In past small animal studies, x-ray dark-field imaging showed a potential benefit in diagnosing lung diseases [31,32,33,34,35,36,37,38,39,40,41,42, 54]. Subsequent large animal and human cadaver measurements showed the possibility to translate x-ray dark-field radiography to the human scale [43,44,45,46], and a first study on dark-field chest radiographs of human bodies showed that the dark-field signal could be reliable quantified [55]. However, further applications of x-ray dark-field imaging has not been extensively investigated on large animal models or human cadavers yet.
In this study, we presented the first x-ray dark-field image of a complete human body and compared findings in the transmission, dark-field and CT images. A high dark-field signal was found in the lung. Furthermore, the bone, calcification in the femoral arteries, implants and foreign bodies also produce a dark-field signal. Phantom measurements showed that a strong visibility reduction signal can be produced by objects which do not generate small-angle scatter.
In this study, the ED values resulted to be higher than typical ED in some organs. The typical ED are given for radiographs where only the organ in question is imaged. As a consequence, image parameters like tube current were chosen to be optimal for imaging this organ. In our study, the imaging parameters were the same for all organs. Thus, the tube current had to be chosen in such a way that enough photons reach the detector behind the more absorbing body parts like head, pelvis, and abdomen. For those regions, the ED in this study resulted to be similar to the typical ED. Higher ED in less absorbing organs is a consequence of this experimental setup.
Our results concerning the thorax region are similar to the results reported in other studies [43, 44, 46]. The considerable dark-field signal in the lung originates from small-angle scattering on the numerous air-tissue interfaces in the lung parenchyma, whereas the high transmission is due to the weak attenuation of the mainly air-filled lungs. In this study, the dark-field signal of the lung was lower compared to the dark-field signal of the pig lung reported by Hauke et al. [43]. The authors euthanised a pig a few minutes before image acquisition, whereas in this study, the images were taken 4 days postmortem. Therefore, the lung of the human body was more strongly collapsed with partially fluid-filled alveoli compared to the pig lung and not as many air-tissue interfaces were present anymore in the human lung.
In contrast to the lungs, barely any air-tissue interfaces exist in the stomach and, as a consequence, no dark-field signal originates in the stomach. The low attenuation signal is a result of the stomach being extensively filled with air due to a previous incorrect positioning of the endotracheal tube within the oesophagus.
Dark-field signal is generated by the individual beads of the antibiotic bead chain, as these are made from a spongious material which acts as a substrate for the antibiotic agent. Spongious materials contain large numbers of interfaces with air, leading to an increase in scattering and therefore to an increased dark-field signal.
Calcification, i.e., accumulation of calcium salt, in particular of arterial walls (as it can be seen in atherosclerosis), is a common finding, especially in patients with cardiovascular risk factors. Atherosclerosis can lead to stenosis of the affected arteries, resulting in hypoperfusion of the anatomical region which is supplied by the respective artery. Depending on the affected arteries, patients can suffer from coronary or peripheral artery disease, as it was for the patient in our study. Previous studies showed that calcification in the breast is visible in the dark-field image, potentially improving breast cancer detection [56,57,58]. We were also able to see calcification of the arteries in the dark-field image. However, whether the calcification is better and earlier visible on dark-field than common transmission images and if dark-field imaging has the potential to improve the diagnostic value has to be determined in further studies.
Bony structures are visible in both transmission and dark-field images, especially the areas of increased calcium content: subchondral sclerosis as a cause of bilateral osteoarthritis of the hips was highly visible in the dark-field image. Similarly, subchondral sclerosis due to osteoarthritis of both knees also resulted in a high signal on the dark-field images.
As visible in Fig. 4, even non-scattering materials can cause a reduction in visibility which is due to beam hardening. The interferometric visibility of an x-ray dark-field imaging setup is not only dependent on the setup arrangement (i.e., grating periods and inter-grating distances), but also on the spectrum of the x-ray beam. In our setup, visibility was high for photon energies up to 40 keV and decreases continuously for higher energies. When measuring with energy integrating detectors, the average visibility results from the detector signal of the energy dependent visibility weighted with the detected signal from each photon energy [59]. Since the low-energy x-rays, which contribute the most to the measured visibility, are preferentially attenuated by most materials, the measured, average visibility decreases. As a result, non-scattering, strongly attenuating materials can induce a decrease of visibility, which is indistinguishable from a true dark-field signal due to small-angle x-ray scatter. For soft and adipose tissues, this effect could largely be eliminated by a correction, which is based on visibility reduction measurements of POM. As POM has spectral attenuation properties similar to soft tissue, which differs significantly from those of bone and metal, the correction however fails to accurately model their contribution to beam hardening, and a residual visibility reduction signal remains.
It has been shown that the bones also generate a dark-field signal [47, 48, 60]. The dark-field signal of the bones is lower as the signal of the lungs. Our setup showed a lower sensitivity than the ones used for bone measurements. Therefore, in our opinion, the dark-field signal of the bones in the here presented measurements mainly result from residual visibility reduction signal.
Our study has some limitations. First, only one human body was imaged. Furthermore, the complete image was stitched together from six individual scans and the body had to be repositioned during the acquisition procedure. This experimental setup is not practical as an imaging approach in clinical routine. This problem could be overcome by changing the acquisition of the dark-field images: instead of moving the interferometer, the sample could be moved as proposed by Seifert et al. [61]. Thus, the patient can be imaged in one scan similar to a full body CT scan. Lastly, the beam hardening correction was performed with respect to the soft tissue component. Removal of beam-hardening-related artefact depends on the knowledge of the material composition of the body part in question. Calibration data and algorithms used for correction must therefore be tailored to the measured body part. In particular, superposition of materials with very different spectral properties (e.g., bone and soft tissue) presents a special challenge for beam-hardening-related visibility correction in dark-field radiography, since the individual material contributions to the total attenuation cross-section along the x-ray path are unknown and vary over the field of view. A useful approach in such a case may be to use dual-energy acquisition, which may allow calculating spatial maps of such cross-section fractions and thus enabling correction of beam-hardening-related dark-field artefact for two different materials.
In conclusion, we gave an overview over the dark-field signal strength of different organs of the human body presenting images and data from a female cadaver. However, before dark-field imaging can be included into the clinical routine, further studies will have to be conducted. Besides lung imaging, bone imaging and the possibility of diagnosing calcification might be of special interest.