Date post: | 12-Feb-2017 |
Category: |
Health & Medicine |
Upload: | taleb-alashkar |
View: | 209 times |
Download: | 1 times |
Taleb ALASHKAR 1
Medical Multi-signal Signature Recognition Appliedto Cardiac Diagnosis
Supervisor: Eric Fauvet & Olivier Laligant
Centre universitaire Condorcet,Université de Bourgogne
June-13-2012
Taleb ALASHKAR
Outline
Taleb al-Ashkar 2
1. Introduction2. Motivation3. Methodology 4. Results5. Conclusion
Introduction
Taleb al-Ashkar 3
What is Electrocardiogram (ECG) Signal
Fig 1. ECG Phases
Introduction
Taleb al-Ashkar 4
ECG Interpretation
• RR Line• QT Interval• PR Interval• ST Segment• TP Segment
Fig 2. ECG Interpretation
Motivation
Taleb al-Ashkar 5
Statistics• 1/3 people in US has Cardiac Problem• Main reason of mortality in developed countries• Costs of healing an caring of patients
Fig 3. Cardiac Problems Costs in US
Motivation
Taleb al-Ashkar 6
Why we need automatic analysis system
• Decrease costs • Increase efficiency of diagnosis systems
Fig 4. ECG Automatic Analysis
Methodology
Taleb al-Ashkar 7
Methodology of ECG Features Detection
• 1D Nonlinear Filtering Scheme (NLFS)• Mathematical Model for ECG Features Detection• Real Approach for ECG Features Detection
Methodology
Taleb al-Ashkar 8
1D Nonlinear Filtering Scheme (NLFS)
• Edge Detection Approach• Decomposing signal into two signals by:
Y +(z)=T(F +(z)S(z))
Y -(z)=T(-F -(z)S(z))
T: Threshold to select the responseF(z): Detector FilterS(z): Original Signal
Methodology
Taleb al-Ashkar 9
1D Nonlinear Filtering Scheme (NLFS)
Original Signal
Y+
Y-
Fig 5. NLFS Signals
Methodology
Taleb al-Ashkar 10
Mathematical Model for ECG Features Detection
• Applied on: Synthetic & free of noise ECG Signal
Fig 6. Synthetic ECG
Methodology
Taleb al-Ashkar 11
Mathematical Model for ECG Features Detection• QRS Peak Detection
1. NLFS on ECG: Y+
2. Differentiation: difY+
3. Thresholding: TdifY+
4. Linear search: Peaks5. RR line
ECG
Y+
difY+
TdifY+
Fig 7. QRS Peak Detection
Methodology
12
Mathematical Model for ECG Features Detection• Onset of P or T-wave Detection
1. Defining search window (w0 ,w1 )
Fig 8. ECG waves
Methodology
13
Mathematical Model for ECG Features Detection• Onset of P or T-wave Detection
2. Y+ = Y+ [w0 ,w1 ]3. Differentiation difY+ 4. Linear search
P-wave
Y+
difY+
Fig 9. Onset Detection
Methodology
14
Mathematical Model for ECG Features Detection• End of P or T-wave Detection
1. Defining search window (w0 ,w1 )2. Y- = Y- [w0 ,w1 ]3. Differentiation difY- 4. Linear search
T-wave
Y-
difY-
Fig 10. End Detection
Methodology
15
Mathematical Model for ECG Features Detection• Challenges1. Work for free of noise signal
Fig 11. a) Real ECG, b) Synthetic
Methodology
16
Mathematical Model for ECG Features Detection• Challenges2. Variable Morphologies'
Fig 12. a) Inverted T-wave, b) biphasic T-wave
Methodology
17
Real Approach for ECG Features Detection
• Starting from previous Mathematical Model• Modification to overcome challenges
Fig 13. Real ECG
Methodology
18
Real Approach for ECG Features Detection
• QRS Peak Detection
1. Smoothing ECG: Average Filtering2. 1D NLFS: to get Y+
3. Y+ Differentiation 4. Thresholding
5. Linear Search: C0 is the end of each peak6. Search window : QRS peak = max (ECG[C0 -5, C0 +5])7. Repeat 5, 6 steps up to end of ECG signal8. Defining RR line
Fig 14. QRS Peak Detection
Methodology
19
Real Approach for ECG Features Detection
• Synthetic ECG vs Real ECG
Fig 15. a) Synthetic ECG b) Real ECG
Methodology
20
Real Approach for ECG Features Detection• Onset of P or T-wave Detection
1. Defining w0 ,w1
2. Y+ = Y+ [w0 ,w1 ]3. Differentiation by 8 samples step
4. Onset is the index of max value of S(i)Fig 16. S(i) fro Y+
Methodology
21
Real Approach for ECG Features Detection• End of P or T-wave Detection
1. Defining w0 ,w1
2. Y- = Y- [w0 ,w1 ]3. Differentiation by 8 S(i)
4. Index min value of S(i)5. Shifting by 8 samples to get the End
point Fig 17. S(i) for Y-
Results
22
Testing on Standard Database Testing Database• 12 Records of QTMIT Standard Database• Contains Manual Annotations by expert• Each record contains about 30 annotated beats
Results
23
Testing on Standard Database Evaluation Parameters1. Sensitivity:
2. Positive Predictivity
3. Mean Error
4. Standard Deviation
Results
24
Testing on Standard Database Standard Accepted ErrorFor deciding Automatic Detection is TP, FP or FN
Table 1. Maximum Accepted Error
Results
25
Testing on Standard Database Testing Results
ME (ms) SD (ms) Se % P+ %
P-onset 1.48 11.55 75.16 75.16
P-end -1.747 13.57 71 71
R peak -3.251 2.487 98.43 98.88
T-end -7.93 12.396 90.7 90.7
Table 2. Results
Results
26
Testing on Standard Database Comparing with other methods
Method Parameters P-onset P-end QRS T-end
This workSe (%)P+ (%)m±s (ms)
75.1675.16
1.48±11.5
7171
-1.7±13.5
98.8898.88
-3.2±2.48
90.790.7
-7.9±12.3
WTSe (%)P+ (%)m±s (ms)
98.8791.03
2.0±14.8
98.7591.03
1.9±12.8
`99.9299.88NA
99.7797.79
-1.6±18.1
LPD Se(%)P+ (%)m±s (ms)
97.791.17
14±13.3
97.7091.17
-0.1±12.3NA
99.9097.71
13.5±27.0
BayesSe (%)P+ (%)m±s (ms)
99.6NA
1.7±10.8
99.6NA
2.5±11.2NA
100NA
2.7±13.5
Table 3. Comparing Results
Results
27
Testing Against NoiseECG signal with 8 Different Level of additive noise.
Noise Level L1 L2 L3 L4 L5 L6 L7 L8PSNR (dB) 113 106.9 100.9 97.4 94.7 86.9 80.8 77.4
L1 L2 L3 L4 L5 L6 L7 L80
20
40
60
80
100
120
PonPoffRTonToff
Noise Levels
Se(%)
Table 4. Noise Levels
Fig 18. Testing against noise results
Conclusion
28
Pros
1. Exploiting 1D NLFS in ECG features Detection2. Fast & Robust to noise approach3. Possibility to improve performance
Cons:
4. Less performance than other method5. Not sufficient for special ECG cases
Conclusion
29
Future Work1. Developing to detect QRS onset/end 2. Detection of P and T-waves peaks3. Handling special clinical cases in ECG
30
THANKS
Q&A