to the detection of obstructive a signal processing ... oral...osa: obstructive sleep apnea 2 sleep...

18
A Signal Processing Approach to the Detection of Obstructive Sleep Apnea NRP: EEE24A Jovyn Tan Li Shyan Hwa Chong Institution

Upload: others

Post on 07-Apr-2020

6 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

A Signal Processing Approach to the Detection of Obstructive Sleep Apnea

NRP: EEE24A

Jovyn Tan Li ShyanHwa Chong Institution

Page 2: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

OSA: Obstructive Sleep Apnea

2

❏ Sleep disorder where breathing stops for ❏ at least 10 seconds,❏ more than 5 times/hour

1 in 3 Singaporeans

1 billion people worldwide

Heart disease

High blood pressure

Daytime fatigue

Health Complications

Page 3: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

3

Electromyogram Electro-oculogram Electromyogram Electrocardiogram

Current Form of Diagnosis

Polysomnography (PSG)

Page 4: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

4

Current Form of Diagnosis

Polysomnography (PSG)

Apnea Hypopnea Index (AHI)

Problems

Distorts OSA condition

Manual data analysis

Cannot analyse all data collected

No differentiation of event severity

Aims of Research

An automated system without full PSG

1

Automated 3-class system(Healthy, Hypopnea, Apnea)

2

Aims & Objectives

3-class Fisher’s Ratio3

Page 5: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

5

Electromyogram Electro-oculogram Electromyogram Electrocardiogram

OSA Diagnosis

Oro-nasal thermistor Respiratory effort belts

Proposed method

Page 6: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Data

6

14 channels

Data

obs

erva

tions

Sampling rate: 64Hz64Hz * 60s * 60min * 6.2h = 1428480 data points

Page 7: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Methodology

7

Signals extracted from data● Oro-nasal airflow● Rib cage movement● Abdomen movement Normalisation

Preparation of data for machine learning

Page 8: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Methodology

8

Signals extracted from data● Oro-nasal airflow● Rib cage movement● Abdomen movement

Feature Extraction

Segmentationwindows of 1024

data pointsNormalisation

Preparation of data for machine learning

1024 points= 1 window

Page 9: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Feature Extraction

9

❏ 15 features extracted from each signal❏ e.g. mean peak prominence, number of peaks

❏ 5 peak processing thresholds

Page 10: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Preparation of data for machine learning

Methodology

10

Signals extracted from data● Oro-nasal airflow● Rib cage movement● Abdomen movement

By box plot analysis

Feature Selection

By 3-class Fisher’s ratio

Conceptualising 3-class Fisher’s Ratio

Segmentationwindows of 1024

data pointsNormalisation

Feature Extraction

Page 11: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Fisher’s Ratio

11

❏ Measures discriminating power of a variable

❏ 2-Class FR:

Page 12: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

3-Class Fisher’s Ratio

12

Page 13: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Feature Selection by Box Plots

13

Suitable feature Unsuitable feature

Page 14: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Preparation of data for machine learning

Methodology

14

Signals extracted from data● Oro-nasal airflow● Rib cage movement● Abdomen movement

By box plot analysis

Feature Selection

By 3-class Fisher’s ratio

Recursive Feature Elimination

Feature Elimination

Principal Component Analysis (PCA)

Support Vector Machines (SVM) using Matlab

Conceptualising 3-class Fisher’s Ratio

Segmentationwindows of 1024

data pointsNormalisation

Feature Extraction

Page 15: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Classification Results

15

2 Classes

3 Classes

Page 16: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Results [2 classes]

16

Highest Accuracy: 94.8%Sensitivity: 96%Specificity: 93%

Cubic kernel28/28 features

No PCASelection by FR

Healthy

Hea

lthy

Apn

ea

Apnea

True

Cla

ssPredicted Class

Rib

cage

and

Abd

omen

Mov

emen

ts

Oro-nasal Airflow

Standard Deviation of Peak Prominence of 2 Signals

Page 17: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Results [3 class]

17

Highest Accuracy: 83.4%

Medium Gaussian kernel19/28 features

Selection by box plotsPCA enabled

Oro-nasal AirflowHealthy

Hea

lthy

Sev

ere

Apn

ea

Severe Apnea

Mild

Apn

ea

Mild ApneaTr

ue C

lass

Predicted Class

Standard Deviation of Peak Prominence of 2 Signals

Rib

cage

and

Abd

omen

Mov

emen

ts

Page 18: to the Detection of Obstructive A Signal Processing ... Oral...OSA: Obstructive Sleep Apnea 2 Sleep disorder where breathing stops for at least 10 seconds, more than 5 times/hour 1

Conclusion

18

An automated system without full PSG1

An automated 3-class system(Healthy, Hypopnea, Apnea)

3

3-class Fisher’s Ratio2