view article

Figure 4
Comparison of generated and actual tissue structures: (a) Three 256 × 256 × 256 voxel cubes from a PCCT scan of a rat lung FFPE tissue block: (i) cube with air artifacts (top left), (ii) cube with air artifacts suppressed by a trained generator model (top middle), and (iii) actual cube without air artifacts (top right). (b) Histograms of each cube. The cube with air artifacts (blue) shows peaks at 0 and 255, indicating low intensity artifacts and high intensity at their edges. The generated cube (orange) lacks these extremes, aligning with the original cube's intensity distribution. (c) For ten randomly selected ROIs with air artifacts, porosity, pore volume, pore extent and solidity were calculated for the original data (blue) and for data with air artifacts masked (orange). Since there is no ground truth, the artifact-free values (red line, after applying our cGAN) were used as reference. Masking the artifacts resulted in similar measurements to our approach, which increases the average pore size due to blending of alveolar regions (the ^ symbol denotes an outlier).

Journal logoJOURNAL OF
SYNCHROTRON
RADIATION
ISSN: 1600-5775
Follow J. Synchrotron Rad.
Sign up for e-alerts
Follow J. Synchrotron Rad. on Twitter
Follow us on facebook
Sign up for RSS feeds