measure ribcage and abdominal motion can be divided into three classes: i.e. those which measure circumference, linear displacement or cross-sectional area of chest or ribcage (Gribbin 1983). A simple mechanical model of chest-wall movement has been developed to describe combined ribcage and abdominal movements (Konno 1967). The model has two large cylinders separated by a floppy membrane.
Both cylinders have their own pistons, which can move independently. The model can explain movements during normal breathing (thorax and abdomen signals are in phase) as well as during upper airway obstruction during apnea (signals are in opposite phase).
Elastic belts have been applied to record volume changes in human limbs (Whitney 1953). The same technology has been used to record circumference of thorax and abdomen as a way to detect respiratory effort. In the early applications, the belts were rubber tubes filled with mercury. Later, the tubes were replaced with elastic belts containing a small elastic transducer. Most commonly, the transducer was a strain gauge or a piezo-electric sensor (Pennock 1990). Although the use of strain gauges may cause overestimation of central apneas (Boudewyns 1997) they are widely used in the diagnosis of sleep apnea. Strain gauges are sensitive to body movement artefacts and the recorded signal must be filtered to avoid emphasizing of high frequencies. Still these sensors are applicable to measurements of respiratory movements. Electrical impedance plethysmography has been used for recording of breathing movements (Yasuda 2005).
Kuopio University Publications C. Natural and Environmental Sciences 261: 1 - 79 (2009)