Fig. 5: A multi-class, multi-physics uncertainty analysis of a CT scan of a human torso.
From: Quantifying the unknown impact of segmentation uncertainty on image-based simulations
a 2D slices of the CT image labels for the spine and aorta overlaid in green and orange, respectively. b A 3D visualization of the μ-percentile segmentation of both classes (spine and aorta). Visualization of spine axial compression simulation showing c vertical displacements, d lateral displacements, and e the resulting uncertainty distribution in the effective Young’s modulus, including distribution estimates. f Visualization of the wall shear stress in the aorta resulting from a flow simulation. g Prescribed aorta inlet pressure profile (gray) overlaid on the time-dependent outlet side-to-bottom flow rate ratio, including uncertainty bounds. h Aorta wall-shear-stress threshold ratio for each percentile segmentation, with the standard segmentations shown as thick solid curves. The dark gray-shaded region emphasizes the curves bounded between (μ + σ)- and (μ−σ)-percentile segmentations, whereas the light gray-shaded region highlights the space between the 10- and 90-percentile segmentations.
