Towards a more realistic anthropomorphic chest phantom using 3D-printed and cork-integrated components

Abstract

Background: Thorax phantoms for computed tomography (CT) imaging often lack realistic lung parenchyma and bronchovascular anatomy. To improve anatomical accuracy, 3D-printed chest phantoms have been developed as more realistic alternatives to existing models.

Purpose: To evaluate whether an in-house developed anthropomorphic phantom insert realistically represents the anatomical structures and attenuation characteristics compared to the original phantom and human CT data.

Methods: The anthropomorphic chest phantom "Lungman" was modified by integrating a 3D-printed insert, cork-based lung parenchyma, and lung nodules. The phantom was scanned on a CT system and evaluated using qualitative and quantitative CT analyses, comparing attenuation values and histogram distributions to human CT data. Subjective radiologist assessments were conducted to compare anatomical realism between the modified and unmodified phantom.

Results: Qualitative assessment of CT value distribution showed strong similarity between the modified phantom and human lung parenchyma, although the radiodensity characteristics of the 3D-printed bronchovascular insert still require further refinement. Quantitative analysis confirmed that the modified phantom's parenchymal attenuation (-854 Hounsfield unit [HU]) closely matched human lung parenchyma (-872 HU, p > 0.05), whereas the unmodified phantom showed lower attenuation (-997 HU, p < 0.05). However, the bronchovascular insert showed lower attenuation than human vasculature (-41 HU vs. 42 HU, p < 0.05). Radiologists rated the modified phantom higher for anatomical realism.

Conclusions: This study presents a modified, anthropomorphic thorax phantom that integrates realistic 3D-printed bronchial and vascular structures, lung parenchyma, and lung nodules.

Keywords: 3D‐printing; anthropomorphic thorax phantom; computed tomography; lung cancer; pulmonary nodules; quality assurance.

FIGURE 1

The original anthropomorphic chest phantom “Lungman” consisting of a mediastinum, spine and ribs, pulmonary vasculature, and an abdomen block.

FIGURE 2

(a) Combined 3D‐model of mediastinum (orange), bronchi (gray), and vessels (red) in preparation for 3D‐printing and (b) the 3D‐printed insert. (c) A caudo‐cranial view of the insert placed inside the phantom's thoracic cavity and (d) cork granules that were poured into the phantom.

FIGURE 3

Axial CT images of the (a) unmodified phantom and (b) modified phantom with artificial lung nodules, indicated by the red arrows. The display window level and window width were set to −400 and 1500 HU, respectively.

FIGURE 4

Axial CT images of the unmodified phantom (a), modified phantom (b), and human scan (c) in the upper (1), middle (2), and lower (3) scan sections. The display window level and window width were set to −400 and 1500 HU, respectively.

FIGURE 5

Normalized histograms (lung area percentage), split into two sections, showing CT value distributions for (a) lung parenchyma from −1024 to −700 HU and (b) bronchovasculature from −150 to 200 HU for human, unmodified, and modified phantom scans.

FIGURE 6

Boxplots displaying the median and interquartile range (IQR, 25th–75th percentiles) CT values for (a) lung parenchyma and (b) bronchovasculature in the human, unmodified phantom, and modified phantom CT scans, acquired at a tube voltage of 120 kVp.

FIGURE 7

Boxplots displaying the median and interquartile range (IQR, 25th–75th percentiles) CT values measured at tube voltages ranging from 70 to 140 kVp for (a) lung parenchyma and (b) bronchovasculature of the modified phantom.

FIGURE 8

Bar chart displaying the Likert scores (1 = very poor, 2 = poor, 3 = acceptable, 4 = good, 5 = very good) for anatomical realism of lung vasculature and bronchi, parenchyma, lung nodules, and image quality, as assessed by two radiologists. The scores compare the modified and unmodified phantoms across all evaluated aspects.