journal menu
![[Figure 3]](mo5080fig3mag.jpg)
|
|
Figure 3
A direct comparison of the image features for PBI (a) and the TIE phase-retrieved data set (b) is shown, both obtained for the same sample at an object-to-detector distance of 30 cm (this figure depicts only the central part of the reconstruction for convenience). As indicated in the detailed views, PBI presents strong edge effects when compared with PhR. The overall grey value histograms (lower right corner) for PBI (blue) and PhR (red) show that, due to the phase effects overlaying the measured absorption, only one Gaussian-shaped peak can be seen. Therefore, no intensity-based segmentation between air and tissue can be performed. In contrast, after phase retrieval the histogram is clearly composed of two density intervals. The line profile at the position P in PBI (blue) and PhR (red) in the lower left-hand corner shows large variations and a strong edge effect for PBI, whereas in PhR the profile resembles the expected jump-function for a simple air–tissue interface. These drastic edge effects also cause negative values lower than the value for air, within the bronchi (dark contours). Therefore, the airways may appear filled, which is not the case. Note that the δ-to-β ratio for PhR was optimized for soft tissue and did not match the ratio of bone, which is why ribs and spine appear more blurred than in PBI. |
![[Figure 3]](mo5080fig3.jpg)
 | JOURNAL OF SYNCHROTRON RADIATION |
ISSN: 1600-5775
