1.

Introduction

Artificial ventilation is a life-saving therapy in respiratory failure, which at the same time can induce lung injury or impair lung function.¹ The development of protective ventilatory strategies aims to decrease the incidence of ventilator-induced lung injury.² One aid in the development of new protective ventilation strategies is the generation of numerical models of the lung, requiring detailed information on alveolar dynamics during ventilation. However, in vivo dynamic 3-D geometries with adequate resolution have not yet been described in the literature. Hence, most scientists dealing with lung modeling use 2-D intravital microscopy (IVM) images or the results of histological examinations of fixed lung tissue as a basis for their numerical models, even though these do not describe a 3-D in vivo geometry.³ One imaging method for closing the lack of information could be Fourier domain optical coherence tomography (FD-OCT), which is a contact-free and high-speed imaging modality.^{4, 5, 6} FD-OCT allows 3-D imaging even in strong scattering samples such as biological tissue, with a resolution of less than $10\mu \mathrm{m}$ . FD-OCT is particularly suitable for performing investigations of lung tissue of isolated lungs through a thorax window in an in vivo mouse model, as described by our group in previous studies.^{7, 8, 9} To avoid pneumothorax and lung injury during preparation, we carried out a thorax window preparation without harming the parietal pleura. A similar preparation was already described by Wearn ¹⁰ for the visualization of pulmonary blood flow. OCT imaging was performed through the thorax window to provide 3-D geometry of subpleural alveolar structures needed for developing numerical models of the lung.

2.

Methods and Materials

2.2.

Image Acquisition

Imaging through the prepared thorax window was achieved using 2-D intravital microscopy and 3-D FD-OCT. A newly developed scanner head combining both modalities was used.^{6, 8} The subpleural alveolar structures were imaged using IVM, applying an IPP of $18\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ corresponding to a tidal volume of approximately $14\text{-}\mathrm{ml}∕\mathrm{kg}$ bw. A first image series of $15\mathrm{s}$ was taken focused on the inspiratory plateau (see Fig. 1 ). After $5\min$ of regular ventilation (IPP $12\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ ), IVM was repeated, focused on the expiratory plateau. The IPP of $18\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ during IVM was chosen because this is the maximum IPP, which avoids ventilator-induced lung injury (VILI) when applied for a short time. The rabbit was regularly ventilated for $5\min$ between acquisitions of IVM images. Limited by the sample rate of the OCT system, the acquisition of the 3-D image stacks ( $320\times 320\times 512\text{pixels}$ /approximately $1.3\times 1.3\times 2\mu {\mathrm{m}}^3$ ) takes about $8.5\mathrm{s}$ . For data capture, the ventilation was stopped at a constant positive airway pressure (CPAP) of $12\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ applied for, at most, $10\mathrm{s}$ (see Fig. 1). OCT imaging was only carried out at a CPAP of $12\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ to demonstrate the feasibility of OCT imaging in this model. The 3-D reconstructions of the alveolar structures were carried out by the Amira 4.1 software (Visage Imaging, San Diego, California).

Fig. 1

Ventilation pressure during image acquisition by IVM (a) and OCT (b). The IVM was documented by a movie of $15\mathrm{s}$ focused either to the inspiratory $\left(18\mathrm{cm}{\mathrm{H}}_2\mathrm{O}\right)$ or the expiratory $\left(2\mathrm{cm}{\mathrm{H}}_2\mathrm{O}\right)$ plateau. I (b), the ventilation chart for OCT image acquisition for an applied CPAP of $12\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ is displayed. The inspiratory plateau is increased up to $10\mathrm{s}$ to provide a stable situation for 3-D OCT imaging for the required $8.5\text{-}\mathrm{s}$ acquisition time. Before and after OCT imaging, the rabbit is regularly ventilated for at least $5\min$ to provide sufficient ventilation.

3.

Results

Briefly, the model was very stable for the time of investigation, and we detected no negative effects in hemodynamic $(\text{MAP}=80\pm 8\mathrm{mmHg})$ , blood gases ( $\mathrm{p}{\mathrm{O}}_2=86\pm 12\mathrm{mmHg}$ , $\mathrm{pa}\mathrm{C}{\mathrm{O}}_2=28\pm 5\mathrm{mmHg}$ ), and electrocardiogram $(\text{heartbeat}=173\pm 15\mathrm{bpm})$ after investigation. In this feasibility study, the experiments were stopped after $3\mathrm{h}$ . Also, no edema or inflammation caused by surgical preparation was observed.

Figure 2 shows the prepared thorax window (TW) (approximately $10\times 5{\mathrm{mm}}^2$ ), which was used for imaging alveolar lung tissue in vivo. The photograph of the ribs (R) and thorax window (TW) was taken from the intrathoracal side and was taken at the end of the experiment after resection of the tissue. In this figure, the undamaged parietal pleura (P) can be recognized. Figures 2 and 2 present that we are able to image identical alveoli during the whole ventilation cycle. The identical alveoli are denoted in the IVM image acquired in the expiratory phase as well as in the inspiratory phase by labels a, b, c, and d. The alveolar area marginally expanded approximately $7\pm 5\mathrm{\%}$ $(\mathrm{n}=15)$ at $18\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ , compared to $2\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ ventilation pressure. Figure 3 demonstrates that the described animal model is also suitable for 3-D OCT imaging. In Fig. 3, the intact parietal pleura (P) with approximately $50\mu \mathrm{m}$ thickness can be seen on top of the lung surface in the 2-D cross sectional OCT images, acquired with an applied CPAP of $12\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ . In the 2-D cross sectional OCT images, single alveoli (A) and the border between two lobes (B) of the lung can be recognized. Figure 3 shows a 2-D OCT enface image of subpleural lung parenchyma extracted from the acquired 3-D OCT image stack. The enface image represents the subpleural alveolar tissue in a $45\mu \mathrm{m}$ depth beneath the pleura, and was used to quantify alveolar areas.⁸ The mean of the alveoli area is $7.015\mu {\mathrm{m}}^2$ with a minimum of $1.984\mu {\mathrm{m}}^2$ and a maximum of $20.368\mu {\mathrm{m}}^2$ . The 3-D reconstruction of a single alveolar structure for a CPAP of $12\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ can be seen in Fig. 3. Furthermore, the 3-D reconstruction shown in Fig. 3 features a triangle mesh, which can be used as a basis for 3-D numerical models of the lung.

Fig. 2

(a) Prepared thorax window (TW/black ellipse) between the third and fourth rib (R) photographed after dissection from the intrathoracal side showing the transparent uninjured parietal pleura (P). (b) and (c) are IVM images acquired through the uninjured parietal pleura in the expiratory [(b) $2\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ ] and inspiratory [(c) $18\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ ] plateau phase. The identical alveoli labeled a, b, c, and d can be detected in both images. An expansion of approximately 7% of the alveolar area can be recognized.

Fig. 3

(a) Single alveoli (A) and the uninjured parietal pleura (P) with a thickness of approximately $50\mu \mathrm{m}$ are recognizable in the 2-D cross sectional OCT image of the investigated area. The image was despeckled by a median filter. The image was acquired in the inspiratory plateau with IPP of $12\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ . The border (B) of the upper lobe of the lung is demarked. (b) OCT enface image extracted from the 3-D OCT image stack. The enface image shows the lung parenchyma in a depth of about $45\mu \mathrm{m}$ beneath and parallel to the pleura. (c) A 3-D reconstruction of single subpleural alveoli at a constant airway pressure of $12\mathrm{cm}{\mathrm{H}}_2\mathrm{O}$ is shown. The alveolar structures are bordered at the upper flat side by the visceral pleura. At the lower side, the reconstruction is limited by the decreasing depth information, what can be seen in the cross sectional OCT image in (a), too.

4.

Discussion and Conclusion

We present a novel method for thorax window preparation for optical imaging that, unlike other techniques,⁹ can avoid the induction of pneumothorax during preparation and investigation. The described innovative preparation technique allows that when the parietal pleura is retained as a transparent membrane, IVM and 3-D OCT can be performed with an intact chest. This technique allows for tracking the identical alveoli during the whole ventilation cycle in healthy lungs. Furthermore, we have shown that OCT is highly suitable for 3-D in vivo imaging of subpleural alveolar structures within a resolution of less than $10\mu \mathrm{m}$ . Other imaging modalities, like computed tomography (CT) or magnetic resonance tomography (MRT), do not offer images at such high resolution. In contrast to techniques where cover glasses are placed on the surface of the lung for 2-D IVM imaging,¹¹ we have demonstrated a contact-free measurement without irritation of alveolar dynamics during ventilation. Additionally, it could be shown that a 3-D reconstruction based on 3-D OCT datasets is possible. The quantification of the alveolar areas in the IVM images results in an expansion of alveolar areas. This detected increase in alveolar area with increasing ventilation pressure is confirmed by other studies.^{8, 9} The handicap of this model for alveolar imaging is the limited observation area including only a small number of alveoli. Also, the investigated region has to be on the upper lobe, because the dislocation of the inferior lobe is much higher during the ventilation cycle. By gating the 3-D OCT measurement to the inspiratory and/or expiratory plateau, the 3-D OCT image stack can be acquired during ventilation without applying CPAP, as we have previously demonstrated.⁹ A gating to the heartbeat is also possible to minimize moving artifacts. In further studies, we will characterize the alveolar volume change during artificial ventilation in healthy and injured lungs. Additionally, high speed OCT systems allow OCT Doppler measurements,^{12, 13} providing blood flow and blood velocity measurements in the arterioles and investigation of the vascular function in healthy and injured lungs without opening the pleura. The 3-D OCT imaging technique in combination with the described animal model might improve knowledge about alveolar geometry and vascular function during ventilation in an in vivo situation. We conclude that dynamic data on alveolar volume change acquired by OCT might provide the basis for the development of numerical models of the lung, by using the surface mesh obtained by 3-D reconstruction of the 3-D OCT data of fine alveolar structures. As shown, OCT overestimates the alveolar walls and underestimates the air-filled alveolar space.¹⁴ This fact has to be considered using 3-D OCT data as input for numerical models.