Introduction 5 1 4 7 In vitro, the moisture exchanging capacity of the HME can be specified with the ISO 9360 standard. However, this standard does not incorporate a standardized method for determining the heat exchange of the HME, and it is not possible to translate the results obtained with the lung model of the ISO 9360 standard to the true in vivo situation. Therefore several questions remain unanswered. Fundamentally: what is the in vivo heat and moisture exchanging capacity of the HME? And clinically, how does the in vivo heat and moisture exchanging capacity relate to that of the upper airway, i.e. where does the HME stand in the range from open stoma to nose breathing? Is a further improvement of the HMEs heat and moisture exchanging capacity possible and desirable, and does this result in a further decrease of the pulmonary complaints due to the use of this rehabilitation device? 15 6 8 10 11 13 14 10 8 9 Our goal was to develop a verified intra-tracheal airway climate analyser with response characteristics fast enough for assessment of end tidal intra-airway temperature and humidity variations during normal breathing in laryngectomized individuals, and for evaluation of the influence of HMEs on these parameters. Design considerations Placement of a humidity sensor in the airway itself was rejected. Humidity sensors must be kept clean, and a construction to protect a sensor would be so large that the aerodynamics within the airway would be disturbed. A high airflow around the sensor is required to minimize the response time of the sensor. A constant airflow is necessary for a reliable interpretation of the time dependent signal, and in addition facilitates the use of deconvolution to optimize system response characteristics. Therefore, a sensor environment had to be created in which airflow remains constant and high during the whole breathing cycle, i.e. in a “chamber”. In order to disturb airway aerodynamics as little as possible, a small diameter sample catheter for transportation of breathing air outside the airway to a humidity sensor is required. We decided that the lumen surface of the sample catheter should not exceed 10% of the lumen surface of the trachea. A decrease in lumen area of maximal 10% will result, by the law of mass conservation, in an increase in local flow velocity of 10%. This on its turn will lead to a transbronchial pressure drop of maximal 20% which is supposed to be small enough to avoid substantial deformation of the trachea wall. This means that based on a diameter of 2 cm, which is representative for the trachea of an adult person, the maximal diameter of the sample catheter should not be more than 6 mm. 16 http://www.thunderscientific.com/web_humicalc/index.php We decided to limit the sample rate to less than 10% of the respiratory minute volume. Based on a breathing frequency of at least 12 cycles/minute and a tidal volume of 0.5 l for an adult human at rest, the sample rate should be 0.6 l/min at most. Finally the catheter should be easy to manipulate and it should be possible to insert it into the trachea even if an HME filter is in situ. The primary measuring site for the sensor should be about 1 cm behind the HME, but preferentially, measurements deeper in the trachea should be possible as well. Components Airway Climate Explorer 1 Fig. 1 a b  2 3 Fig. 2 I II III IV V VI  Fig. 3 2  12 15 4 Fig. 4 Block diagram of the airway climate explorer  For patient safety, the total assembly is connected to a safety-isolating transformer. Data acquisition and signal processing All signals of the total assembly, i.e. voltage output of the sensor polymer and the aluminium wire on the humidity sensor of the ACE, temperature signal of the thermocouple, sample rate of the mass flow controller, and all ancillary equipment (see below), are read out simultaneously via a multichannel data acquisition system with additional software (Powerlab and Chart 5.4.1 software, ADInstruments Ltd, Oxfordshire, UK) on a PC. For deconvolution of the raw humidity output of the ACE, the software application LabVIEW (National Instruments Netherlands BV, Woerden, The Netherlands) is used. Cleaning and disinfection protocol Prior to every in vivo measurement, the sample catheter is cleaned and disinfected with the following procedure. First the catheter is rinsed with tap water. Subsequently the catheter is cleaned with disinfection hand soap (Baktolin Basic, Bode Chemie, Hamburg, Germany) and rinsed again with tap water. Then the catheter is placed in a solution of Biotex (Sara Lee, Utrecht, The Netherlands) and tap water for 10 min, after which it is placed in a 70% alcohol–water solution for 5 min. During the disinfection procedure, chemicals are prevented from entering the central canal of the sample catheter by establishing retrograde airflow, in order to protect the humidity sensor polymer. The central canal is not sterilized because during use there is continuous suction of air towards the vacuum pump. Finally the catheter is dried in room air. Ancillary equipment for frequent in hospital calibration For frequent, in-house, temperature calibrations a can of water is used in which the thermocouple is placed, simultaneously with a calibrated thermometer (Thermalite, Electronic Temperature Instruments Ltd, Worthing, UK), accuracy: ±0.4°C, acting as a secondary reference. By using water with melting ice or warm tap water, the full temperature range can be quickly calibrated. 3 Assessment of step-response characteristics was performed in a two-stream system, which consisted of two tubes that were mounted side by side, connected to—in absence of compressed room air facilities—the hospital’s oxygen supply. Through one tube, dry oxygen flowed, and through the other the oxygen was heated and 100% humidified with the respiratory humidifier. Subsequently, the sample catheter tip was manually switched between both tube lumina. Leakage testing is performed by checking the time required to completely empty a flexible sack that is filled with a well-defined volume of air (1 l). This volume is pushed into the sack with a calibration syringe (Jaeger GmbH/VIASYS Healthcare GmbH, Hoechberg, Germany). Environmental temperature and humidity are monitored with the heated humidity and the temperature sensor. Calibration and verification  Full calibration of the system was performed with one single sensor beam. However, due to the disposable design of a radiosonde and the possibility of damaging the sensor during the measurements, most likely several sensor beams will be used over time. Therefore, response time behaviour was checked also for a second sensor beam, and the calibration procedure was designed such that these characteristics can be easily determined for a consecutive beam. Operating temperature Raising the temperatures of the system, and therefore of the humidity sensor, above room temperature, we observed an initial decrease in the response time, but above 40°C the response time increased. Therefore 40°C was chosen as the operating temperature. At this temperature, the sampled air in the catheter reaches 40°C well before arrival in the sensor house, and therefore the influence of breathing temperature variations on sensor temperature is negligible. For measurements at 1 cm deep in the trachea the heated part of the tube does not enter the trachea so that artificial heating of the inspired air will not occur. Static calibration 5 Fig. 5 Output of the humidity sensor plotted against the secondary reference  The thermocouple measuring the temperature at the tip of the sample catheter was 2-point calibrated against the values of the reference thermometer in ice water and water at body temperature. This procedure can be performed in the hospital, prior to every in vivo measurement. Given the accuracy of the reference thermometer, the accuracy of the system for measuring temperature values after a 2-point calibration is 0.3°C. Assessing the role of condensation in the measuring system In order to assess the necessity of heating the system in order to prevent condense formation within the device, step response measurements were performed with an unheated system and subsequently with the system heated to 40°C. With an unheated system, a delay of up to several seconds was observed in the humidity trace relative to the thermocouple trace after a downward step. The longer the sensor is kept in a high humidity environment, this delay increases; and it is indicative of the formation and subsequent evaporation of the condensate within the system. Even when simulating clinically relevant breathing frequencies, a noticeable delay of 0.15 s was observed. When measuring high humidity at operating temperature for a long time, after the downward step, a very fast response with a delay of less than 0.05 s was observed. This delay is caused by the transport of sampled air from the tip of the catheter towards the humidity sensor and is not relevant for the measurement of end tidal values. In the downward trace this fast response was followed by a shoulder at about 50% of the step, of which the length again increased with the duration at high humidity. This shoulder probably represents the evaporation of the condensate in the unheated tip of the sensor. During in vivo measurements, however, the system will never be exposed to high humidities for an extended period of time. Therefore we simulated a range of breathing frequencies with the two-stream system resulting in a series of block functions of approximately 9, 18, and 35 cpm (cycles per minute). At 9 cpm a short shoulder is still visible, but the condensate evaporates sufficiently fast that the “end-inspiratory” value is reached. At 18 and 35 cpm no shoulder was observed. Raw response time characteristics 6 Fig. 6 solid black line solid grey line dashed line 2 a b  For the thermocouple, the response to increasing and decreasing temperature steps was <0.2 s. For the first humidity sensor, the response to increasing humidity and decreasing humidity was <0.5 and <0.8 s, respectively, at a frequency of 18 cpm. We verified that the response characteristics at other frequencies were comparable. Both rise and fall times of the second humidity sensor were approximately 60% shorter. The observed raw response times are slightly longer than we had expected based on the specifications of the Vaisala Humidity Sensor, probably because the airflow over the sensor is slower than specified. Although every attempt was made to minimize the volume of the sensor chamber we estimate that the airspeed over the sensor is between 2 and 3 m/s instead of 6 m/s. Deconvolution of the humidity signal \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym} 
\usepackage{amsfonts} 
\usepackage{amssymb} 
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}
$$ f(t) = e^{{ - t/\tau _{{}} }} $$\end{document} For the removal of excessive noise a low pass filter of <300 cpm was applied. Because the raw response characteristics varied in between the humidity sensors and over time, τ was adjusted for each individual measurement. The principle of adjustment of these parameters was to acquire a maximal value for τ, such that no overshoot of the deconvolved humidity trace was visible. We used the thermocouple signal as reference for the true step function. For the first sensor, the deconvolved signal was stable and optimal for increasing humidity with τ = 0.4 s. For the second humidity sensor a τ of 0.1 s appeared to be appropriate. We verified that the raw step response characteristics were reproducible, and that the deconvolution results were not influenced by the step size. 6 6 2 2 At 35 cpm even the amplitude of the fast thermocouple signal does not reach the true peak-to-peak values. The deviation is less than 0.5°C so that we do not consider a correction necessary for the thermocouple signal. The in vivo measurement 7 8 2 . 8 Fig. 7 Sample catheter in situ, pushed through a punched hole in the HME plaster, without HME filter in situ  Fig. 8 black trace grey trace 2  Discussion 6 8 10 11 13 14 6 Although contact of the humidity sensor of the ACE by tracheal secretion is less likely, the central canal of the sample catheter is susceptible to obstruction by tracheal secretion. When this would occur, the sample catheter can be disconnected from the sensor house and subsequently the obstructive plug can be removed by establishing reversed airflow. The electrodes in the bottom of the droplet interceptor appeared to work effectively. If the humidity sensor would be damaged, it can easily be replaced. Due to the “in hospital” calibration procedures, the measurements can be continued within a relatively short time span. 13 10 11 8 9 9 10 11 However, as the two-stream block function experiment has shown, also in the present measuring device, the amplitude of the deconvolved humidity trace depends on the breathing frequency. The clinical implication of this finding is that, at higher breathing frequencies, the true peak-to-peak values are slightly, but progressively underestimated by the ACE, although probably less than in vitro due to more gradual humidity and temperature variation in the latter situation. The variation in end-inspiratory humidity values and temperature signals, as indicated by the deconvolved humidity signal and the thermocouple in the in-vivo experiment may therefore be partially due to a frequency dependency and not due to a real variation in humidity and temperature values. Quality control The first sensor functioned correctly for 1 year and the second for half a year. Both sensors had to be replaced because of accidental damage during cleaning. The two humidity sensors appeared to have different response characteristics. Sensor age, variation in the production process of the humidity sensor etc. may result in variation between the characteristics of different humidity sensors, but also between the characteristics of one sensor measured over time. Linearity between reference humidity values and sensor output in all consecutive sensors is felt to be a prerequisite for serial production of radiosondes. The small climate room and the 2-point calibration is therefore considered appropriate for daily quality control. A potential source of error is leakage of air, causing dilution of sampled tracheal air with room air. Therefore the simple leakage test should be part of the regular quality control. Although both tested humidity sensors revealed the different step response times of the raw humidity signal, the peak-to-peak amplitude of deconvolved signal was comparable between both sensors. The two-stream step response set up facilitates regular control of the sensor response characteristics and, if necessary, adjustment of the deconvolution function can be easily performed. Limitations and further improvements During measurements deeper in the trachea (e.g. 4 cm), contact of the catheter tip with the airway wall may lead to irresistible cough. Preferably the distal end of the sample catheter should be better pliable in the shape of the trachea, in order to facilitate better control of its tip and reducing airway wall contact. Also visual control of the catheter tip may contribute to prevention of wall contact. The response of the sensor when decreasing the humidity is slower than the response to increasing humidity. However, the same simple deconvolution function is used for increasing and decreasing humidity. A more advanced signal processing might further improve response characteristics and decrease the remaining frequency dependence of the peak tot peak amplitude. In particular for the end-inspiratory values this is important because of the relatively short time span of the inspiration time compared to the expiration time, combined with the relatively slow response characteristics to decreased humidity compared to increased humidity. In conclusion, the Airway Climate Explorer is an easy to use, relatively inexpensive tool for intra airway temperature and humidity measurements. When looking at the end-inspiratory and end-expiratory humidity values as measured by our system, the accuracy is adequate for assessment of intratracheal climate.