SlideShare une entreprise Scribd logo
1  sur  28
Flow parameters derived from 

impedance pneumography after nonlinear
calibration based on neural networks
Marcel Młyńczak, MSc, Gerard Cybulski, PhD

Warsaw University of Technology, Faculty of Mechatronics,
Institute of Metrology and Biomedical Engineering
Porto, February 23, 2017
Introduction
Spirometry / Pneumotachometry
• The most reliable methods

• Direct measurements
• The need for using a face mask or mouthpiece with nose clip

• Uncomfortable during exercises, sleep and for children
Figures adapted from chat.stackexchange.com and legio24.pl
Mesh grid of 

known pneumatic
resistance
2
Introduction
Impedance pneumography
Basic idea
Changes of transthoracic bioimpedance are 

connected with changes of the amount of air in the lungs.
Method of
measurements
The nature of the
IP signals
Carried out using tetrapolar method.
Volume-related.
Calibration
Simple linear model provides the best accuracy of volume
parameters for specific electrode configuration.
Flow
measurements
Obtained from differentiated 

impedance pneumography signals.
3
The Problem
The flow IP signals are usually underestimated
4
The Problem
Possible solution
Nonlinear signal calibration?
To recover the dynamics of the flow signal artificially by using 

nonlinear processing of the signal.
5
Objectives
• Improvement of the accuracy of flow parameters:

➡ peak flow values, separately for inspirations and expirations

➡ mean flow values, as above

calculated by impedance pneumography, using neural network approach. 

• Assessment of which neural network structure is the best for the flow
parameters analysis.

• Finding the best calibration procedure in terms of flow-related analysis.
6
Methodology
Subjects - generally healthy students, 10 males
Min Avg Max
Weight [kg] 65.0 77.4 100.0
Height [cm] 171.0 179.3 187.0
BMI 20.75 24.14 33.41
Age 19 23 27
7
Methodology
Pneumonitor 2
• ECG signal to estimate heart rate
and tachogram
• Impedance signal relating to 

main breathing activity
• Portable

• Recording on SD card

• Rechargeable battery
• Motion signal from 3-axis
accelerometer to indicate 

subject’s activity and body position
8
Methodology
Electrode configuration Reference device
Flow Measurement System with 

a Spirometer Unit and a Fleisch-type 

Heatable Flow Transducer 5530, 

with a Conical Mouthpiece 

(Medikro Oy, Finland)
9
IP electrodes
Methodology
• The simplest and the quickest one

➡ Free breathing registered for 30 seconds.

• To evaluate the impact of longer measurement

➡ Free breathing registered for 2 minutes.

• To check, whether adding various rates and depths of breathing may 

improve the calibration quality meaningly 

➡ Fixed breathing, shallow and deep alternately, 4 times each, 

for three frequencies: 6, 10 and 15 breaths per minute (BPM).

Each calibration procedure was repeated for three body positions:

• supine

• sitting

• standing
Calibration procedures
10
1
2
3
Methodology
Test procedure
11
Consisting of 6 breaths with two subjectively different depths:

➡ normal

➡ deep

for three breathing rates:

➡ 6 BPM

➡ 10 BPM

➡ 15 BPM

and for three body positions:

➡ supine

➡ sitting

➡ standing.
Methodology
Signal processing
12
• Finding breathing phases

• Differentiation

➡ by second-order method.

• Smoothing

➡ Moving average filter with a 500 ms window, striking a balance
between (necessary) removal of the cardiac component and 

(undesirable) deterioration of the signal dynamics.
Methodology
Neural Network
Perceptron architecture with classic Levenberg-Marquardt backpropagation
learning method and default division of data into sets: 

• training (70%) 

• validation (15%) 

• test (15%)
13
Methodology
Postprocessing
14
• Post-hoc smoothing (to preserve the discontinuities that may arise)

➡ Moving average filter with a 125 ms window

• MATLAB 2016b toolboxes
Methodology
Considered calibration approaches ( I )
15
• Simple linear modeling based on flow-related signals (as a reference)

• Neural network approach, trained individually:
➡ single hidden layer with 10 or 20 neurons

➡ two hidden layers of 5 or 10 neurons

• Simple linear modeling and neural network correction, trained individually:
➡ single hidden layer with 10 or 20 neurons

➡ two hidden layers of 5 or 10 neurons

• Simple linear modeling and neural network correction, trained globally:
➡ single hidden layer with 10 or 20 neurons

➡ two hidden layers of 5 or 10 neurons
Results
Comparison for the most accurace calibration approaches in each group ( I )
Calibration Procedure 1 - Free breathing registered for 30 seconds
16
Linear 

model
(reference)
NN, trained
individually
(10, 10)
Linear model
+ NN, trained
individually
(10)
Linear model
+ NN, trained
globally 

(20)
Peak flow
Mean absolute error [ml/s] 240.7 220.9 216.7 174.3
Mean relative error [%] 34.6 24.8
Mean flow
Mean absolute error [ml/s] 187.6 169.8 173.5 145.2
Mean relative error [%] 43.7 30.3
Results
Comparison for the most accurace calibration approaches in each group ( II )
Calibration Procedure 2 - Free breathing registered for 2 minutes
17
Linear 

model
(reference)
NN, trained
individually
(10, 10)
Linear model
+ NN, trained
individually
(10, 10)
Linear model
+ NN, trained
globally 

(10)
Peak flow
Mean absolute error [ml/s] 142.6 175.6 200.1 128.1
Mean relative error [%] 26.6 18.9
Mean flow
Mean absolute error [ml/s] 130.8 147.5 155.2 118.5
Mean relative error [%] 31.6 26.0
Results
Comparison for the most accurace calibration approaches in each group ( III )
Calibration Procedure 3 - Fixed breathing
18
Linear 

model
(reference)
NN, trained
individually
(10, 10)
Linear model
+ NN, trained
individually
(10)
Linear model
+ NN, trained
globally 

(20)
Peak flow
Mean absolute error [ml/s] 144.3 118.4 118.5 133.0
Mean relative error [%] 28.6 20.8
Mean flow
Mean absolute error [ml/s] 102.8 90.5 87.7 84.3
Mean relative error [%] 25.8 21.4
Results
Comparison for the most accurace calibration approaches in each group ( IV )
19
Linear 

model
(reference)
NN, trained
individually
(10, 10)
Linear model
+ NN, trained
individually
(10)
Linear model
+ NN, trained
globally 

(20)
Peak flow
Mean absolute error [ml/s] 144.3 109.1 109.7 123.7
Mean relative error [%] 28.6 18.4
Mean flow
Mean absolute error [ml/s] 102.8 91.4 88.6 85.0
Mean relative error [%] 25.8 21.6
Calibration Procedure 3 - Fixed breathing (with post-hoc smoothing)
Results
Calibration Procedure 3

Fixed breathing 

(with post-hoc smoothing)

Neural network, trained indivudually 

(10, 10)
Bland-Altman plots for the 2 best approaches
20
Calibration Procedure 3

Fixed breathing
Linear modeling and neural network, 

trained indivudually
(10)
Results
Time of analysis
• Individual neural network learning over 14 times faster than global

• Average 36s for individual vs 4min 12s for global learning
• Intel i5 class processor

21
Summary
Recovery of dynamics
22
Summary
80% accuracy of peak and mean flow parameters estimation:

➡ for NN with two hidden layers of 10 neurons, 

➡ based on the data from the longest, the most complex 

calibration procedure with fixed breathing

versus 72.5% for simple linear modeling.

Neural networks trained individually seem to be more reliable and
provide better results than ones trained with a global set.
23
Discussion
Quantitative respiratory parameters
Figures adapted from Clevend Clinic Medical & Wikicommons materials
24
Breathing frequency [ l/min ]
Flow-volume parametersTidal volume [ l ]
Minute ventilation [ l/min ]
Discussion
Quantitative respiratory parameters
25
What about flows during natural breathing?
Breathing frequency [ l/min ]
Flow-volume parametersTidal volume [ l ]
Minute ventilation [ l/min ]
Discussion
Ambulatory respiratory monitoring
Sleep Physiology Sport medicine
• Hypo-, normo-, and
hyperventilation monitoring
in the obese and those with
neuromuscular diseases 

• Cardiorespiratory coupling
analysis
• In-house diagnostics

• Training control

• Determining the level of
exercice
Figures adapted from ”Pulmonary Function Testing” Rolf M. Schlegelmilch, Rüdiger Kramme, Springer, 2011
• Monitoring of breathing
disorders

• Analysis of the effects of
pharmacological treatment
26
Discussion
• Validation of different sequence of
preprocessing and other nonlinear
algorithms.

• Preparation of models established
separately for inspirations and
expirations, and for different depths
of breathing.
• Measurements only in static
conditions, without considering
motion artifacts.
• Neural network approach was not
compared with other nonlinear
methods.
Limitations of this study Further plans
27
Flow parameters derived from 

impedance pneumography after nonlinear
calibration based on neural networks
Porto, February 23, 2017
Marcel Młyńczak, MSc

mlynczak@mchtr.pw.edu.pl

Contenu connexe

En vedette

A review on different technical specifications of respiratory rate monitors
A review on different technical specifications of respiratory rate monitorsA review on different technical specifications of respiratory rate monitors
A review on different technical specifications of respiratory rate monitorseSAT Journals
 
Ppp085 Impedance Educational
Ppp085 Impedance EducationalPpp085 Impedance Educational
Ppp085 Impedance Educationallwannall
 
Jade A Global Witness investigation into Myanmar's "Big State Secret"
Jade A Global Witness investigation into Myanmar's "Big State Secret"Jade A Global Witness investigation into Myanmar's "Big State Secret"
Jade A Global Witness investigation into Myanmar's "Big State Secret"MYO AUNG Myanmar
 
قانون رقم (27) لسنة 2016 العفو العام
قانون رقم (27) لسنة 2016 العفو العامقانون رقم (27) لسنة 2016 العفو العام
قانون رقم (27) لسنة 2016 العفو العامAyad Haris Beden
 
Communication skills
Communication skillsCommunication skills
Communication skillskiran9696
 
Las redes sociales
Las redes socialesLas redes sociales
Las redes socialesXOAN459
 
Ambulatory Devices Measuring Cardiorespiratory Activity with Motion
Ambulatory Devices Measuring Cardiorespiratory Activity with MotionAmbulatory Devices Measuring Cardiorespiratory Activity with Motion
Ambulatory Devices Measuring Cardiorespiratory Activity with MotionMarcel Młyńczak
 
الرياضيات للصف الاول المتوسط
الرياضيات للصف الاول المتوسطالرياضيات للصف الاول المتوسط
الرياضيات للصف الاول المتوسطAyad Haris Beden
 
Patagonia Flooring inauguró su sede Design & Art Center.
Patagonia Flooring inauguró su sede Design & Art Center.Patagonia Flooring inauguró su sede Design & Art Center.
Patagonia Flooring inauguró su sede Design & Art Center.Proyecciones Digitales S. A.
 
Agile test-management-test-rail-lastest
Agile test-management-test-rail-lastestAgile test-management-test-rail-lastest
Agile test-management-test-rail-lastestOnur Baskirt
 
01 tuong phan
01 tuong phan01 tuong phan
01 tuong phanDuy Quang
 
Analysis of cardiorespiratory coupling in athletes in rest using causal approach
Analysis of cardiorespiratory coupling in athletes in rest using causal approachAnalysis of cardiorespiratory coupling in athletes in rest using causal approach
Analysis of cardiorespiratory coupling in athletes in rest using causal approachMarcel Młyńczak
 
Unit 4 biomedical
Unit 4 biomedicalUnit 4 biomedical
Unit 4 biomedicalAnu Antony
 
Advanced+Haemodynamic+Monitoring+and+support+-+part+1+vs+1+0.pptx
Advanced+Haemodynamic+Monitoring+and+support+-+part+1+vs+1+0.pptxAdvanced+Haemodynamic+Monitoring+and+support+-+part+1+vs+1+0.pptx
Advanced+Haemodynamic+Monitoring+and+support+-+part+1+vs+1+0.pptxcicmelearning
 
medical electronics 5 module ppt
medical electronics 5 module pptmedical electronics 5 module ppt
medical electronics 5 module pptAnu Antony
 

En vedette (17)

Physiological measurements
Physiological measurementsPhysiological measurements
Physiological measurements
 
A review on different technical specifications of respiratory rate monitors
A review on different technical specifications of respiratory rate monitorsA review on different technical specifications of respiratory rate monitors
A review on different technical specifications of respiratory rate monitors
 
Ppp085 Impedance Educational
Ppp085 Impedance EducationalPpp085 Impedance Educational
Ppp085 Impedance Educational
 
Jade A Global Witness investigation into Myanmar's "Big State Secret"
Jade A Global Witness investigation into Myanmar's "Big State Secret"Jade A Global Witness investigation into Myanmar's "Big State Secret"
Jade A Global Witness investigation into Myanmar's "Big State Secret"
 
قانون رقم (27) لسنة 2016 العفو العام
قانون رقم (27) لسنة 2016 العفو العامقانون رقم (27) لسنة 2016 العفو العام
قانون رقم (27) لسنة 2016 العفو العام
 
Communication skills
Communication skillsCommunication skills
Communication skills
 
Las redes sociales
Las redes socialesLas redes sociales
Las redes sociales
 
Ambulatory Devices Measuring Cardiorespiratory Activity with Motion
Ambulatory Devices Measuring Cardiorespiratory Activity with MotionAmbulatory Devices Measuring Cardiorespiratory Activity with Motion
Ambulatory Devices Measuring Cardiorespiratory Activity with Motion
 
الرياضيات للصف الاول المتوسط
الرياضيات للصف الاول المتوسطالرياضيات للصف الاول المتوسط
الرياضيات للصف الاول المتوسط
 
Patagonia Flooring inauguró su sede Design & Art Center.
Patagonia Flooring inauguró su sede Design & Art Center.Patagonia Flooring inauguró su sede Design & Art Center.
Patagonia Flooring inauguró su sede Design & Art Center.
 
Agile test-management-test-rail-lastest
Agile test-management-test-rail-lastestAgile test-management-test-rail-lastest
Agile test-management-test-rail-lastest
 
01 tuong phan
01 tuong phan01 tuong phan
01 tuong phan
 
Analysis of cardiorespiratory coupling in athletes in rest using causal approach
Analysis of cardiorespiratory coupling in athletes in rest using causal approachAnalysis of cardiorespiratory coupling in athletes in rest using causal approach
Analysis of cardiorespiratory coupling in athletes in rest using causal approach
 
Unit 4 biomedical
Unit 4 biomedicalUnit 4 biomedical
Unit 4 biomedical
 
Advanced+Haemodynamic+Monitoring+and+support+-+part+1+vs+1+0.pptx
Advanced+Haemodynamic+Monitoring+and+support+-+part+1+vs+1+0.pptxAdvanced+Haemodynamic+Monitoring+and+support+-+part+1+vs+1+0.pptx
Advanced+Haemodynamic+Monitoring+and+support+-+part+1+vs+1+0.pptx
 
Tiang pancang
Tiang pancangTiang pancang
Tiang pancang
 
medical electronics 5 module ppt
medical electronics 5 module pptmedical electronics 5 module ppt
medical electronics 5 module ppt
 

Similaire à Flow Parameters Derived from Impedance Pneumography after Nonlinear Calibration based on Neural Networks

DESIGN AND IMPLEMENTATION OF EMBEDDED MONITOR SYSTEM FOR DETECTION OF A PATIE...
DESIGN AND IMPLEMENTATION OF EMBEDDED MONITOR SYSTEM FOR DETECTION OF A PATIE...DESIGN AND IMPLEMENTATION OF EMBEDDED MONITOR SYSTEM FOR DETECTION OF A PATIE...
DESIGN AND IMPLEMENTATION OF EMBEDDED MONITOR SYSTEM FOR DETECTION OF A PATIE...Abhishek Somayaji
 
Cardiopulmonary exercise testing
Cardiopulmonary exercise testingCardiopulmonary exercise testing
Cardiopulmonary exercise testingAvinash Arke
 
Cardiopulmonary exercise testing
Cardiopulmonary exercise testingCardiopulmonary exercise testing
Cardiopulmonary exercise testingAvinash Arke
 
The Future of Metabolic Phenotyping Using data bandwidth to maximize N, analy...
The Future of Metabolic Phenotyping Using data bandwidth to maximize N, analy...The Future of Metabolic Phenotyping Using data bandwidth to maximize N, analy...
The Future of Metabolic Phenotyping Using data bandwidth to maximize N, analy...InsideScientific
 
An algorithm for obtaining the frequency and the times of respiratory phases...
An algorithm for obtaining the frequency and the times of  respiratory phases...An algorithm for obtaining the frequency and the times of  respiratory phases...
An algorithm for obtaining the frequency and the times of respiratory phases...IJECEIAES
 
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES Touradj Ebrahimi
 
20211008 修論中間発表
20211008 修論中間発表20211008 修論中間発表
20211008 修論中間発表Tomoya Koike
 
Galaxy For Productive Sleep Diagnosis and Lab Management
Galaxy For Productive Sleep Diagnosis and Lab ManagementGalaxy For Productive Sleep Diagnosis and Lab Management
Galaxy For Productive Sleep Diagnosis and Lab ManagementAnand Kumar
 
ELM and K-nn machine learning in classification of Breath sounds signals
ELM and K-nn machine learning in classification  of Breath sounds signals  ELM and K-nn machine learning in classification  of Breath sounds signals
ELM and K-nn machine learning in classification of Breath sounds signals IJECEIAES
 
IEEE Bio medical engineering 2016 Title and Abstract
IEEE Bio medical engineering  2016 Title and Abstract IEEE Bio medical engineering  2016 Title and Abstract
IEEE Bio medical engineering 2016 Title and Abstract tsysglobalsolutions
 
Pressure Prediction System in Lung Circuit using Deep Learning and Machine Le...
Pressure Prediction System in Lung Circuit using Deep Learning and Machine Le...Pressure Prediction System in Lung Circuit using Deep Learning and Machine Le...
Pressure Prediction System in Lung Circuit using Deep Learning and Machine Le...IRJET Journal
 
Respiratory Rate Measurement
Respiratory Rate MeasurementRespiratory Rate Measurement
Respiratory Rate Measurementgoverdhan765
 
Evaluation of Thresholding Based Noncontact Respiration Rate Monitoring using...
Evaluation of Thresholding Based Noncontact Respiration Rate Monitoring using...Evaluation of Thresholding Based Noncontact Respiration Rate Monitoring using...
Evaluation of Thresholding Based Noncontact Respiration Rate Monitoring using...IRJESJOURNAL
 
Ab experiments of fluid flow in polymer microchannel
Ab experiments of fluid flow in polymer microchannelAb experiments of fluid flow in polymer microchannel
Ab experiments of fluid flow in polymer microchannelShaelMalik
 
Basic modes of mechanical ventilation
Basic modes of mechanical ventilationBasic modes of mechanical ventilation
Basic modes of mechanical ventilationdrsangeet
 

Similaire à Flow Parameters Derived from Impedance Pneumography after Nonlinear Calibration based on Neural Networks (20)

DESIGN AND IMPLEMENTATION OF EMBEDDED MONITOR SYSTEM FOR DETECTION OF A PATIE...
DESIGN AND IMPLEMENTATION OF EMBEDDED MONITOR SYSTEM FOR DETECTION OF A PATIE...DESIGN AND IMPLEMENTATION OF EMBEDDED MONITOR SYSTEM FOR DETECTION OF A PATIE...
DESIGN AND IMPLEMENTATION OF EMBEDDED MONITOR SYSTEM FOR DETECTION OF A PATIE...
 
Cardiopulmonary exercise testing
Cardiopulmonary exercise testingCardiopulmonary exercise testing
Cardiopulmonary exercise testing
 
Cardiopulmonary exercise testing
Cardiopulmonary exercise testingCardiopulmonary exercise testing
Cardiopulmonary exercise testing
 
The Future of Metabolic Phenotyping Using data bandwidth to maximize N, analy...
The Future of Metabolic Phenotyping Using data bandwidth to maximize N, analy...The Future of Metabolic Phenotyping Using data bandwidth to maximize N, analy...
The Future of Metabolic Phenotyping Using data bandwidth to maximize N, analy...
 
An algorithm for obtaining the frequency and the times of respiratory phases...
An algorithm for obtaining the frequency and the times of  respiratory phases...An algorithm for obtaining the frequency and the times of  respiratory phases...
An algorithm for obtaining the frequency and the times of respiratory phases...
 
sleep apnea detection
sleep apnea detectionsleep apnea detection
sleep apnea detection
 
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES
RECENT ADVANCES IN BRAIN-COMPUTER INTERFACES
 
Respiratory gating
Respiratory gatingRespiratory gating
Respiratory gating
 
20211008 修論中間発表
20211008 修論中間発表20211008 修論中間発表
20211008 修論中間発表
 
Galaxy For Productive Sleep Diagnosis and Lab Management
Galaxy For Productive Sleep Diagnosis and Lab ManagementGalaxy For Productive Sleep Diagnosis and Lab Management
Galaxy For Productive Sleep Diagnosis and Lab Management
 
ELM and K-nn machine learning in classification of Breath sounds signals
ELM and K-nn machine learning in classification  of Breath sounds signals  ELM and K-nn machine learning in classification  of Breath sounds signals
ELM and K-nn machine learning in classification of Breath sounds signals
 
Close loop Ventilation
Close loop VentilationClose loop Ventilation
Close loop Ventilation
 
IEEE Bio medical engineering 2016 Title and Abstract
IEEE Bio medical engineering  2016 Title and Abstract IEEE Bio medical engineering  2016 Title and Abstract
IEEE Bio medical engineering 2016 Title and Abstract
 
Pressure Prediction System in Lung Circuit using Deep Learning and Machine Le...
Pressure Prediction System in Lung Circuit using Deep Learning and Machine Le...Pressure Prediction System in Lung Circuit using Deep Learning and Machine Le...
Pressure Prediction System in Lung Circuit using Deep Learning and Machine Le...
 
Respiratory Rate Measurement
Respiratory Rate MeasurementRespiratory Rate Measurement
Respiratory Rate Measurement
 
Weaning: The art and science
Weaning: The art and scienceWeaning: The art and science
Weaning: The art and science
 
Evaluation of Thresholding Based Noncontact Respiration Rate Monitoring using...
Evaluation of Thresholding Based Noncontact Respiration Rate Monitoring using...Evaluation of Thresholding Based Noncontact Respiration Rate Monitoring using...
Evaluation of Thresholding Based Noncontact Respiration Rate Monitoring using...
 
Ab experiments of fluid flow in polymer microchannel
Ab experiments of fluid flow in polymer microchannelAb experiments of fluid flow in polymer microchannel
Ab experiments of fluid flow in polymer microchannel
 
Fo3610221025
Fo3610221025Fo3610221025
Fo3610221025
 
Basic modes of mechanical ventilation
Basic modes of mechanical ventilationBasic modes of mechanical ventilation
Basic modes of mechanical ventilation
 

Dernier

(Rocky) Jaipur Call Girl - 09521753030 Escorts Service 50% Off with Cash ON D...
(Rocky) Jaipur Call Girl - 09521753030 Escorts Service 50% Off with Cash ON D...(Rocky) Jaipur Call Girl - 09521753030 Escorts Service 50% Off with Cash ON D...
(Rocky) Jaipur Call Girl - 09521753030 Escorts Service 50% Off with Cash ON D...indiancallgirl4rent
 
All Time Service Available Call Girls Marine Drive 📳 9820252231 For 18+ VIP C...
All Time Service Available Call Girls Marine Drive 📳 9820252231 For 18+ VIP C...All Time Service Available Call Girls Marine Drive 📳 9820252231 For 18+ VIP C...
All Time Service Available Call Girls Marine Drive 📳 9820252231 For 18+ VIP C...Arohi Goyal
 
Call Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service AvailableDipal Arora
 
Night 7k to 12k Navi Mumbai Call Girl Photo 👉 BOOK NOW 9833363713 👈 ♀️ night ...
Night 7k to 12k Navi Mumbai Call Girl Photo 👉 BOOK NOW 9833363713 👈 ♀️ night ...Night 7k to 12k Navi Mumbai Call Girl Photo 👉 BOOK NOW 9833363713 👈 ♀️ night ...
Night 7k to 12k Navi Mumbai Call Girl Photo 👉 BOOK NOW 9833363713 👈 ♀️ night ...aartirawatdelhi
 
Call Girls Bhubaneswar Just Call 9907093804 Top Class Call Girl Service Avail...
Call Girls Bhubaneswar Just Call 9907093804 Top Class Call Girl Service Avail...Call Girls Bhubaneswar Just Call 9907093804 Top Class Call Girl Service Avail...
Call Girls Bhubaneswar Just Call 9907093804 Top Class Call Girl Service Avail...Dipal Arora
 
Lucknow Call girls - 8800925952 - 24x7 service with hotel room
Lucknow Call girls - 8800925952 - 24x7 service with hotel roomLucknow Call girls - 8800925952 - 24x7 service with hotel room
Lucknow Call girls - 8800925952 - 24x7 service with hotel roomdiscovermytutordmt
 
Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...
Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...
Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...Call Girls in Nagpur High Profile
 
Call Girls Horamavu WhatsApp Number 7001035870 Meeting With Bangalore Escorts
Call Girls Horamavu WhatsApp Number 7001035870 Meeting With Bangalore EscortsCall Girls Horamavu WhatsApp Number 7001035870 Meeting With Bangalore Escorts
Call Girls Horamavu WhatsApp Number 7001035870 Meeting With Bangalore Escortsvidya singh
 
Top Quality Call Girl Service Kalyanpur 6378878445 Available Call Girls Any Time
Top Quality Call Girl Service Kalyanpur 6378878445 Available Call Girls Any TimeTop Quality Call Girl Service Kalyanpur 6378878445 Available Call Girls Any Time
Top Quality Call Girl Service Kalyanpur 6378878445 Available Call Girls Any TimeCall Girls Delhi
 
Top Rated Hyderabad Call Girls Erragadda ⟟ 6297143586 ⟟ Call Me For Genuine ...
Top Rated  Hyderabad Call Girls Erragadda ⟟ 6297143586 ⟟ Call Me For Genuine ...Top Rated  Hyderabad Call Girls Erragadda ⟟ 6297143586 ⟟ Call Me For Genuine ...
Top Rated Hyderabad Call Girls Erragadda ⟟ 6297143586 ⟟ Call Me For Genuine ...chandars293
 
Call Girls Nagpur Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Nagpur Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Nagpur Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Nagpur Just Call 9907093804 Top Class Call Girl Service AvailableDipal Arora
 
Call Girls Kochi Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Kochi Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Kochi Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Kochi Just Call 9907093804 Top Class Call Girl Service AvailableDipal Arora
 
Call Girls Gwalior Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Gwalior Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Gwalior Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Gwalior Just Call 9907093804 Top Class Call Girl Service AvailableDipal Arora
 
Call Girls Coimbatore Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Coimbatore Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Coimbatore Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Coimbatore Just Call 9907093804 Top Class Call Girl Service AvailableDipal Arora
 
Call Girls Visakhapatnam Just Call 9907093804 Top Class Call Girl Service Ava...
Call Girls Visakhapatnam Just Call 9907093804 Top Class Call Girl Service Ava...Call Girls Visakhapatnam Just Call 9907093804 Top Class Call Girl Service Ava...
Call Girls Visakhapatnam Just Call 9907093804 Top Class Call Girl Service Ava...Dipal Arora
 
Call Girls Bareilly Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Bareilly Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Bareilly Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Bareilly Just Call 9907093804 Top Class Call Girl Service AvailableDipal Arora
 
Night 7k to 12k Chennai City Center Call Girls 👉👉 7427069034⭐⭐ 100% Genuine E...
Night 7k to 12k Chennai City Center Call Girls 👉👉 7427069034⭐⭐ 100% Genuine E...Night 7k to 12k Chennai City Center Call Girls 👉👉 7427069034⭐⭐ 100% Genuine E...
Night 7k to 12k Chennai City Center Call Girls 👉👉 7427069034⭐⭐ 100% Genuine E...hotbabesbook
 
VIP Hyderabad Call Girls Bahadurpally 7877925207 ₹5000 To 25K With AC Room 💚😋
VIP Hyderabad Call Girls Bahadurpally 7877925207 ₹5000 To 25K With AC Room 💚😋VIP Hyderabad Call Girls Bahadurpally 7877925207 ₹5000 To 25K With AC Room 💚😋
VIP Hyderabad Call Girls Bahadurpally 7877925207 ₹5000 To 25K With AC Room 💚😋TANUJA PANDEY
 
The Most Attractive Hyderabad Call Girls Kothapet 𖠋 6297143586 𖠋 Will You Mis...
The Most Attractive Hyderabad Call Girls Kothapet 𖠋 6297143586 𖠋 Will You Mis...The Most Attractive Hyderabad Call Girls Kothapet 𖠋 6297143586 𖠋 Will You Mis...
The Most Attractive Hyderabad Call Girls Kothapet 𖠋 6297143586 𖠋 Will You Mis...chandars293
 
Call Girls Jabalpur Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Jabalpur Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Jabalpur Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Jabalpur Just Call 9907093804 Top Class Call Girl Service AvailableDipal Arora
 

Dernier (20)

(Rocky) Jaipur Call Girl - 09521753030 Escorts Service 50% Off with Cash ON D...
(Rocky) Jaipur Call Girl - 09521753030 Escorts Service 50% Off with Cash ON D...(Rocky) Jaipur Call Girl - 09521753030 Escorts Service 50% Off with Cash ON D...
(Rocky) Jaipur Call Girl - 09521753030 Escorts Service 50% Off with Cash ON D...
 
All Time Service Available Call Girls Marine Drive 📳 9820252231 For 18+ VIP C...
All Time Service Available Call Girls Marine Drive 📳 9820252231 For 18+ VIP C...All Time Service Available Call Girls Marine Drive 📳 9820252231 For 18+ VIP C...
All Time Service Available Call Girls Marine Drive 📳 9820252231 For 18+ VIP C...
 
Call Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Ludhiana Just Call 9907093804 Top Class Call Girl Service Available
 
Night 7k to 12k Navi Mumbai Call Girl Photo 👉 BOOK NOW 9833363713 👈 ♀️ night ...
Night 7k to 12k Navi Mumbai Call Girl Photo 👉 BOOK NOW 9833363713 👈 ♀️ night ...Night 7k to 12k Navi Mumbai Call Girl Photo 👉 BOOK NOW 9833363713 👈 ♀️ night ...
Night 7k to 12k Navi Mumbai Call Girl Photo 👉 BOOK NOW 9833363713 👈 ♀️ night ...
 
Call Girls Bhubaneswar Just Call 9907093804 Top Class Call Girl Service Avail...
Call Girls Bhubaneswar Just Call 9907093804 Top Class Call Girl Service Avail...Call Girls Bhubaneswar Just Call 9907093804 Top Class Call Girl Service Avail...
Call Girls Bhubaneswar Just Call 9907093804 Top Class Call Girl Service Avail...
 
Lucknow Call girls - 8800925952 - 24x7 service with hotel room
Lucknow Call girls - 8800925952 - 24x7 service with hotel roomLucknow Call girls - 8800925952 - 24x7 service with hotel room
Lucknow Call girls - 8800925952 - 24x7 service with hotel room
 
Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...
Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...
Book Paid Powai Call Girls Mumbai 𖠋 9930245274 𖠋Low Budget Full Independent H...
 
Call Girls Horamavu WhatsApp Number 7001035870 Meeting With Bangalore Escorts
Call Girls Horamavu WhatsApp Number 7001035870 Meeting With Bangalore EscortsCall Girls Horamavu WhatsApp Number 7001035870 Meeting With Bangalore Escorts
Call Girls Horamavu WhatsApp Number 7001035870 Meeting With Bangalore Escorts
 
Top Quality Call Girl Service Kalyanpur 6378878445 Available Call Girls Any Time
Top Quality Call Girl Service Kalyanpur 6378878445 Available Call Girls Any TimeTop Quality Call Girl Service Kalyanpur 6378878445 Available Call Girls Any Time
Top Quality Call Girl Service Kalyanpur 6378878445 Available Call Girls Any Time
 
Top Rated Hyderabad Call Girls Erragadda ⟟ 6297143586 ⟟ Call Me For Genuine ...
Top Rated  Hyderabad Call Girls Erragadda ⟟ 6297143586 ⟟ Call Me For Genuine ...Top Rated  Hyderabad Call Girls Erragadda ⟟ 6297143586 ⟟ Call Me For Genuine ...
Top Rated Hyderabad Call Girls Erragadda ⟟ 6297143586 ⟟ Call Me For Genuine ...
 
Call Girls Nagpur Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Nagpur Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Nagpur Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Nagpur Just Call 9907093804 Top Class Call Girl Service Available
 
Call Girls Kochi Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Kochi Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Kochi Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Kochi Just Call 9907093804 Top Class Call Girl Service Available
 
Call Girls Gwalior Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Gwalior Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Gwalior Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Gwalior Just Call 9907093804 Top Class Call Girl Service Available
 
Call Girls Coimbatore Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Coimbatore Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Coimbatore Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Coimbatore Just Call 9907093804 Top Class Call Girl Service Available
 
Call Girls Visakhapatnam Just Call 9907093804 Top Class Call Girl Service Ava...
Call Girls Visakhapatnam Just Call 9907093804 Top Class Call Girl Service Ava...Call Girls Visakhapatnam Just Call 9907093804 Top Class Call Girl Service Ava...
Call Girls Visakhapatnam Just Call 9907093804 Top Class Call Girl Service Ava...
 
Call Girls Bareilly Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Bareilly Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Bareilly Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Bareilly Just Call 9907093804 Top Class Call Girl Service Available
 
Night 7k to 12k Chennai City Center Call Girls 👉👉 7427069034⭐⭐ 100% Genuine E...
Night 7k to 12k Chennai City Center Call Girls 👉👉 7427069034⭐⭐ 100% Genuine E...Night 7k to 12k Chennai City Center Call Girls 👉👉 7427069034⭐⭐ 100% Genuine E...
Night 7k to 12k Chennai City Center Call Girls 👉👉 7427069034⭐⭐ 100% Genuine E...
 
VIP Hyderabad Call Girls Bahadurpally 7877925207 ₹5000 To 25K With AC Room 💚😋
VIP Hyderabad Call Girls Bahadurpally 7877925207 ₹5000 To 25K With AC Room 💚😋VIP Hyderabad Call Girls Bahadurpally 7877925207 ₹5000 To 25K With AC Room 💚😋
VIP Hyderabad Call Girls Bahadurpally 7877925207 ₹5000 To 25K With AC Room 💚😋
 
The Most Attractive Hyderabad Call Girls Kothapet 𖠋 6297143586 𖠋 Will You Mis...
The Most Attractive Hyderabad Call Girls Kothapet 𖠋 6297143586 𖠋 Will You Mis...The Most Attractive Hyderabad Call Girls Kothapet 𖠋 6297143586 𖠋 Will You Mis...
The Most Attractive Hyderabad Call Girls Kothapet 𖠋 6297143586 𖠋 Will You Mis...
 
Call Girls Jabalpur Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Jabalpur Just Call 9907093804 Top Class Call Girl Service AvailableCall Girls Jabalpur Just Call 9907093804 Top Class Call Girl Service Available
Call Girls Jabalpur Just Call 9907093804 Top Class Call Girl Service Available
 

Flow Parameters Derived from Impedance Pneumography after Nonlinear Calibration based on Neural Networks

  • 1. Flow parameters derived from 
 impedance pneumography after nonlinear calibration based on neural networks Marcel Młyńczak, MSc, Gerard Cybulski, PhD Warsaw University of Technology, Faculty of Mechatronics, Institute of Metrology and Biomedical Engineering Porto, February 23, 2017
  • 2. Introduction Spirometry / Pneumotachometry • The most reliable methods • Direct measurements • The need for using a face mask or mouthpiece with nose clip • Uncomfortable during exercises, sleep and for children Figures adapted from chat.stackexchange.com and legio24.pl Mesh grid of 
 known pneumatic resistance 2
  • 3. Introduction Impedance pneumography Basic idea Changes of transthoracic bioimpedance are 
 connected with changes of the amount of air in the lungs. Method of measurements The nature of the IP signals Carried out using tetrapolar method. Volume-related. Calibration Simple linear model provides the best accuracy of volume parameters for specific electrode configuration. Flow measurements Obtained from differentiated 
 impedance pneumography signals. 3
  • 4. The Problem The flow IP signals are usually underestimated 4
  • 5. The Problem Possible solution Nonlinear signal calibration? To recover the dynamics of the flow signal artificially by using 
 nonlinear processing of the signal. 5
  • 6. Objectives • Improvement of the accuracy of flow parameters: ➡ peak flow values, separately for inspirations and expirations ➡ mean flow values, as above calculated by impedance pneumography, using neural network approach. • Assessment of which neural network structure is the best for the flow parameters analysis. • Finding the best calibration procedure in terms of flow-related analysis. 6
  • 7. Methodology Subjects - generally healthy students, 10 males Min Avg Max Weight [kg] 65.0 77.4 100.0 Height [cm] 171.0 179.3 187.0 BMI 20.75 24.14 33.41 Age 19 23 27 7
  • 8. Methodology Pneumonitor 2 • ECG signal to estimate heart rate and tachogram • Impedance signal relating to 
 main breathing activity • Portable • Recording on SD card • Rechargeable battery • Motion signal from 3-axis accelerometer to indicate 
 subject’s activity and body position 8
  • 9. Methodology Electrode configuration Reference device Flow Measurement System with 
 a Spirometer Unit and a Fleisch-type 
 Heatable Flow Transducer 5530, 
 with a Conical Mouthpiece 
 (Medikro Oy, Finland) 9 IP electrodes
  • 10. Methodology • The simplest and the quickest one ➡ Free breathing registered for 30 seconds. • To evaluate the impact of longer measurement ➡ Free breathing registered for 2 minutes. • To check, whether adding various rates and depths of breathing may 
 improve the calibration quality meaningly ➡ Fixed breathing, shallow and deep alternately, 4 times each, 
 for three frequencies: 6, 10 and 15 breaths per minute (BPM). Each calibration procedure was repeated for three body positions: • supine • sitting • standing Calibration procedures 10 1 2 3
  • 11. Methodology Test procedure 11 Consisting of 6 breaths with two subjectively different depths: ➡ normal ➡ deep for three breathing rates: ➡ 6 BPM ➡ 10 BPM ➡ 15 BPM and for three body positions: ➡ supine ➡ sitting ➡ standing.
  • 12. Methodology Signal processing 12 • Finding breathing phases • Differentiation ➡ by second-order method. • Smoothing ➡ Moving average filter with a 500 ms window, striking a balance between (necessary) removal of the cardiac component and 
 (undesirable) deterioration of the signal dynamics.
  • 13. Methodology Neural Network Perceptron architecture with classic Levenberg-Marquardt backpropagation learning method and default division of data into sets: • training (70%) • validation (15%) • test (15%) 13
  • 14. Methodology Postprocessing 14 • Post-hoc smoothing (to preserve the discontinuities that may arise) ➡ Moving average filter with a 125 ms window • MATLAB 2016b toolboxes
  • 15. Methodology Considered calibration approaches ( I ) 15 • Simple linear modeling based on flow-related signals (as a reference) • Neural network approach, trained individually: ➡ single hidden layer with 10 or 20 neurons ➡ two hidden layers of 5 or 10 neurons • Simple linear modeling and neural network correction, trained individually: ➡ single hidden layer with 10 or 20 neurons ➡ two hidden layers of 5 or 10 neurons • Simple linear modeling and neural network correction, trained globally: ➡ single hidden layer with 10 or 20 neurons ➡ two hidden layers of 5 or 10 neurons
  • 16. Results Comparison for the most accurace calibration approaches in each group ( I ) Calibration Procedure 1 - Free breathing registered for 30 seconds 16 Linear 
 model (reference) NN, trained individually (10, 10) Linear model + NN, trained individually (10) Linear model + NN, trained globally 
 (20) Peak flow Mean absolute error [ml/s] 240.7 220.9 216.7 174.3 Mean relative error [%] 34.6 24.8 Mean flow Mean absolute error [ml/s] 187.6 169.8 173.5 145.2 Mean relative error [%] 43.7 30.3
  • 17. Results Comparison for the most accurace calibration approaches in each group ( II ) Calibration Procedure 2 - Free breathing registered for 2 minutes 17 Linear 
 model (reference) NN, trained individually (10, 10) Linear model + NN, trained individually (10, 10) Linear model + NN, trained globally 
 (10) Peak flow Mean absolute error [ml/s] 142.6 175.6 200.1 128.1 Mean relative error [%] 26.6 18.9 Mean flow Mean absolute error [ml/s] 130.8 147.5 155.2 118.5 Mean relative error [%] 31.6 26.0
  • 18. Results Comparison for the most accurace calibration approaches in each group ( III ) Calibration Procedure 3 - Fixed breathing 18 Linear 
 model (reference) NN, trained individually (10, 10) Linear model + NN, trained individually (10) Linear model + NN, trained globally 
 (20) Peak flow Mean absolute error [ml/s] 144.3 118.4 118.5 133.0 Mean relative error [%] 28.6 20.8 Mean flow Mean absolute error [ml/s] 102.8 90.5 87.7 84.3 Mean relative error [%] 25.8 21.4
  • 19. Results Comparison for the most accurace calibration approaches in each group ( IV ) 19 Linear 
 model (reference) NN, trained individually (10, 10) Linear model + NN, trained individually (10) Linear model + NN, trained globally 
 (20) Peak flow Mean absolute error [ml/s] 144.3 109.1 109.7 123.7 Mean relative error [%] 28.6 18.4 Mean flow Mean absolute error [ml/s] 102.8 91.4 88.6 85.0 Mean relative error [%] 25.8 21.6 Calibration Procedure 3 - Fixed breathing (with post-hoc smoothing)
  • 20. Results Calibration Procedure 3
 Fixed breathing 
 (with post-hoc smoothing)
 Neural network, trained indivudually 
 (10, 10) Bland-Altman plots for the 2 best approaches 20 Calibration Procedure 3
 Fixed breathing Linear modeling and neural network, 
 trained indivudually (10)
  • 21. Results Time of analysis • Individual neural network learning over 14 times faster than global • Average 36s for individual vs 4min 12s for global learning • Intel i5 class processor 21
  • 23. Summary 80% accuracy of peak and mean flow parameters estimation: ➡ for NN with two hidden layers of 10 neurons, ➡ based on the data from the longest, the most complex 
 calibration procedure with fixed breathing versus 72.5% for simple linear modeling. Neural networks trained individually seem to be more reliable and provide better results than ones trained with a global set. 23
  • 24. Discussion Quantitative respiratory parameters Figures adapted from Clevend Clinic Medical & Wikicommons materials 24 Breathing frequency [ l/min ] Flow-volume parametersTidal volume [ l ] Minute ventilation [ l/min ]
  • 25. Discussion Quantitative respiratory parameters 25 What about flows during natural breathing? Breathing frequency [ l/min ] Flow-volume parametersTidal volume [ l ] Minute ventilation [ l/min ]
  • 26. Discussion Ambulatory respiratory monitoring Sleep Physiology Sport medicine • Hypo-, normo-, and hyperventilation monitoring in the obese and those with neuromuscular diseases • Cardiorespiratory coupling analysis • In-house diagnostics • Training control • Determining the level of exercice Figures adapted from ”Pulmonary Function Testing” Rolf M. Schlegelmilch, Rüdiger Kramme, Springer, 2011 • Monitoring of breathing disorders • Analysis of the effects of pharmacological treatment 26
  • 27. Discussion • Validation of different sequence of preprocessing and other nonlinear algorithms. • Preparation of models established separately for inspirations and expirations, and for different depths of breathing. • Measurements only in static conditions, without considering motion artifacts. • Neural network approach was not compared with other nonlinear methods. Limitations of this study Further plans 27
  • 28. Flow parameters derived from 
 impedance pneumography after nonlinear calibration based on neural networks Porto, February 23, 2017 Marcel Młyńczak, MSc mlynczak@mchtr.pw.edu.pl