Measuring Chest Movement Using an Array of Unobtrusive Pressure Sensors
INTRODUCTION <ul><li>Chronic diseases and mobility problems have serious health implication for older adults. </li></ul><ul><li>Transferring healthcare services from in-hospital to home or smart home based monitoring is aligned with older adults expressed preference to remain in their homes. </li></ul>
OBJECTIVE <ul><li>The objective of this work is to detect and measure chest wall motion using pressure sensor array data and compare it to measurements derived from RIP bands in bed based data collection systems. </li></ul><ul><li>We present a non-invasive measurement system for breathing movement analysis with detection of single breath events using a non-contact unobtrusive fiber optic pressure sensor array. </li></ul>
BACKGROUND <ul><li>Currently, the gold standard for disordered breathing during nighttime is for a patient to spend the night in a sleep lab. </li></ul><ul><li>Patient is fitted with a respiratory inductance plethysmography (RIP) sensor that detect breathing abnormalities only while they are in bed. </li></ul><ul><li>RIP bands are used to detect movement in the chest and abdomen . </li></ul>
Respiratory inductance plethysmography (RIP) sensors <ul><li>Inductance Plethysmography employs sensors that are able to measure changes in a cross-sectional area of the patient. </li></ul><ul><li>The RIP sensor consists of a belt with a wire woven or sewn in a sine wave or zigzag pattern along its length. </li></ul><ul><li>Driver module with a circuit board, oscillator and battery that passes a weak current through the wire in t.he band creating a small magnetic field. </li></ul><ul><li>This change in cross-section produces a slight change in the magnetic field that results in a change in the frequency of the current. </li></ul><ul><li>This change can be measured and converted to a voltage output that creates the waveform on the PSG recorder. </li></ul>
NEED FOR UNOBSTRUSIVE SENSORS . <ul><li>Compared to contact sensors, unobtrusive sensors could be </li></ul><ul><li>They are also preferred for long-term use and virtually eliminate the intentional and unintentional compliance problems that have been documented with wearable sensors. </li></ul><ul><li>Unobtrusive sensors can collect long-term trend information about the fluctuations of a disease or condition without increasing the physical or cognitive load on the user. </li></ul><ul><li>Block scheme of the new thermometer: IZ - </li></ul><ul><li>insulating amplifier, Ulf (flU) - voltage-to-frequency </li></ul><ul><li>(frequency-to-voltage) converter, Ml(2) optical to </li></ul><ul><li>electrical signal converer, F - filter </li></ul>
<ul><li>Measurement Equipment </li></ul><ul><li>The measuring system consists of two pressure sensor arrays placed beneath a mattress, and two commercial RIP bands which are used as the reference signal . </li></ul><ul><li>Each sensor array has 24 fiber optic pressure sensors located every 10 cm in an 8x3 framework. </li></ul><ul><li>It is constructed of nonlinear foam sandwiched in two rigid plastic layers. </li></ul><ul><li>The intensity-based sensor’s output is related non-linearly to applied pressure. Its output is a pressure value with an 11 bit resolution. </li></ul><ul><li>Measurement Protocol </li></ul>MEASURING SYSTEM a
MEASURING SYSTEM Measurement Protocol <ul><li>Five healthy participants (20-35 yrs old) volunteered to lie on a mattress while wearing RIP bands. </li></ul><ul><li>Two pressure sensor arrays were placed beneath the single mattress, 10 cm apart to maximize sensor coverage, as shown in fig. </li></ul><ul><li>Participants were asked to produce normal, deep and shallow breaths in 3 positions (side, supine & prone) during 2 hours. </li></ul><ul><li>The pressure sensor arrays and RIP bands were connected to an Armada 700 MHz laptop. </li></ul>Sensor positioning beneath volunteer and change in pressure signal shown for 4 sensors during normal breathing.
PROPOSED APPROACH (A.) Aligning Array Data <ul><li>Each pressure sensor collects information about the pressure detected in a local area. </li></ul><ul><li>That local area may provide ample information about chest motion or little, depending on </li></ul><ul><li>its placement relative to the chest and abdomen of the participant. </li></ul><ul><li>Relative to the chest and abdomen of the participant. Multiple sensors can have common </li></ul><ul><li>information, and if combined properly, create a higher amplitude signal that is more easily </li></ul><ul><li>analyzed. </li></ul><ul><li>Increasing pressure on an individual sensor can also cause the plastic around the foam </li></ul><ul><li>to bend upwards and create an artifact of pressure reduction on an adjacent sensor. </li></ul>
Recorded pressure in two adjacent sensors . <ul><li>Combining the array data into one signal is a three step process . </li></ul><ul><li>each signal is inverted if its cross-correlation with the signal of largest amplitude is stronger with the inverted signal than with the original signal. </li></ul><ul><li>Signals are aligned using the delay value that maximizes the cross-correlation value. </li></ul><ul><li>Lastly, all 48 signals are summed together to produce the combined signal. </li></ul>PROPOSED APPROACH
(B.) Measuring Chest Movement and Signal Synchronization <ul><li>Signal amplitude. </li></ul><ul><li>Delays between signals. </li></ul><ul><li>Timing and duration of individual chest wall movements. </li></ul>PROPOSED APPROACH
RESULTS A. Aligned Array Data <ul><li>The increase in the pressure array data signal amplitude </li></ul><ul><li>after aligning was from 1 to 7 pressure units (1 to 9 dB). </li></ul><ul><li>The delays between sensors in the pressure array were found using </li></ul><ul><li>cross-correlation and were between 0.1 and 0.8 seconds . </li></ul>B. Signal Amplitude <ul><li>After the signals from the array were inverted, aligned </li></ul><ul><li>and summed, the amplitude of the combined signal was </li></ul><ul><li>greater than that of individual sensors. </li></ul>
Pressure sensor array data before alignment Signal amplitude for deep and regular breathing
C. Delays Between Signals RESULTS ∆ T delay values for 400s of regular breathing ∆ T calculation showing delayed pressure signal
D. Timing and Duration of Movements RESULTS <ul><li>The peaks and valleys of the pressure and RIP signals were automatically identified and used to flag a change in the direction of chest movements, as well as calculate the timing and duration of individual inhalation and exhalation segments. </li></ul><ul><li>Only local peaks that were greater than 2 pressure units above or below a previous one were flagged to avoid finding many rapid ‘breaths’ in a noisy signal . </li></ul>
The overlap between labeled inspiration segments of the three signals were measured both with and without adjusting for the ∆T value. Shifting the delayed signal improves the overlap percentage and brings the pressure to RIP overlap to within 11% of the overlap between the RIP bands. Table suggests that when corrected for ∆T delay, the direction change is synchronized. RESULTS
FUTURE WORKS Future work will focus on chest movement analysis and determine particular monitoring needs of older adults in the smart home context.
CONCLUSIONS <ul><li>An unobtrusive pressure sensor array beneath the mattress to determine the duration of each inspiration phase, expiration phase and initiation time of each movement. </li></ul><ul><li>cross-correlation was used to align signals from a pressure array, and local signal peaks were detected to identify chest wall movements from two types of sensors. </li></ul><ul><li>convenient assessments and long-term monitoring of population groups with a greater prevalence of lung and respiratory disorders through passive and discrete sensors. </li></ul>
REFERENCES  R. Steele, C. Secombe and W. Brookes, "Using Wireless Sensor Networks for Aged Care: The Patient's Perspective," Pervasive Health Conference and Workshops, 2006, pp. 1-10, 2006.  D. Foley, S. Ancoli-Israel, P. Britz and J. Walsh, "Sleep disturbances and chronic disease in older adults: results of the 2003 National Sleep Foundation Sleep in America Survey," J. Psychosom. Res., vol. 56, pp. 497-502, May. 2004.  P. Varady, S. Bongar and Z. Benyo, "Detection of airway obstructions and sleep apnea by analyzing the phase relation of respiration movement signals," Instrumentation and Measurement, IEEE Transactions on, vol. 52, pp. 2-6, Feb. 2003.  K. P. Cohen, W. M. Ladd, D. M. Beams, W. S. Sheers, R. G. Radwin, W. J. Tompkins and J. G. Webster, "Comparison of impedance and inductance ventilation sensors on adults during breathing, motion, and simulated airway obstruction," Biomedical Engineering, IEEE Transactions on, vol. 44, pp. 555-566, July. 1997.