Micro-CT scanning

TPU powder had lower attenuation compared to the vessels (both under clinical CT and micro-CT imaging), indicating the difference in material properties between the powder and its fused form. However, bright spots within the powder, corresponding to areas of highest attenuation, were observed in micro-CT images, suggesting localised regions of increased density or material variation (Fig. 4). The detailed material composition is not provided by 3D-manufacturers, in general. The CT attenuation properties of the 3D-printed phantom were presented elsewhere [7], showing that solid HP-TPU had an attenuation similar to that of blood (without contrast), while HP-TPU in powder form was in the range of ground glass opacification.

Fig 4: TPU powder had lower attenuation compared to the vessels (both under clinical CT and micro-CT imaging), and appears as diffuse grey area in the images. Bright spots within the powder, corresponding to areas of highest attenuation, were observed in micro-CT images (highlighted in circle in the figure), suggesting localised regions of increased density or material variation.

Printing accuracy evaluation

Colour map from Measure Thickness  tool (Fig. 5) shows that the printed model was within the 3D printer’s resolution limit, indicating satisfactory structural integrity.

Fig 5: Image of 3D-printed phantom is shown on the left. The thickness of the CT-phantom was measured using the Measure Thickness tool in MeshInspector, with a threshold set to 0.25 mm, which corresponds to the printer’s resolution for minimum details as specified by the manufacturer. From the phantom thickness colour map, the entire mesh appears green, indicating that the thickness is within the printer's resolution limit.

The CT-phantom vessel diameters were consistent with the design within the measurement error for large vessels (Test_1, Design_model: 7.20±2.32 mm, CT_model: 6.98±2.14 mm). Small vessel diameters were consistent between the design and the CT-phantom model at the 3D printer's minimum wall thickness threshold of 1 mm [6].

Coloured distance maps show in general excellent agreement between both meshes (Test_2, Fig. 6). Global distance analysis was satisfactory, with AAD=0.21±0.38 mm. AAD values from distance comparison within the VOIs containing large vessels (of diameter ~7 mm, Fig. 7) and small vessels (of diameter ~1.4 mm, Fig. 8) were 0.15±0.37 mm and 0.08±0.09 mm respectively. This indicates better accuracy for smaller objects, which is consistent with 3D printer's stated precision of 0.9% in the XY plane and up to 1.8% in the Z direction, with minimum error limits of 1 mm in XY and up to 1.5 mm in Z (the uncertainty is higher for larger objects) [6].

Fig 6: The distance comparison map used for the global assessment of printing accuracy illustrates the differences between the design and scan models. The global distance analysis shows that discrepancies are most noticeable along the edges, caused by some misalignment of a few small vessels. Deviations resulting from unprinted vessels are not shown in the distance comparison map. Average absolute distance (AAD) = 0.21±0.38 mm.

Fig 7: The distance comparison map, used for the local evaluation of printing accuracy, visually demonstrates the excellent agreement between the design and scan models in the large vessels (diameter ~7 mm). Average absolute distance (AAD) = 0.15±0.37 mm.

Fig 8: The distance comparison map, used for the local evaluation of printing accuracy, visually demonstrates the excellent agreement between the design and scan models in the small vessels (diameter ~1.4 mm). Average absolute distance (AAD) = 0.08±0.09 mm.

The global distance analysis shows that discrepancies are most noticeable along the edges next to the elliptical rim and at the top and bottom of the phantom, caused by some misalignment of a few small vessels and a localised fraction of small vessels that were not printed. Misalignment of vessels between the CT-phantom and design models was likely due to physical deformation of the vessels, as they are made of soft and flexible rubber material (TPU). As for the future application of this lung phantom, the accuracy of the print is determined not by the spatial distribution of the vessels, but by the printer’s ability to reproduce the specified diameters and lengths from the design file. The misaligned vessels were isolated, registered, and then compared using distance analysis. After accounting for misalignment errors, the printing accuracy with respect to vessel dimensions was confirmed. AAD from one of such measurements was 0.07±0.06 mm.