13
Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

Embed Size (px)

Citation preview

Page 1: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

Classification of the Pulse Signals Based on

Self-Organizing Neural Network for the

Analysis of the Autonomic Nervous System

Present by: Yu Yuan-Chu

 

Page 2: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 2

OutlineAutonomic Nerve System(ANS)

The test function

The relationship between heart rate & blood pressure

R-R interval variability

Data acquisition

System Architecture

Experimental ResultsClinical procedures

Spectral analysis result

Classification of pulse signal result

Correlation between ECG and Pulse signal

Page 3: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 3

ANS Test FunctionANS function

The movement of many internal organs The tempture, blood pressure, heart rate, endocrine and emotion Opposing the outside pressure

Elementssympathetic nerves, parasympathetic nerves, and α,βreceptors

ANS test functionSympathetic:

• BP change in the state of supine and standing • The test of the sustained handgrip • Dark-adapted pupil size after parasympathetic blockade Parasympathetic :

• deep breathing • HR response to standing • Pancreatic polypeptide concentration

Return

Page 4: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 4

The Relationship amongHeart Rate, Blood Pressure and Baroreflex

Blood Pressure(BP):

mainly mechanically induced

Heart Rate Variability (HRV): under baroreflex control via the vagus nerves

BP and RR oscillations occurring at respiratory or Mayer wave(0.1Hz) frequencies is mediated by a baroreflex mechanism

Arterial pressure increase

Arterial baroreceptorsFiring

Sympatheticoutflow to

heart, arterioles,veins decrease

Parasympatheticoutflow to

heart decrease

Reflex via medullarycardiovascular center

Return

Page 5: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 5

The R-R interval variability Y-

Axis

(am

plitu

de)

X-Axis (sec)

Y-Ax

is(s

ec)

X-Axis(sec)

L1 L4L3L2

L1 L2 L3 L4

T T T

(a)

Plasma Epinephrine increase

Activity ofparasympatheticnerves to heart

decrease

Activity ofsympathetic nerves

to heartincrease

Activity of the sinoaterial node (SA node)increase

Heart rate increasei.e.

R-R interval variability decrease

•HRV derived from the ECG signals

T1

NLn

n 1

N

•Sympathetic and Parasympathetic activities regions in PSD

•The relationship between R-R interval variability and autonomic nerves

Return

Page 6: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 6

Data AcquisitionHardware

Finapres: Finger arterial pressure utilizes the principle of arterial wall unloading ECG(12 leads): 12 different potential differences from the body surfaceSCXI-1140: signal conditioning module, 8-channel differential amplifier AT-MIO-16F-5: DAQ board, 200 kHz, its resolution is 12 bits

Software(LabVIEW):Data acquisition systemData analysis systemp

• PSD, 3D PSD, baroreflex analysis and ART2 analysis system

SCXI-1140

Finapres

AT-MIO-16F-5CARD

RS 232

PC AT-586

ECG

(signal conditioning) (Multi-functional I/O card)

•Hardware Architecture •electrodes connected in an leadΠconfiguration

Return

Page 7: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 7

System ConfigurationMain Purpose:

Signal validation between ECG & BP• Hamming windows, Autoregression, PSD

Improve the analytic results• Preprocessor, Adaptive Resonance Theorem of Version 2(ART2)

Finapres

DerivedR-R interval

variability

ART2recognition

sytstem

ECG

DerivedR-R interval

variability

HammingWindows

andAutogression

Preprocessor

Signal validationbetween

ECG&Finapres

Power SpectralDensity Analysis

Recognition pattern(LTM)

Non-invasivedata acqisition

Source arterialpressure

variability data

HammingWindows

andAutogression

•R-R intervals from ECG and Pulse signals Return

Page 8: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 8

Power Spectral Density Analysis

Attenuate the spectral leakage Describe the signal “parsimoniously” by a small number of coefficients

•Hamming Window(Time Domain)

•Hamming Window(Freq. Domain)

•Autoregressive spectrum: Linear Predict Coefficients(LPC)

Return

Page 9: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 9

ART2

Blood Pressure ParametersQ-U: the pulse transmission timeV-D: the diastolic shut timeU-P: the systolic ejection time U-U’: the one cardiac timeP-V: the slow time of ejection

Self-organizes stable pattern recognition codes in real-time Continuous speech recognition and synthesis, pattern recognition, classification of noisy data, nonlinear feature detection Not affected by factors: human fatigue, emotional states, and habituation Return

Page 10: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 10

Clinical ProceduresSix young controlled subjects(23-26 years old) without any clinically evident disease were examined Two standard autonomic tests were undertaken:

Rest- All subjects were asked to lie quietly for 5 minutes with spontaneous breathTilting- recorded over 5 minutes following passive tilting to 75 degree position by the electrically rotating table

Studies were performed between 2:00 PM and 5:00 PMTemperture

The environment tempture was controlled on 24.1 ° C Body temperatures of all subjects were at the range of 35 ° C to 38 ° C

•The validation testing between the ECG and arterial pulse variability is 97.81 + 1.38%

(a) ECG (b) Pluse Return

Page 11: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 11

Spectral Analysis

Indices

LF HF T-test value

Area ︽  ﹀  p = 0.001

Mean ︽  ﹀  p = 0.002

Max ︽  ﹀  p < 0.001

SD ︽  ﹀  p = 0.002

(a) ECG/Rest (b) Pulse/Rest (a) ECG/Tilt (a) Pulse/Tilt

•ECG in the state of tilting up, T-test value between LF and HF ︽ : increase significantly, ﹀ : decrease not significantly

Index LF HF

Area 0.91 0.95Mean 0.95 0.98Max 0.57 0.74SD 0.95 0.88

•Correlation between ECG and Finapres,• Index of Area is best for the PSD in the HRV tests

Return

Page 12: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 12

Classification of pulse signal result

Return

•48.8%, sitting up 60 degree

•27.8%, deep breathing •Status Distribute Plot

•Deep Breathing(Original)

•Deep Breathing(after ART2)

•Sitting up 60 degree(Original)

•Sitting up 60 degree(after ART2)

Page 13: Classification of the Pulse Signals Based on Self-Organizing Neural Network for the Analysis of the Autonomic Nervous System Present by: Yu Yuan-Chu

NCTU BCI Group 13

Correlation between ECG and Pulse signal

Subject 1Date : 03/19/97Time : 09:55 PMState : Tilting upBody temperature=36.5Environment temperature=24.1, Man, Birthday : 65.5.15, Years : 22

Return