Simple low dose radiography allows precise lung volume assessment in mice

Abstract

X-ray based lung function (XLF) as a planar method uses dramatically less X-ray dose than computed tomography (CT) but so far lacked the ability to relate its parameters to pulmonary air volume. The purpose of this study was to calibrate the functional constituents of XLF that are biomedically decipherable and directly comparable to that of micro-CT and whole-body plethysmography (WBP). Here, we developed a unique set-up for simultaneous assessment of lung function and volume using XLF, micro-CT and WBP on healthy mice. Our results reveal a strong correlation of lung volumes obtained from radiographic XLF and micro-CT and demonstrate that XLF is superior to WBP in sensitivity and precision to assess lung volumes. Importantly, XLF measurement uses only a fraction of the radiation dose and acquisition time required for CT. Therefore, the redefined XLF approach is a promising tool for preclinical longitudinal studies with a substantial potential of clinical translation.

Figure 1

Schematic illustration of the set-up for correlative XLF, WBP and micro-CT measurements. One end of the WBP chamber has an isoflurane inlet/outlet, the other end is connected to the differential pressure sensor (DPS) which in turn is connected to a reference chamber, a Powerlab data acquisition device and a portable computer (PC). The chamber is placed inside the gantry of the micro-CT imaging system on the sample stage. The mouse is positioned inside the chamber such that the chest cavity is within the field of view (FOV). A piezoelectric (PZT) acoustic sensor that transduced the sound of the CT door interlock is used to synchronize data acquisition from WBP with XLF or micro-CT.

Figure 2

Comparison of traces from relative X-ray transmission function (rXTF) at lung region and WBP. (a) XLF-image in inspiration. The white box indicates the region of interest (ROI) that was used to measure changes in intensity of rXTF. (b) Data from rXTF and WBP. Red trace: rXTF at the ROI (arbitrary units). Blue trace: band pass filtered (0.5–20 Hz) flow trace (µl/s) from the WBP. Green trace: WBP Volume trace (V_t, µl) derived from the integral of the flow (reset each cycle). (c) Averaged data from (b). (d–f) Correlation of the time course and signal derived from rXTF and WBP, (d) time rise (t_rise) of the signal (20–80%), (e) decay of the signals (t_fall) (20–80%) and (f) tau (τ) using a mono-exponential fit (peak to baseline, LabChart 8.0, ADInstruments). p values and respective coefficients of determination (R²) from linear regression analysis are shown on the graph.

Figure 3

Quantification of 3D lung volumes at expiration and inspiration. (a) The red line marks the exemplary ROI selected near the diaphragm for representing the lung motion in the micro-CT projections. (b) A graph for the lung motion at the ROI is generated where the green and yellow lines show the threshold lines for maxima and minima selection, respectively. (c) A representative single breathing cycle is shown with an exemplary selection of one frame for expiration phase (red circle) and two frames for inspiration phase (yellow circles). (d) Following the selection of multiple breathing phases from the entire breathing curve, the frames from the whole micro-CT projection are sorted to reconstruct two segregated 3D lung volumes for expiration and inspiration. (e) A region inside and outside the lung (red and black squares in d) were selected to generate histograms showing the distribution of grey values for each region. (f) The lung volume ($V_{\left[\mu ,C,T\right]}$) (red) is segmented through a region growing method. (g) The calculated tidal volumes from micro-CT ($V_{\left[\mu ,C,T\right]}^{\mathit{TV}}$) from the segmented lung region show a strong positive correlation with the increasing stroke volumes used for mechanically ventilating euthanized mice. The body weight of the mice, p values and respective coefficients of determination (R²) from linear regression analysis are shown on the graph. Scale bars in (d) represent 1 cm.

Figure 4

Correlation between lung volumes measured using micro-CT and WBP in living mice. Graph showing a weak positive correlation between TV_[WBP] with $V_{\left[\mu ,C,T\right]}^{\mathit{TV}}$ (R² = 0.5061) performed in-vivo (n = 8) using the correlative set-up. All points are labelled with the respective weights of the mice. The p values and respective coefficients of determination (R²) from linear regression analysis is shown on the graph.

Figure 5

Quantification of advanced XLF parameters from the X-ray transmission over time. (a) A representative radiograph is shown, demonstrating background correction based on the selection of ROI at the lung lobes marked in cyan normalized by selection of the background marked in red (top panel). The modified background correction at ROI is accomplished by applying an adaptive moving average filter which requires a smaller FOV (bottom panel). (b) Two exemplary breathing cycles extracted from the averaged X-ray transmission at the ROI are shown for a healthy mouse over a period of 3 s. The peak intensities representing maximum inhalation I are marked by yellow circles while the red circles represent the beginning of a new breathing event and the maximum exhalation phase E of the breathing cycle. The breathing curve is split at threshold T (marked by a cyan circle and a grey horizontal line) to obtain a breathing (A) and a passive expiration (B) phase. XLF parameters are shown including average inhalation time (t_in), average breathing length (L), X-ray-based end-inspiration volume (EIV_[XLF]), relative X-ray transmission at end-expiration (rXTE) and exponential function (τ, calculated in region indicated by blue dotted line). Scale bars in (a) represent 1 cm.

Figure 6

Correlation of XLF measurements with micro-CT and WBP. (a) Graph showing a strong positive correlation between EIV_[XLF] with $V_{\left[\mu ,C,T\right]}^{\mathit{insp}}$ performed on dead mechanically ventilated mice (n = 3). (b) Graphs showing a strong positive correlation between EIV_[XLF] and $V_{\left[\mu ,C,T\right]}^{\mathit{insp}}$ (R² = 0.94) and (c) a weak correlation of TV_[WBP] with EIV_[XLF] (R² = 0.32). Data for both b and c is shown for living mice (n = 8). The respective p values and coefficients of determination (R²) from linear regression analysis are shown on the graph. (b,c) All points are labelled with the respective weights of the mice.