Docstoc

Electrical Impedance Tomography and the Lungs

Document Sample
Electrical Impedance Tomography and the Lungs Powered By Docstoc
					Electrical Impedance Tomography and the Lungs

Mayneord-Phillips:July 03

Summary
What I intend to do is to start by trying to explain why tissue conducts electricity. I will look at lungs and see if we can predict how well they might conduct electricity I will talk about Impedance Pneumography and then what Electrical Impedance Tomography might offer

I will then look at some in vivo data from normals and from some patients. Finally I will ask if we can relate lung impedance directly to lung structure and see if we can quantify both pathological and normal changes with age.
Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Tissue
CSF Blood Skeletal muscle - longitudinal - transverse Lung - inspired - expired Neural tissue - grey matter - white matter Fat Bone

Resistivity
0.650 m 1.46 - 1.76 m 1.25 - 3.45 m 6.75 - 18.0 m 17.0 m 8.0 m 2.8 m 6.8 m 20 m >40 m

Frequency
1 kHz - 30 kHz 1 kHz - 100 kHz 100 Hz - 1 kHz " 100 kHz " " " 1 kHz - 100 kHz "

Mayneord-Phillips:July 03

relative permittivity (e)

b

g
102 106
1010
Mayneord-Phillips:July 03

frequency (Hz)

conductivity (s)

a

Typical resistivities
• Copper • Steel • Silicon 2 x 10-8 m 5 x 10-7 m 1 m (depends on purity) 1 x 103 m

• Soil

Mayneord-Phillips:July 03

What determines tissue resistance?

Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Tissue impedance consists of: RESISTANCE and CAPACITANCE

Mayneord-Phillips:July 03

1

Conduction current
0

Ic

log Ic and Id

-1 -2 -3

Id
Displacement current

1 2 3 4 5 6 7 8 9 10 log f (Hz)

Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Effect of freezing on plant tissue
1500

c2

ohms

1000
c1

a1 b1

500

a2

b2

Courgettes - Zucchini
3 4 5 6 7 8

0 0 1 2

frequency
The resistance of three courgettes (zuccini), was measured at seven frequencies in binary steps from 10 kHz (point 1) to 640 kHz (point 7). The initial measurements, a1, b1 and c1, all gave low frequency readings of about 750 ohms. In all three cases the resistance fell with increasing frequency because of the impedance presented by the cell membranes. 24 hours later the measurement c2 made on the vegetable that had been kept in a refrigerator showed an increase in low frequency resistance because of some dehydration. However, the readings a2 and b2 made on the two vegetables that had been frozen and then thawed showed a dramatic fall in resistance and a total absence of cellular structure. Mayneord-Phillips:July 03

I

V

I

V

Two-electrode configuration

Four-electrode configuration

Mayneord-Phillips:July 03

Impedance Pneumography

Mayneord-Phillips:July 03

V

impedance

Inspiration

time

Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Can we produce an image of tissue resistivity?

Mayneord-Phillips:July 03

Terminology
• • • • • Impedance Imaging (II) Applied Potential Tomography (APT) Electrical Impedance Tomography (EIT) Electrical Impedance Spectroscopy (EIS) Electrical Impedance Tomographic Spectroscopy (EITS)
Mayneord-Phillips:July 03

History of EIT
Electrical Impedance Tomography (EIT) now has quite a long history.

You can consider that the first publication of an impedance image was that of Henderson and Webster in 1976 & 1978. They used a 2-D matrix of 100 electrodes on one side of the thorax and a single large electrode on the other to produce a transmission image of the tissues. Henderson R P and Webster J G: An Impedance Camera for spatially specific measurements of the thorax, IEEE Trans. on BME 25, 250-254, 1978. Mayneord-Phillips:July 03

Transmission impedance image of theAthorax produced by image of Fig 1 transmission impedance Henderson and Webster. the thorax produced by Henderson and Webster. Iso-conductance contours show the lung regions.

Mayneord-Phillips:July 03

If you consider that these images were not tomographic then the first published images were those of Brown and Barber in 1982 & 1983. They used 16 electrodes, current injection between adjacent electrodes and a back-projection method of image reconstruction along isopotentials. Brown B H, Tissue impedance methods. In D F Jackson, Imaging with Non-ionising Radiations (Surrey University Press: Guilford) 1983.

Fig. 2 EIT image of an arm showing the ulnar and radius bones and peripheral fat.

Mayneord-Phillips:July 03

Since that time there have been many research programmes and a number of general publications in the area.
 There were two European Concerted Action programmes on EIT (called Applied Potential Tomography – APT in the early years). These gave rise to special issues of the journal ‘Physiological Measurement’ between 1986 and 1996 following meetings in Sheffield, Lyon, Copenhagen, York, Barcelona, Ankara and Heidelberg.  A book introducing the technology was published in 1990. Electrical Impedance Tomography, Edit. J G Webster, Adam Hilger, Bristol and New York.
Mayneord-Phillips:July 03

 A book considering the possible clinical applications was published in 1993. Clinical and Physiological Applications of EIT, Edit D S Holder, UCL Press, London.  A comprehensive review article appeared following the completion of the European Union Concerted Action programmes in 1997. K Boone, Imaging with electricity: report of the European Concerted Action on Impedance Tomography, Jn. Med. Eng. & Tech., 21, 6, 201-232.

Mayneord-Phillips:July 03

• A recent review followed the 10th International Conference on Electrical Bio-impedance held in Barcelona, in 1998. ‘Electrical Bioimpedance Methods, Applications to Medicine and Biotechnology’, Annals of the New York Academy of Sciences, 873, 1999. 11th was held in Oslo. • Last year a book called ‘Bioimpedance & Bioelectricity – BASICS’ by Grimnes and Martinsen appeared. Academic Press. •There is also a wide range of Web based sources on EIT (see www.eit.org.uk; www.tomography.umist.ac.uk; www.geocities.com/CapeCanaveral/9710/Eitlink.html).
Mayneord-Phillips:July 03

What does EIT involve?

What can we see?

Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Flexible belt of hydrogel electrodes produced Mayneord-Phillips:July 03 by N.I.B.E.C. – Dr Eric McAdams

N(N-3)/2 independent measurements
Mayneord-Phillips:July 03

X-ray CT

EIT

Mayneord-Phillips:July 03

A difference image of inspiration with respect to expiration

Changes in resistivity of 5-10% can be seen.
Mayneord-Phillips:July 03

The number of independent measurements will determine the best spatial resolution which will be possible. N=8 20 measurements
50 100 150 200 250 300 350 400 450 500 550
50 100

100

200

300

400

500

600

700

N=16

104 measurements

150 200 250 300 350 400 450 500 550

50

100

200

300

400

500

600

700

N=64 1952 measurements

100 150 200 250 300 350 400 450 500 550

Mayneord-Phillips:July 03
100 200 300 400 500 600 700

So what determines the set of transfer impedances which we can measure?

We will look first at what we might expect to measured from an empty thorax

Mayneord-Phillips:July 03

12

10 transfer impedance in ohms

8

6

4

2 0 5 10 15 20 25 30 35 Drive 1 Receive 3,4,5,6,7 ; Drive 2 Receive 4,5,6,7,8 etc. 40

3-D finite difference model of the neonatal thorax. The thorax is uniform at 5 ohms m. Major axis 132 mm, minor axis 110 mm. 8 electrodes have been used.

Mayneord-Phillips:July 03

We can compare a large thorax with a small one –

This is easy. All the transfer impedances scale in inverse proportion to size.

Mayneord-Phillips:July 03

12

10

transfer impedance in ohms

8

6

4

2

0 0 5 10 15 20 25 30 35 Drive 1 Receive 3,4,5,6,7 ; Drive 2 Receive 4,5,6,7,8 etc. 40

3-D finite difference model of the uniform thorax. Red curve is for Major axis 330 mm. Blue curve is for Major axis 132 mm.
Mayneord-Phillips:July 03

We can compare a thorax which is elliptical in cross-section with a more circular shape.

Mayneord-Phillips:July 03

15

transfer impedance in ohms

10

5

0 0 5 10 15 20 25 30 35 Drive 1 Receive 3,4,5,6,7 ; Drive 2 Receive 4,5,6,7,8 etc. 40

Data from a 3-D finite difference model of the thorax. The thorax is uniform. Ratio of major to minor axes 1.2 (blue curve) and 1.4 (red curve).

Mayneord-Phillips:July 03

Now what is the effect of adding some lungs, a heart, fat and bones?

Mayneord-Phillips:July 03

20 18

16
transfer impedance in ohms 14

12
10

8
6 4 2 0 0 5 10 15 20 25 30 35 Drive 1 Receive 3,4,5,6,7 ; Drive 2 Receive 4,5,6,7,8 etc. 40

Data from a 3-D finite difference model of the thorax. Red: organs included: Shape ratio 1.3. Blue: uniform thorax: Shape ratio 1.3.

Mayneord-Phillips:July 03

So what determines the set of transfer impedance which we can measure?

Size and shape matter, at least as much as content.

Mayneord-Phillips:July 03

By collecting two sets of data (reference and data sets ) a difference image can be produced showing fractional change in resistivity.

An image from a patient with lung water. Anterior is at the top of the image. The left side of the thorax appears on the right of the image.

This is a difference image. The reference was the mean of data collected from a large group of normal subjects. The data was from a patient with heart failure and associated lung water. The two lungs appear as regions of relatively low resistivity (blue). The difference between the normal and wet lungs is about 60%.
Mayneord-Phillips:July 03

Resistivity images collected at increasing levels of inspiration.
Mayneord-Phillips:July 03

EIT Lung images made from data collected at increasing frequency
Mayneord-Phillips:July 03

Anterior

Normal subjects: maximum change c. 20%

Right

Left

Patients with LVF: maximum change c. 60%

Posterior

From: T J Noble, A H Morice, K S Channer, P Milnes, N D Harris and B H Brown, (1999 Monitoring patients with left ventricular failure by electrical impedance tomography, Mayneord-Phillips:July 03 European Journal of Heart Failure, 379-384.

Changes with time

Mayneord-Phillips:July 03

3D mesh and electrode configuration
Sheffield MK3b system (64 electrodes)
32 current drive pairs 32 voltage receive pairs  1024 measurement data set ( gn)
Right
Posterior Anterior

Left

Drive electrode

Receive electrode

Electrode Plane 4 Electrode Plane 3 Electrode Plane 2 Electrode Plane 1

p8 p7

p6
p5

p4
p3 p2 p1

Mayneord-Phillips:July 03
V

Electrode Attachment

Plane 4 Plane 3
Plane 2 Plane 1

axilla electrodes

xiphoid process
Mayneord-Phillips:July 03

Sheffield MK3b Data Collection System
Specifications :
• 8 freqs (9.6kHz - 1.2MHz) • 64 electrodes • 16.6 frames s-1

current driver, multiplexer & receive amplifiers

saline filled phantom

data collection computer & ADC

demodulators

logic & control
power supply unit

triaxial cables

Ventilation Images
Option 1
use TLC as ref. and reconstruct others to this
RV FRC PTV
p8 p7

Option 2

Option 3
functional ventilation images

p6 p5 p4
p3

p2
p1
Mayneord-Phillips:July 03
+4% -40%

Ventilation Images
Option 1
use TLC as ref. and reconstruct others to this
RV FRC PTV
p8 p7

Option 2

Option 3
functional ventilation images

p6 p5 p4
p3

p2
p1
Mayneord-Phillips:July 03
+4% -40%

Breath held

Tidal breathing

A single set of transfer impedance measurements Mayneord-Phillips:July 03 made across the thorax over 40 seconds.

Cardio synchronous Images
Voxel Impedance Time Activity Curve
2

% resistivity change

1 0 -1
-2

0

3

6

9

12

15

High Pass Filtered Voxel Impedance
0.2

% resistivity change

0.1 0 -0.1 -0.2

ECG Monitor 0 3 6 9 12 15

time (s)

R wave events are recorded simultaneously with the EIT data

RV

TB

TLC
Residual Volume
Heart Rate 80 bpm (9 bpm) R wave Events 46 -1.75% +1.75%

Cardio synchronous Images

Tidal Breathing
Heart Rate 66 bpm (5 bpm) R wave Events 50 -1.50% +1.50%

Total Lung Capacity
Heart Rate 62 bpm (3 bpm) R wave Events 58 -1.00% +1.00%

Mayneord-Phillips:July 03

Some areas of possible clinical application
 Lung ventilation monitoring in intensive care  Optimisation of ventilation during anaesthesia  Lung water assessment in neonates and in adults with heart failure  Pulmonary embolus detection using both ventilation and perfusion measurement  Tissue characterisation

Mayneord-Phillips:July 03

Changes with frequency

Mayneord-Phillips:July 03

‘Frequency images’.

The images use the data collected at 9.6 kHz as a reference and then show the changes with respect to this data at frequencies from 19.2 kHz up to 1.2 MHz. The lung regions can be seen because the resistivity does not fall with frequency at the same rate as other tissues. The maximum change with frequency is about 60%. Frequency images show how different tissues can be characterised by the way in which impedance changes with frequency.
Mayneord-Phillips:July 03

Impedance Spectrum over the centre of the right lung. The frequency range was 9.6 kHz to 1.2MHz. Changes are referenced to the lowest frequency image.

Mayneord-Phillips:July 03

Hardware

Mayneord-Phillips:July 03

A small body-worn EIT system used to monitor thoracic fluid shifts in micro gravity.

Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Mk3.5 EIT specifcation

Number of electrodes

8

Number of receive/drive channels (m’plexed) Drive current

8 (in parallel)
200 Ap-p at each frequency: 212 A rms for each of 3, 10 frequency sets. This is 38% below what is allowed by IEC601 were the current to be applied continuously.

Multi-frequency range

2 kHz to 1620 kHz in 30, 3rd-octave steps

Speed of data collection
Method of demodulation

25 frames/sec
Digital by FFT

Outputs: High speed serial.

Electrically isolated to IEC601

Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Anterior
2 4 6 8

Right
10 12 14 16 2 4 6 8 10 12 14 16

Left

Posterior

Functional image from a 1 day old full term child Mayneord-Phillips:July 03

Mayneord-Phillips:July 03

Why is absolute EIT imaging so difficult?

Mayneord-Phillips:July 03

Size and shape matter, at least as much as content.

Cross sections of thorax
Mayneord-Phillips:July 03

Can we compromise or make assumptions to facilitate absolute imaging?
• (Measure) . Measure

•

•

both size and shape as accurately as possible. (Use a priori info.) . Estimate the positions of internal organs and make assumptions about their conductivities. (Be modest) .Only attempt to obtain whole organ absolute resistivity.
Mayneord-Phillips:July 03

Methods (Imaging)
• Measure circumference and major and minor axes of the actual thorax.

• Distort a Finite Difference model for size and shape to correspond to shape and size of the actual thorax • Reconstruct an image of the measured data set against the model.
• Use a region of interest over the lungs to compare the measured and computed data.

• Minimise the difference by adjusting the lung resistivity in the model. to that of the actual lung resistivity.
• Use a frequency imaging technique to obtain changes with frequency and hence obtain the absolute impedance spectrum.
Mayneord-Phillips:July 03

Tissue
Fat

Resistivity (m) 16

Muscle Bone
Blood

3.2 50
1.5 3-80

Lung

The resistivity values assigned to the tissues. Thymus and myocardium were assigned the same resistivity as muscle. No distinction was made between skeletal, vascular and smooth muscle
Mayneord-Phillips:July 03

Methods (patients)
Measurements were made on 155 normal children. Year 1 84 Year 2 31 Year 3 40 Four 1 minute epochs of data were collected during tidal breathing. The ‘Real’ part of the measured transfer impedances was used.

Mayneord-Phillips:July 03

Age (months)

Mean lung resistivity (m)

Number of measurements.

<1

5.7(1.7)

114

1-3.5 4-12
13-24

12.5(5.1) 15.3(6.3)
16.1(4.9)

49 149
139

Results

25-38

20.7(8.8)

163

The lung resistivity values derived from measurements made on children in different age groups. The figure given in brackets in each case is one standard deviation . These resistivity values are based upon the measured data sets which are an average of all frequencies between 4 kHz and 812 kHz.
Mayneord-Phillips:July 03

50

Lung resistivity (Ohm metres)

40

Measured spectra

30

20

10

0 30 20 10 Age (months) 0 10
1 3

10

2

10

Frequency (kHz)

Lung resistivity as a function of frequency and age. Lines show the mean within five age groups. It can be seen that resistivity increases and Mayneord-Phillips:July 03 characteristic frequency decreases with age.

• Lung resistivity increases with postconceptual age. • Spectral shape changes with age. • Can we relate resistivity and the shape of the spectra to the structure of lungs?

Mayneord-Phillips:July 03

Model simulation for neonatal lungs
• • • • •
The Nopp model for adult lungs has been modified as follows: Extra-capillary mean radius reduced. Extra-capillary blood volume reduced Number of alveoli reduced Ratio of air to condensed matter reduced Alveolar diameter reduced
Mayneord-Phillips:July 03

50

Lung resistivity (Ohm metres)

40

Modelled spectra
30 20 10 0 30 20 10

Age(months

0

10

1

10

2

10

3

Frequency (kHz)

Resistivity spectra calculated using the Nopp model. The frequency and age ranges are the same as for the in vivo measurements presented earlier. Mayneord-Phillips:July 03

50

Measured spectra
Lung resistivity (Ohm metres)
40

30

20

10

0 30 20 10 Age (months) 0 10
1 3

10

2

10

Frequency (kHz)

Mayneord-Phillips:July 03

We concluded that the modified Nopp model fitted the measured data very well. So can it be used to make predictions of maturational changes?
Mayneord-Phillips:July 03

Parameters

0-1 mth

1-3.5 mths

4-12 mths

13-24 mths

25-38 mths

Adult

Extra-capillary vessels: radii (m) Extra-capillary blood volume (ml) Number of Alveoli

30

38

42

46

50

70

17 50M
124

40 60M
126

60 70M
135

80 80M
145

118 90M
155

500 300M
180

Size of alveoli (m)

Lung filling factor

1.35

1.55

1.65

1.8

1.9

2.3

Mayneord-Phillips:July 03

Parameters
Extra-capillary vessels: radii (m) Extra-capillary blood volume (ml) Number of Alveoli Size of alveoli (m) Lung filling factor Intercellular conductivity (Sm-1) and permittivity
Intracellular conductivity (Sm-1) and permittivity Cell membrane conductivity (Sm-1) and permittivity

Adult value

Neonatal value

70 500 300M 200 2.3 2/80
1/80

30 17 50M 126 1.35 2/80
1/80

10-7/10 4

10-7/10 4

Some of the parameters which are required for the Nopp Model [Nopp et al, 1997]. In addition to the above Nopp assumes proportions of 0.85, 0.12 and 0.03 for blood, membrane and intercellular fluid for the alveolar walls in adults. We have used figures of 0.67, 0.24 and 0.09 for neonatal lungs. Nopp also uses an equation which gives blood conductivity as a function of frequency.
Mayneord-Phillips:July 03

Cell membrane thickness (nm)

20 Nopp model: Adults(red) Neonates(blue)

18
16

14
lung impedance 12

10
8 6 4 2 0 1 2 3 4 5 6 7 frequency 2kHz to 1.024 MHz 8 9 10

Measured adult (red) and neonatal (blue) spectra.
0.8

Adult (red) and Neonatal (blue) spectra predicted by the model.
Lung impedance (Real part)

0.6

0.4

0.2

0

-0.2

-0.4 blue-adult; red-32m; green-5m; cyan-3d; black-1d; -0.6 0 5 10 15 20 Frequency points 2kHz to 1.6MHz

Mayneord-Phillips:July 03
25

30

Conclusions
• Absolute organ imaging is possible if shape and size are measured. • Neonatal lung impedance spectra change consistently over the first three years of life • The measured neonatal lung spectra are consistent with the known changes in lung composition with maturation
• The results suggest that lung maturation is not complete at three years of age.

Mayneord-Phillips:July 03

In addition to absolute lung resistivity then what other information can we collect?

Mayneord-Phillips:July 03

The 8 electrodes have been applied around the thorax. A reference has been acquired of 25 frames. A lung region of interest (ROI) has been loaded. Tidal breathing can be observed. The trace shows the % changes in resistivity within the ROI

The red spot against electrode number 6 indicates a faulty lead or electrode.
Mayneord-Phillips:July 03

The mouse is set to acquire 1500 frames of data over 60 seconds. Data is recorded for all 30 frequencies from 2kHz to 1.536MHz.
Mayneord-Phillips:July 03

Data analysis
Pulmonary measurement

Mayneord-Phillips:July 03

In Data Analysis we can LOAD the previously acquired file called Pete_1_Dec02. Comments attached at the time of data acquisition are displayed.

This shows the raw data when current is driven on channel 2 i.e. between electrodes 2 and 3. In addition to the inspiratory changes small cardiac related changes can be seen on some of the receive channels.

This shows more of the raw data for drive channel 5 at a frequency of 2kHz. Note that the ‘Real part’ of the impedance has been selected.

The raw data is displayed as 8 ‘profiles by selecting all 8 drive channels. Profiles are usually U shaped. The display shows the Data recorded at the 13 selected frequencies. Mayneord-Phillips:July 03

Three ‘frequency images’ are shown with respect to a reference average of the frequencies 10079 – 20159Hz over a period at inspiration. Parts of the lungs and heart appear dark as they change resistivity less with frequency than the surrounding tissues.

Right lung

Left lung

Using one of the frequency images a region of interest(ROI) can be set over the lungs at 50% below the maximum change. Dynamic images(see next slide) are usually more suitable to define an ROI.

A reference has been taken from frames 1 to 250 (expiration). The dynamic image is shown for frame 390 after inspiration. The lungs appear clearly as an increase in resistivity. Anterior is at the top and the left lung appears on the right.

The recorded ECG has been added below this time plot of resistivity. It can be seen that the cardiac related changes show a fall in resistivity Mayneord-Phillips:July 03 corresponding to systole.

The cardiac related changes can be seen more clearly by showing just part of the recording. Mayneord-Phillips:July 03

This set of dynamic images show clearly that the changes in lung resistivity on inspiration depend upon frequency. From this display the ‘ABS R’ button was selected.

The shape of the thorax described by e, the ratio of the width to the depth, must be entered as well as the chest circumference.

The Cole parameters Fc and R/S have been calculated from the impedance spectrum. Data quality can be assessed from the mean reciprocity value. The Functional image shows no clear changes as no ventilation took place.

The absolute value of lung resistivity in ohm metres is now shown as function of frequency. The changes with time are also shown but there is little change as this was a breath hold on inspiration.

No time changes are seen as both data sets were recorded whilst the breath was held. For this reason the Functional Image does not show an clear changes. See next slide.

The upper figure again shows the lung resistivity at inpiration. On The right the resistivity for the breath hold at expiration is shown. It can be seen that the mean change is from 11.1 to 33.8 ohm metres on inspiration.
Mayneord-Phillips:July 03

In this case the absolute lung resistivity during the period of inspiration is shown. This appears on the time plot. Because ventilation has occurred the functional image now shows the lungs clearly

Respiratory Manoeuvres
1000 frames of data (16.6 frames s-1) collected from 10 subjects (seated), during:

Total Lung Capacity (TLC)
1.85

Residual Volume (RV)

Tidal Breathing (TB)

TLC
1.8

mean voltage (V)

1.75 1.7 1.65 1.6

Inspiration

PTV
FRC

1.55
1.5 1.45

RV
0 100 200 300 400 500 600 700 800 900 1000

Expiration

time (frames) Inspiration levels, Functional Residual Capacity (FRC) and Peak Tidal Volume (PTV) extracted from tidal breathing sequence

Conclusions
• Tissue conducts electricity because it contains ions • Impedance changes with frequency because of membranes • EIT can show changes of lung tissue impedance These changes can be with time or they can be with frequency. • Lung impedance can be modelled and related to structural arrangements • Absolute lung resistivity imaging is possible
Mayneord-Phillips:July 03

Tanker on the ice road across the Mackenzie River in Canada

THE END

Mayneord-Phillips:July 03


				
DOCUMENT INFO