Study design
Using EIT, ventilation distribution and regional lung filling were measured in seven rats in each of four body positions and with three gas mixtures of differing density randomly inhaled.
Animal preparation
Animal ethics approval was obtained from The University of Queensland. Seven Wistar rats (8 to 10 weeks of age, 279 ± 36 g of either sex) were studied. The rats were anaesthetized, intubated and prepared accordingly to our standard protocol
[11]. Three gas mixtures were used, in random order, to ventilate the rat – air (ρ = 1.2 gL-1, viscosity: 186 μP); -, 70% helium with 30% oxygen (Heliox, ρ = 0.51 gL-1, viscosity: 199 μP); - SF6, 20% sulphur hexafluoride with 20% oxygen and 60% air (ρ = 2.4 gL-1, viscosity: 270 μP (calculated)). All gas mixtures had compressibility factors within 1% (
http://encyclopedia.airliquide.com/encyclopedia.asp). With each gas mixture, the rat was randomly placed in each of four postures: prone, supine, left- and right-lateral and ventilated using a time-cycled, pressure-limited ventilator based on that of Hedlund
[12] with a respiratory rate of 80 breaths per minute and a tidal volume of ~10 mL/kg.
Electrical impedance tomography (EIT)
A Göttingen GoeMF II EIT tomograph (Sensormedics/ VIASYS Healthcare, Netherlands) was used
[13]. The basic principles of EIT have been published elsewhere
[14, 15]. The rats were circumferentially shaved around the chest and 16 epicardial pacing wires (Medtronic Inc, Minneapolis, MN, USA) were sutured in an equidistant fashion through the skin and the panniculosus carnosus
[11]. EIT measurements were made with a 100 kHz injected current at 44 images per second. A sensitivity-weighted back-projection algorithm
[16] was used to reconstruct a 32×32 pixel image of the distribution of relative impedance changes. A minimal data set length of 60 second or at least 60 breaths for analysis were required.
Data analysis
Functional EIT data was analysed offline using custom developed software (MATLAB, Mathworks, 7.2, Natick, MA, USA). Data were filtered using a band pass filter including the first and second harmonic of the respiratory rate
[17–19]. With this filter in place, the ventilated regions were defined as regions in which the impedance signal was greater than 20% of the peak impedance signal
[20, 21].
Ventilation distribution (VD)
Three measures of VD were employed – the amplitudes of regional impedance change, the geometric centre (GC) as a measure of regional VD and the global inhomogeneity index (GI) as a measure of global VD.
Amplitudes of regional impedance changes for the anterior, posterior, right and left side of the ventilated regions were calculated by averaging the end expiratory to end inspiratory impedance differences for each pixel in the region of interest (ROI)
[22]. To account for the unequal number of pixels analysed in the different ROIs, the average amplitude of each ROI was reported.
The GC of the EIT image was calculated for the entire image
[23]. The GC defines the centre of ventilation using a balanced averaging of pixel values from right to left and from anterior to posterior.
The GI quantifies the tidal volume distribution within the ventilated region
[24]. A median value for all pixel amplitudes was calculated across the entire image and the sum of the absolute difference between the median and every pixel indicated the tidal volume distribution in the ventilated region. The value was then normalised to the number of pixels included. The lower the GI value the more homogenous the ventilation is distributed.
Regional (temporal) filling characteristics
The filling index describes the rate of volume change between different ventilated regions. In theory, gas density should impact on the temporal filling of lung regions with the rate of volume change in the dependent lung increased with more dense inhaled gas. Regional volume change was compared to global volume change by plotting the impedance change of the relevant region against that of the global signal to form a curve
[25]. The slope (g) was fitted, using a Levenberg-Marquardt method, to the following equation
[25]:
where I(g) is the regional impedance change, g is the global impedance change, FI is the regional filling index and a and c are constants.
The filling index, FI, describes the shape of the curve. A linear relationship (FI = 1) is found if the rate of volume change of a ROI is the same as the global lung. If the rate of change in a ROI is initially less but increases as inspiration continues, the curve has a concave shape (FI > 1). If the rate in a ROI during the initial phase of the inspiration is greater than the global lung but decreases towards the end, then the curve has a convex shape (FI < 1).
Statistics
A general linear model was used to seek interactions between position or gas composition on measured parameters. Results were described using the mean and confidence intervals. An ANOVA with Bonferoni for repeated measurements was used to compare parameters within each position. Data was described using mean and standard error of the mean. For statistical analysis SPSS version 15.0 (SPSS Inc., Chicago, IL) was used. Significance was accepted at p < 0.05.