In this study we evaluated the use of micro-CT for assessment of pulmonary fibrosis induced by adenoviral gene transfer of biologically active TGF-β1 in mice. We found micro-CT imaging to be a highly valuable tool to study pulmonary fibrosis progression in mice using both visual inspection of lung images and a semi-automated segementation algorithm that estimates aerated lung areas as a surrogate marker inversely correlating with developing lung fibrosis in mice.
Micro-CT allows a detailed assessment of lung morphology due to high density differences and well defined borders between air and lung tissue. However, one limitation of this method is that the activity of inflammatory processes can only be assessed by secondary characteristics by micro-CT, such as affection of adjacent strucutre or pleural effusion. Other imaging techniques such as magnetic resonance imaging (MRI), micro-positron emission tomography (PET) and optical imaging could provide functional data, with micro-PET beeing an especially promising candidate to robustly assess inflammatory activity by increase of glucose metabolism as detected by 18F-fluoro-deoxyglucose micro-PET [17–19], which might be particularly relevant in terms of monitoring acute exacerbations of pulmonary fibrosis in small laboratory animals.
Image quality is of equal importance in small animal imaging as in human imaging. However, data on imaging protocols are still not extensive. One major issue addressed in previous studies relates to respiratory gating. Various gating techniques, including prospective and retrospective as well as intrinsic and extrinsic techniques have revealed that mostly, gated imaging resulted in increased image quality [20, 21]. In the current study, we found that mean values for image quality were better in gated exams for all observers, however differences were minor and not significant. Both gated and ungated exams resulted in a highly significant correlation with histology-based Ashcroft scores. Therefore, we conclude that gating should be applied whenever possible, but is not an absolutely critical issue when this setup is not available.
A further important aspect of micro CT imaging of normal or fibrosing mouse lungs is the time-point in the respiratory cycle when images are acquired. Due to the typical respiratory pattern in anesthesia, with a short breath followed by a long expiratory plateau, imaging is mostly performed in expiration. Under those experimental conditions, where intratracheal intubation and ventilation is performed, micro-CT imaging can also be performed in inspiration, as a different and consistent respiratory pattern can be created . Without gating, imaging is effectively performed during expiration, due to the longer period of time as compared to inspiration that only represents a small fraction contributing to the average imaging time-point.
The semi-automated segmentation routine described in this study was found to assess consolidations in a quantitative way, as proven by the significant correlation of segmentation values with Ashcroft-based histological grading. To a minor degree, observer interaction is possible, as the observer has to place its own seed points for the region growing algorithm. As seed points have to be placed carefully within aerated lung areas, inadvertent positioning of seed points within non-aerated lung areas may increase the likelihood that consolidated aereas will be included in the segmentation volume and thus may cause assessment bias, which however, according to the presented data, appears to be very low.
In the current study, segmentation volumes were not correlated with total lung volume or body weight of individual mice. However, when considering that differences in expiratory total lung volumes between individual mice of the same age and strain are neglectable, it may be largely excluded that this aspect may have confounded the reported correlations. Future algorithms might take segmentation assessments of the entire lung into account, thereby allowing the correlation of measured consolidations with total lung volumes, or simply by employing a quotient taking the body weight into account.
Additionally, different techniques of fibrosis assessment could be applied. A promising approach is the measurement of CT densities expressed by Hounsfield Units and assessing fibrosis not only by consolidation but also by evaluation of density distribution . Morphological pattern assessment as applied in human fibrosis assessment has also been described for imaging of lung fibrosis in small laboratory animals [24, 25]. However, such approaches would be dependent on an image quality that is consistently comparable to human imaging, which is technically still difficult to achieve in in vivo imaging of small laboratory animals.
In another study we evaluated the applied radiation dose for the protocols used in this study applying phantom and cadaver thermoluminescence dosimetric measurements (own unpublished observations). The expected mean doses in an average C57BL/6 mouse for the respiratory gated and ungated protocols used in this study were 201 mGy and 194 mGy respectively. The expected mean value for an estimated time of 1 minute fluoroscopy to determine the scan field of view was 22 mGy. Day et al. reported that high radiation doses of 7 - 9 Gy did not result in significant lung fibrosis in mice after 90 days, just few microscopically small collagen-rich foci were detected subpleurally . Therefore, it appears that radiation-induced fibrosis can be largely excluded even when serial examinations are performed.
Human lung fibrosis differs from the animal model reported here, as it is irreversible and shows typical morphological patterns in computed tomography representing definite fibrotic changes of the lung. The application of the methods reported for imaging in human lung fibrosis therefore is not intended. We believe that the currently applied technique will primarily be helpful to better assess e.g. novel therapeutic strategies for the treatment of pulmonary fibrosis in rodent model systems.