Discussion
X-ray dark-field radiography is a novel imaging modality with high potential for lung imaging that has been translated from an experimental method to clinical applicability in recent years. However, so far, its feasibility for imaging of the human lungs has only been demonstrated in a single post-mortem chest radiograph. Therefore, the purpose of our study was to assess the imaging features of nine post-mortem dark-field chest radiographs from a clinical point of view as a final step before evaluation of dark-field radiography in clinical studies.
In the visual assessment of post-mortem dark-field chest radiographs, we observed a gradient of dark-field signal strength from the apex to the base of the lungs. This can be attributed to an increasing amount of lung parenchyma in the x-ray beam path, as small-angle x-ray scattering increases with the amount of scattering material. However, in conventional chest x-rays, the transmission did not correlate with the dark-field signal, demonstrating that this presents a unique imaging feature of dark-field radiography. This finding is in accordance with results of animal studies [12, 13, 21]. In a clinical context, this would have to be considered when investigating pathologies that decrease dark-field signal. In centrilobular emphysema, a form of chronic obstructive pulmonary disease that primarily affects the upper lobes, knowledge of this finding will be essential for correct image interpretation.
X-ray dark-field radiography is an imaging modality which we believe will primarily be evaluated by visual assessment. Hence, for its clinical application, it is of major importance that imaging findings are consistently reported. We could demonstrate that visual grading of x-ray dark-field signal in the lungs shows substantial to almost perfect intra- and interobserver agreement, comparable to visual assessment of transmission in conventional chest x-rays. These results confirm outcomes of reader studies performed on x-ray dark-field radiographs of different lung pathologies in small animals [12, 14] and underline clinical applicability.
Diagnostic image quality is mandatory for reliable reporting of imaging findings and an insufficient image quality may reduce diagnostic confidence of the reporting radiologist [30]. The image quality of x-ray dark-field and simultaneously acquired conventional radiographs were graded as good and very good, respectively. Therefore, we deduce that x-ray dark-field imaging provides sufficient image quality for its application and evaluation in clinical trials.
Numerous preclinical animal studies have demonstrated the effect of specific lung pathologies on the dark-field signal. For example, in an animal model of idiopathic pulmonary fibrosis [13, 14], x-ray dark-field imaging allowed visualisation of early fibrotic changes in the lungs with dark-field images showing circumscribed areas of markedly reduced dark-field signal in lung parenchyma affected by the pathologic process next to normal areas with high dark-field signal. In our study, the correlation of pathological findings in conventional x-ray and CT images of individual human bodies with dark-field signal intensity showed comparable results in the human lungs. Areas of pulmonary consolidation may contribute to a major reduction of dark-field signal, whereas ground-glass opacities, representing interstitial and alveolar infiltrates, showed an inconsistent effect on the reduction of dark-field signal.
In one human body, an enlargement of the heart correlated with a strong decrease of the dark-field signal in the left lung lower zone, probably by reducing the amount of lung parenchyma in the x-ray beam path. This is a relevant finding as it shows that extrapulmonary pathologies can also influence the dark-field signal intensity and would have to be considered when interpreting clinical dark-field radiographs.
There are several limitations to our study. The application of x-ray dark-field chest radiography to human bodies limited our control over parameters that potentially alter the dark-field signal. Emphysematous changes, interstitial and alveolar infiltrates, pulmonary consolidation, pleural effusion, and an enlargement of the heart are pathologies we observed in the human bodies. Additionally, decomposition processes after death may influence dark-field signal intensity. In this study, it was not possible to differentiate and quantify the individual contribution of a single finding to the change in dark-field signal strength, because none of the lungs was free of pathologies and could have served as a reference. Furthermore, due to lung anatomy, it is difficult to correlate a CT finding in a pulmonary lobe to dark-field signal changes in a lung zone of a radiograph. However, we gained insight into findings that may have a more pronounced effect on dark-field signal reduction, e.g., pulmonary consolidation, that has not been addressed in animal studies. X-ray dark-field imaging was performed in supine position, which leads to dystelectasis in the dorsal basal parts of the lungs and may influence dark-field signal. To at least partially compensate for this, we performed endotracheal intubation and kept airway pressure constant during x-ray dark-field imaging. Since x-ray dark-field radiography is a novel imaging modality, the possibility to train the readers for the visual evaluation of the dark-field images was limited. The number of human bodies included in our study is relatively small. Still, our results are in accordance with previous animal studies and demonstrate clinical applicability.
In conclusion, our study on post-mortem human x-ray dark-field chest radiography demonstrates that x-ray dark-field images provide complementary information of the lungs to conventional x-ray, allow reliable visual quantification of dark-field signal strength, and have reached an image quality warranting an evaluation in clinical trials.