«PEKKA TIIHONEN Novel Portable Devices for Recording Sleep Apnea and Evaluating Altered Consciousness Doctoral dissertation To be presented by ...»
5.1 Patients and methods The clinical evaluation of Venla (Study I) was carried out in the Department of Clinical Neurophysiology in Kuopio University Hospital in two phases. The subjects were clinical patients with suspected sleep disorders, especially OSAS and they were judged to require a sleep study. All measurements were part of a normal clinical investigation.
The first part of the clinical evaluation was arranged by recording 19 patients (11 males and 8 females) simultaneously with Venla and with a clinical reference instrument
(Embla, Embla Co., Broomfield, CO, USA) in a sleep laboratory. Permission for the experiment was obtained from the ethical committee of Kuopio University Hospital.
The patients were given oral and written information about the trial protocol, and they provided written consent. The second part of the clinical evaluation included a total of 323 ambulatory home recordings and 275 of those were diagnostically acceptable while 48 recordings failed and were excluded due to missing signals. The numbers of successful recording with Venla and with the commercial ambulatory device (Embletta, Embla Co., Broomfield, CO, USA) were 106 (77 males and 29 females) and 169 (116 males and 53 females), respectively.
The first part of Study II was carried out in the Department of Clinical Neurophysiology with the same indications and ethical justifications as in Study I. Ten ambulatory recordings (5 males and 5 females) were carried out simultaneously with the APV2 device and the reference instrument Embla, in the sleep laboratory. In the second part of Study II, the technical reliability of APV2 was evaluated on the basis of 149 consecutive home recordings. As a reference, a total of 169 patients having suspected OSA were studied with a commercial ambulatory device (Embletta).
In Study III, the recordings made in Study I were re-analysed. However, a small number (N = 8) of recordings were omitted from Study III since the signal quality was not good enough for automatic analysis. The start and stop times for the automatic analysis were set manually by the researcher reviewing the traces with the help of the patient’s sleep log and after visual inspection of the data. The default analysis parameters are listed in Tables 5.3 and 5.4. After the automatic analysis, a manual analysis was performed and the results of the analyses were compared.
In Study IV, a preliminary clinical evaluation of the Emma device was done by recording auditory ERPs in four male volunteers. Two of them also participated in SEP measurements and one in the recording of ECG and spontaneous EEG.
5.2 Technical description of evaluated devices
In Studies I – III, four different devices (Venla, APV2, Embletta and Embla) were used to record sleep apnea. In the laboratory recordings the reference instrument was a commercial polysomnography device Embla N7000 (Embla Co., Broomfield, CO, USA), which is a Type 1 device (Table 2.6). The other three devices belong to Type 3, and their technical properties are summarized in Table 5.2. The device developed for evoked potential recordings in Study IV is described separately in Chapter 5.2.3.
The Venla device (Figure 1 in Study I) consists of a main unit containing electronics and a connector panel for attaching the transducers. The device is powered with two AA-size 1.5 V batteries and has a liquid crystal display (LCD) screen to display the real clock time, oxygen saturation, heart rate, number of recordings in memory, maximum available recording time, body position and push button press.
$ VKDSHG WUDQVGXFHU FRQVLVWLQJ RI WKUHH WKHUPLVWRUV N 17& UHVLVWRU W\SH RH16, Mitsubishi Materials Co., Tokyo, Japan) (Figure 5.1) is used for recording of the oronasal airflow. According to the manufacturer’s specifications, the thermal time constant of an individual element is 6 s and the heat dissipation constant is 0.6 mW/oC.
A modified pressure sensor chamber (PTAFlite, Pro-Tech Services Inc., Mukilteo, WA, USA) is used as the pressure transducer for recording of airflow with a nasal pressure cannula (Embla Co., Broomfield, CO, USA). The chamber contains a sensitive piezo element that converts small pressure changes into voltage values. The first amplifier stage (U1) in Figure 5.2 is direct-current (DC) coupled and it has a bias offset trimming resistor, RV1. During recordings, the thermistor sensor and the nasal pressure cannula are taped together to help the patient to fix the transducers in front of the mouth and the nostrils for recording.
Figure 5.2: Nasal pressure transducer and the circuit diagram of the first amplifier stage.
For the recording of respiratory effort via movements of thorax and abdomen, a special strain gauge transducer was developed (Figure 5.3). They consisted of a 0.5 mm thick double-sided printed-circuit laminate plate (15 mm in width and 95 mm in length) and two small strain-gauge elements (KFG-5-350-C1-11, Kyowa Electronic Instruments Co. Ltd., Tokyo, Japan) glued to both sides of the laminate (Figure 5.4). The thorax transducer is taped onto the patient’s skin between the left mamilla and the left collar bone. The abdominal transducer is taped above the lowest rib on the right side.
Figure 5.3: Printed circuit board of a strain gauge transducer and a sensing element.
Blood oxygen saturation and averaged heart rate were measured with a commercial pulse oxymeter (Nonin XPOD 3011, Nonin Medical Inc., Plymouth, MN, USA) connected via a serial port to the processor of the device. The serial port uses a baud rate of 9600 and an eight-bit data length. The saturation and heart rate values are updated
once per second and displayed on the LCD screen, which enables monitoring of the signal quality at the start of the recording. The oxymeter provides the saturation and heart rate values as integer numbers and thus the resolution is 1% and 1 beat per minute, respectively.
Figure 5.4: Cross section of a strain gauge transducer (on the left) and the circuit diagram of its electronic connection (on the right).
The resistive elements, R30 and R31, are connected in series with the protective biasing resistors, R32 and R33. The direct-current (DC) coupled output from between the elements is connected via a protective resistor, R34, to the first amplifier stage. Resistor R35 connected to virtual ground (VGND1) is for zero biasing when the transducer is not connected.
Figure 5.5: Cross section of a body position sensor in supine position (switches A and B are closed).
When the patient is lying on the right side, the ball inside B goes to the outer end and the switch opens, but switch A remains closed. When the patient is lying on the left side, the states of the switches A and B are reversed. When the patient is in the prone position, both switches are open.
A hemisphere-shaped (39 mm in diameter) body position sensor (Figure 5.5) has two internal gravitation sensitive mechanical tilt switches (CW 1300-1, Conus International, Clifton, NJ, USA). The switches are fixed at a perpendicular angle to each other and at a 45-degree angle to the skin plane (plane X in Figure 5.5). With this arrangement, all main sleeping positions – supine, prone, left, and right – can be detected. Upright and upside down positions cannot be detected with this sensor.
Usually the patient is in the upright position only when he/she is awake and this situation is apparent from other signals (e.g. from respiration effort signals). The
common contacts of the switches A and B are connected to each other and to the system ground. The remaining contacts of A and B are biased through 100 k: resistors to the operating voltage of the processor and connected to different digital inputs of the processor producing two-bit binary information on the body position. One extra bit is added by driving a digital line through the sensor connectors to detect the presence of the transducer. Thus, three bits are used to present position information in the data memory. During the recording, the sensor is taped onto the skin overlying the sternum.
A biased electret microphone is taped onto the patient’s nose to record snoring sound. In order to improve visual inspection of weak snoring sound signals and to prevent signal clipping with high volume snoring, the microphone signals are processed with a special analogue circuit U4 (SSM2165, Analog Devices Inc., Norwood, MA, USA) (Figure 5.6). This circuit effectively produces an analogue voltage that represents a logarithmic value of the input-signal. The logarithmic signal is further amplified and low pass filtered before the analogue-to-digital (AD) conversion. With this arrangement, the final recorded signal represents voice pressure levels in the decibel scale and the sampling frequency can be relatively low, e.g. 8 Hz.
Figure 5.6: Circuit diagram of the biasing, signal amplification and logarithmic conversion circuits of the electret microphone.
The Venla device is designed and constructed around an eight-bit embedded processor, Atmega 128 (Atmel Corp., San Jose, CA, USA), driving three two-channel 12-bit analogue-to-digital (ADC) converter chips, a universal serial bus (USB), an LCD, a random access memory (2.5 MB), a real time clock and a calendar (Figure 2, Study I).
The embedded program is written in C in CodeVisionAVR programming environment (HP InfoTech S.R.L., Bucharest, Romania) and saved to the flash memory of the processor. The electronic erasable programmable memory of the processor is used to store setup parameters and start and stop times and dates of the recordings. The external RAM stores the recorded data of all (maximally 99) separate recordings in a special packaged form. The device has a back-up battery to guarantee the security of internal clock, calendar and data in the RAM.
Kuopio University Publications C. Natural and Environmental Sciences 261: 1 - 79 (2009)
- 50 Methodology For data transfer, conversion and analysis, three computer programs were developed with C++ (Borland C++ 5.01 Development Suite, Borland International, Inc., Cupertino, CA, USA). One program is used for transferring the data from the RAM of the device to the computer. This program also automatically synchronises the clock and calendar of the device with the computer and enables the user to erase all the recorded data in the RAM. Another program was developed for review and analysis of the recorded data. The third program provides conversion to European data format (Kemp
2003) and to a format readable with the Somnologica 3.2 (Embla Co., Broomfield, CO, USA) PSG analysis software.
Prior to any clinical recording, the device was thoroughly technically inspected with reliable laboratory instruments in our own laboratory. In order to ensure that the requirements of the International Special Committee for Radio Interference (CISPR 11 EN 55,011) were fulfilled (Williams 1996), the radio frequency emissions were inspected in an electromagnetic compatibility (EMC) laboratory (Pohjois-Savo Polytechnic, Kuopio, Finland).
The APV2 device (dimensions: 86 mm u 116 mm u 27 mm, weight 0.239 kg) evaluated in Study II is a successor of the Venla device. It records nasal and oral air flow (when an oronasal cannula is used), respiratory effort movements (thorax and abdomen), body position, snore sound level, blood oxygen saturation and heart rate. The air flow is measured with an internal pressure sensor which also detects the snore sound level as a ripple on the measured air pressure signal. The strain gauge transducer developed in Study I was used for the recording of abdominal respiratory movements. Thoracic movements are recorded as a ripple signal with an internal accelerometer sensor that primarily detects body position. An external commercial pulse oxymeter (Nonin XPOD 3011, Nonin Medical Inc., Plymouth, MN, USA) is used to record blood oxygen saturation and averaged heart rate. The device has an LCD screen to display oxygen saturation, heart rate, and other parameters, as well as basic operating instructions for the patient. An internal compact flash (CF) memory card is used for storing the data.
All transducers (except the oronasal cannula) are fixed permanently to the APV2 device and the device contains no buttons. The purpose of this simple design is to enable the operation with minimal or no training. For example, the recording is started and stopped by inserting and removing the batteries, respectively. The recorded data is stored in an encrypted format and no confidential patient information is included in the file. No special program is needed to transfer the data from the CF card to a computer or directly to the internet server. The transfer can be done with ordinary Windows tools or by means of a web browser. On the server, the encrypted data can be converted to EDF and Somnologica 3.2 formats for review and analysis.
The Emma device (dimensions: 344 mm u 173 mm u 308 mm, weight 8.6 kg) designed and constructed in Study IV consists of two main blocks: a data logger and an audio stimulator (Figure 5.7). The liquid crystal displays (LCDs), pushbuttons, opening for a memory card and switches of both blocks are arranged on the front panel. A battery
compartment, an electrode connection panel and a connector for headphones are placed at the back.
Figure 5.7: The block diagram of the Emma device.
The core of the data logger is a 16-channel data logger card containing four fourchannel 22-bit sigma-delta analogue-to-digital converter (ADC) chips (AD7716, Analog Devices Inc., Norwood, MA, USA), a digital signal processor (DSP, ADSP-2181 KSAnalog Devices Inc., Norwood, MA, USA) and a couple of logic and memory chips. The logger card is accompanied by a preamplifier card having four preamplifier channels for EEG and one for ECG. These channels have an amplification of 20u and 10u, respectively, input impedances of 10 M:, and bandwidths from 0.3 Hz to 10 kHz.
The analogue inputs are protected against static charge spikes, which are common in clinical situations, and against other forms of electromagnetic interference (e.g. radio transmissions).