Medinfo2013 Decision Support Systems and Technologies - II Non-contact Screening System with Two Microwave Radars in the Diagnosis of Sleep Apnea-Hypopnea Syndrome 21 August 2013 M. Kagawa 1, K. Ueki 1, A. Kurita 2, H. Tojima 3, and T. Matsui 1 1 Faculty of System Design,, Japan 2 Fukuinkai, A Social Welfare Corporation, Japan 3 Department of Chest Medicine, Tokyo Rosai Hospital, Japan 1
Outline 1. Introduction Sleep Apnea-Hypopnea Syndrome and diagnosis 2. Methods Non-contact Vital Signs Measurement with Radars Criteria for apnea/hypopneas with radars 3. Clinical Tests Comparison between Non-contact and Contact tests 4. Results Obstructive Sleep Apnea with paradoxical movements 5. Conclusions and future work 2
1. Introduction Sleep Apnea-Hypopnea Syndrome (SAHS) High prevalence: 2-4% of the adult population (still underdiagnosed) (1) Obstructive Sleep Apnea (OSA) the most common type Collapse of the upper airway in the presence of respiratory efforts (2) Central Sleep Apnea (CSA) Failure of the brain to initiate a breath No respiratory effort (3) Mixed Sleep Apnea (MSA) Apnea (pauses in breathing) Hypopnea (low breathing) airflow chest abdomen SpO 2 airflow chest abdomen SpO 2 Low blood oxygen 3
The gold standard for diagnosis of SAHS Polysomnography (PSG) Overnight monitoring of sleep and variety of body functions during sleep a Apnea-Hypopnea Index (AHI) (the total number of apneic events per hour of sleep) 5 15/hr = mild; 15 30/hr = moderate; and > 30/h = severe brain (EEG) eye movements (EOG) muscle activity (EMG) ECG SpO2 airflow chest abdomen snoring 4
Motivation and Aim of the Project Motivation: PSG with many electrodes restrain the patient and obtrusive inspection methods could disturb the measurement itself. Need for diagnostic alternatives Hypothesis: Displacements of respiratory movement measured by microwave radars have a linear relationship to tidal volumes. Main goal: Developments of non-contact SAHS-diagnostic tools based on microwave radar sensors. Identify novel SAHS features based on the amplitude analysis of chest-abdominal radar signals. 5
2. Methods R1: Chest mattress DC Amp. A/D Conv. 50cm 25cm R2: Abdomen Control unit I - channel Q - channel pillow mattress R1: Chest Body surface oscillation Respiration and Heartbeat Chest Abdomen R2: Abdomen BPF PC system for analysis (LabVIEW) Apnea/Hypopnea Event AGC FFT SSA SAS Diagnosis Respiration & Heart rate BPF : Band Pass Filter AGC : Automatic Gain Control FFT : Fast Fourier Transform SSA : Spectrum Shape Analysis (a) Measurement setup and block diagram LPF I-channel Figure 1 Monitoring system setup LPF LO Q-channel (b) Doppler-radar system and I/Q channels 6
(1) Characteristic of the radar Microwave Doppler radars (NJR4262) Output power 7mW Quadrupole plane antenna (24.11GHz, 24.15GHz) Coverage area Pillow 30cm Radar1:chest Radar2: abdomen (9.5 5.8 1.7cm) Compact microwave radar Incident power density: 1.5 10-2 mw/cm 2 < 1mW/cm 2 Bed the Japanese electromagnetic wave safety guideline The electric wave of this radar is weaker than that of the mobile phone and PHS. 7
(2) Heart Beat Measurement with Radars Detecting the small vibration caused by pulse wave (heartbeats) Radar Radar Doppler detection Entry from heart Expansion A: Arteries (Systole) Entry from heart Contraction B: Arteries (Diastole) 8
(3) Samples of radar output signals respiration pulse Amplitude (V) Amplitude (V) apnea hypopnea 1 5 10 15 Time (sec) (a) Slight vibrations due to heartbeat are superimposed Figure 2 Radar output signals 1 20 40 60 80 Time (sec) (b) Apnea and hypopnea signal 9
(4) Criteria for apnea/hypopneas with radars the American Academy of Sleep Medicine Our original criteria in accordance with the clinical guidelines of AASM* Apnea event: radar amplitude < 20% of the baseline* Hypopnea event 1: radar amplitude < 50% of the baseline* Hypopnea event 2: radar amplitude is 50% to 80% of the baseline* and rise of heart rates is accompanied. Amplitude (V) baseline Apnea event 50% line 20% line Hypopnea event 1 All events last 10 seconds or more. Baseline changes dynamically according to the mean value of the amplitudes of normal breathing radar signal in latest 60 seconds. 1 20 40 60 80 Time (sec) Figure 3 Criteria for Apnea and hypopnea events 10
(5) Flow of the signal processing Four radar channel signals AGC : Automatic Gain Control Heartbeat signal Sampling 100Hz Band-Pass Filter AGC AGC FFT SSA Respiratory signal Heart rates Peak Detection SSA : Spectrum Shape Analysis Apnea Hypopnea Event SAHS Diagnosis Rise of heart rate Respiratory effort Paradoxical movement OSA CSA MSA: Mixed Sleep Apnea Hypopnea Figure 4 Block diagram of SAHS diagnosis 11
(6) Prototype of the system Radars Chest I/Q Abdomen I/Q Control unit (12 12 6.5cm) hypopnea hypopnea LAN Remote monitoring (b) Sample of monitoring display (a) Monitoring system setup Figure 5 Prototype of the monitoring system 12
3. Clinical tests Subjects: eight outpatients at the sleep disorder center (mean age 63; range 30-90, two females and six males) Overnight test: between 9 PM and 6 AM the following day PSG as reference data Monitoring PC R1: Chest R2: Abdomen Figure 6 The setup of the overnight test for SAHS 13
4. Results: OSA with paradoxical movements Amplitude (V) No respiratory effort 1 10 20 30 40 50 60 (a) a typical central sleep apnea (CSA) Time (sec) Amplitude (V) paradoxical movement of the reversed phase 1 10 20 30 40 50 60 (b) an obstructive sleep apnea (OSA) Time (sec) Figure 7 Features of CSA and OSA 14
Results: Hypopnea accompanied by rise of heart rates Radar Amplitude (V) Hypopnea Hypopnea Radar Heart rate (bpm) 86 82 78 74 Delay time 15 sec Rise of HR Delay time 15 sec Rise of HR Heart rate 0 30 60 90 120 Time(sec) Normalized Airflow Blood oxygen level (%) 100 98 96 94 Low blood oxygen Low blood oxygen SpO2 (PSG) 0 30 60 90 120 Time(sec) Figure 8 Features of Hypopnea 15
Results: RDIs calculated with the radar and the PSG RDI: the Respiratory Disturbance Index (the total number of apnea and hypopnea events per hour of recording ) The difference of these is small. Subject S1 S2 S3 S4 Time Radar PSG Radar PSG Radar PSG Radar PSG 22:00~ 91 88 5 4 41 49 20 17 23:00~ 96 89 2 1 48 46 11 16 0:00~ 73 79 11 9 38 46 8 9 1:00~ 58 54 24 20 26 29 2 1 2:00~ 9 6 4 6 38 40 18 19 3:00~ 17 22 32 40 15 17 63 55 4:00~ 33 31 17 16 37 31 25 26 5:00~ 57 59 27 33 35 31 34 31 mean 54.3 53.5 15.3 16.1 34.8 36.1 22.6 21.8 std dev 32.4 31.4 11.5 14.1 10.1 11.0 19.2 16.3 p-value 0.66 0.57 0.48 0.54 Subject S5 S6 S7 S8 Time Radar PSG Radar PSG Radar PSG Radar PSG 22:00~ 19 21 24 30 17 23 5 0 23:00~ 42 36 3 0 33 30 95 92 0:00~ 75 73 7 15 9 12 91 99 1:00~ 60 58 25 20 25 23 74 81 2:00~ 21 29 9 12 30 33 82 78 3:00~ 39 38 10 11 53 50 77 73 4:00~ 48 49 28 24 6 7 73 74 5:00~ 28 21 14 12 32 29 66 64 mean 41.5 40.6 15.0 15.5 25.6 25.9 70.4 70.1 std dev 19.4 18.3 9.4 9.2 15.1 13.2 28.1 30.4 p-value 0.62 0.78 0.85 0.89 16
Results: Linear association and Bland Altman analysis RDI by radars RDIs obtained from radars and PSG correlated very closely, with a Pearson correlation coefficient of 0.98 (n=64). 100 80 60 40 20 0 0 20 40 60 80 100 r = 0.98 RDI by PSG [times/hour] Difference Between RDIs 10 8 6 4 2 0-2 -4-6 -8-10 8.65: Agreement 95% confidence line -8.72: Agreement 95% confidence line 0 20 40 60 80 100 Mean value of RDI between radar and PSG [times/hour] Fig. 9 Comparison between radar test and PSG -0.03: mean 17
5. Conclusions and future work (1) (2) We have identified SAHS features based on the radar amplitude analysis. Developed SAHS diagnostic system can deal with body movement noises and positional changes in bed during sleep. In a practical test at the hospital, we achieved an accurate diagnosis with RDI of SAHS (r = 0.98). Future work: Features abstraction of mixed sleep apnea, detection of consecutive hypopnea events. Our non-contact diagnostic system for SAHS shows great promise as a new screening system. Thank you for your attention! 18