S. Mohamad-Samuri1, M. Mahfouf1, M. Dena

1

• S. Mohamad-Samuri1, M. Mahfouf1, M. Denaï2, J.J.
  Ross3 and G.H. Mills3
• 1 Dept of Automatic Control and Systems Eng,
  University of Sheffield, Sheffield, UK
• 2 School of Science and Eng, Teesside University,
  Middlesbrough, UK
• 3Dept of Critical Care and Anaesthesia, Northern
  General Hospital, Sheffield, UK

2
Outline

• Absolute Electrical Impedance
• Tomography (aEIT) of the lungs Overview

• Clinical trial of aEIT

• Modelling of Mean End Expiratory lung
• Volumes (MEEV) Neuro-Fuzzy Approach

• Overview of the SOPAVent

• Coupling aEIT and SOPAVent

• Conclusion and future work

3
aEIT of the Lungs Overview

• Hardware

Research prototype
Drive pattern Adjacent
No. of electrodes 8
Frequencies 302 kHz 1.6 MHz
Technology Digital
Date 2000
Commercial EIT System
ESCTAIC 6-9 October 2010, Amsterdam Netherland
4
aEIT of the Lungs Overview

• Steps to determine the absolute lung resistivity

Collection of impedance measurements from the
subject
Simulate reference image data using 3D thoracic
model
The modelled data are compared with the real
measurements over a pre-determined region of
interest
The value of lung resistivity which minimizes the
mean difference between these data sets is
returned as the value of the absolute lung
resistivity
ESCTAIC 6-9 October 2010, Amsterdam Netherland
4
5
aEIT of the Lungs Overview

• Absolute lung resistivity flow chart

current injection and EIT data measurement
Real EIT data
Patient
Model predicted EIT data
Y
Match ?
N
3D finite difference model adjusted to the real
EIT data
Absolute lung resistivity
ESCTAIC 6-9 October 2010, Amsterdam Netherland
6
Clinical trial of aEIT

• Objective

To validate the ability of the Mk3.5 aEIT system
to reflect ventilator settings (PEEP)-induced
changes on the lung absolute volume and
resistivity in ITU patients

• Methods

Equipment Sheffield EIT MK 3.5 , with 8 Skintact Premier ECG Electrodes
No. of Subjects Eight (8) ITU patients
Position Supine
Extracted aEIT measurement Mean End Expiratory Lung Volume (MEEV)
ESCTAIC 6-9 October 2010, Amsterdam Netherland
7
Clinical trial of aEIT

• Demographic information of the patients

Gender Height (cm) Chest Circumference (cm) Elipse ratio
Mean S.D 7 males, 1 female 169.8 6.41 94.6 4.10 1.38 0.09

• An example of patients ventilator settings, MEEV
  and MVT

Day Ventilation mode Ventilator settings Ventilator settings Ventilator settings Ventilator settings Ventilator settings Ventilator settings EIT Outputs EIT Outputs
Day Ventilation mode ?ASB PEEP (cmH2O) Pinsp (cmH2O) FiO2 () VT (litre) MV (litre) MEEV (litre) MVT (litre)
1 BIPAP 0 12 30 55 0.65 11.6 6.21 0.77
2 BIPAP 12 12 30 40 0.72 13.3 5.31 0.62
2 BIPAP 12 10 22 40 0.66 10.4 4.72 0.8
3 BIPAP 10 10 20 50 1.11 9.5 3.59 1.12
4 CPAP 3 10 20 45 0.92 13.6 4.36 0.82
ESCTAIC 6-9 October 2010, Amsterdam Netherland
7
8
Clinical trial of aEIT

• Lung absolute resistivity and air volume measured
  by aEIT at different PEEP levels on an ITU patient

ESCTAIC 6-9 October 2010, Amsterdam Netherland
8
9
ANFIS modelling of MEEV

• What is ANFIS?

• Stands for Adaptive Neural-Fuzzy Inference
  Systems 1

• Hybrid system that operates on both linguistic
  descriptions
• of the variables and the numeric values

• Neural-Fuzzy model incorporate human expertise
  as well
• as adapt itself through repeated learning

1 Jang, J. S. R. (1993). "ANFIS
adaptive-network-based fuzzy inference system."
Systems, Man and Cybernetics, IEEE Transactions
on 23(3) 665-685.
ESCTAIC 6-9 October 2010, Amsterdam Netherland
9
10
ANFIS modelling of MEEV

• ANFIS architecture

• ANFIS consists of a set of TSK-type fuzzy
  IF-THEN rules

ESCTAIC 6-9 October 2010, Amsterdam Netherland
10
11
ANFIS modelling of MEEV

• ANFIS model structure

ANFIS Structure 6 inputs, 1 output 4 membership
functions for each input 5 fuzzy rules
example of Gaussian MF
ESCTAIC 6-9 October 2010, Amsterdam Netherland
11
12
ANFIS modelling of MEEV

• Results

ANFIS architecture has demonstrated a good
performance in modelling the MEEV
ESCTAIC 6-9 October 2010, Amsterdam Netherland
12
13
Overview of SOPAVent

• What is SOPAVent?

• Simulation of Patients under Artificial
  Ventilation

• The model represents the exchange of O2 and CO2
  in the lungs and
• tissues together with their transport through
  the circulatory system
• based on respiratory physiology and mass
  balance equations

• The model uses a compartmental structure, where
  the circulatory
• system is represented by lumped arterial,
  tissue, venous and
• pulmonary compartments.

ESCTAIC 6-9 October 2010, Amsterdam Netherland
13
14
Overview of SOPAVent

• The lung is sub-divided into three compartments
  
• an ideal alveolus compartment,
• where all gas exchange takes place
• with a perfusion-diffusion ratio of unity.
• b) a dead space compartment
• representing lung areas that are
• ventilated but not perfused
• c) a shunt compartment that is a fraction of
  cardiac output, representing both anatomical
  shunts and lung areas that are perfused but not
  ventilated.

ESCTAIC 6-9 October 2010, Amsterdam Netherland
14
15
Overview of SOPAVent

• What are the inputs and outputs of the model?

• The inputs of the model are the ventilator
  settings (FiO2, PEEP, PIP, RR,
• Tinsp) and the outputs are the arterial
  pressures PaO2 and PaCO2

• The model parameters are patient-specific and
  the model can therefore
• be matched to each patient provided the
  parameters are known.

ESCTAIC 6-9 October 2010, Amsterdam Netherland
15
16
Coupling aEIT and SOPAVent

• Objective

To simulate the effect of reducing PEEP to
changes of MEEV (predicted from ANFIS model),
PaO2 and PaCO2 (predicted from SOPAVent model)

• Method

• Loading patients specific data (ex ventilator
  parameters etc)
• The models were run for 300 seconds. PEEP was set
  at the initial value of
• 12 cmH2O and gradually decreased to 11cmH2O,
  10cmH2O, 9 cmH2O and
• 8 cmH2O, while all other ventilator settings
  remain constant
• Changes in MEEV, PaO2 and PaCO2 were observed
  and recorded

ESCTAIC 6-9 October 2010, Amsterdam Netherland
16
17
Coupling aEIT and SOPAVent

• Results

ESCTAIC 6-9 October 2010, Amsterdam Netherland
17
18
Conclusion

• aEIT is capable of tracking local changes in
  pulmonary air contents and thus
• can be used to continuously guide the
  appropriate setting of mechanical
• ventilation in critical care patients

• Mean end-expiratory lung volume (MEEV)
  calculated from aEIT is a feature
• parameter that reveals volume of air present
  in the lungs at the end of
• patients expiration

• Both models are capable of providing information
  on patients lung
• behaviour in response to ventilation therapy

• More ventilated patients EIT data are needed to
  further improve the accuracy of
• MEEV prediction

ESCTAIC 6-9 October 2010, Amsterdam Netherland
18
19
Future work
By using information from both aEIT and SOPAVent
models should lead to a better understanding of
phenomena surrounding ventilated patients in
order to support decision-making and
guide ventilator therapy.
Decision support system
ESCTAIC 6-9 October 2010, Amsterdam Netherland
19
20
THANK YOU
ESCTAIC 6-9 October 2010, Amsterdam Netherland
20